EPA/600/R-08-087
July 2008
ANALYSIS OF INDIRECT EMISSIONS
BENEFITS OF WIND, LANDFILL GAS,
AND MUNICIPAL SOLID WASTE
GENERATION
By
Ezra D. Hausman, Jeremy Fisher, and Bruce Biewald
Synapse Energy
22 Pearl Street
Cambridge, MA 02139
Order No.: EP07C000079
Project Officer
Joseph F. DeCarolis
Air Pollution Prevention and Control Division
National Risk Management Research Laboratory
Research Triangle Park, NC 27711
-------
Notice
The U.S. Environmental Protection Agency, through its Office of Research and Development,
funded and managed the research described here under Order Number EP07C000079 to
Synapse Energy Economics. It has been subjected to the Agency's peer and administrative
review and has been approved for publication as an EPA document.
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 of 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 National 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. NRMRL 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 Gutierrez, Director
National Risk Management Research Laboratory
Indirect Emissions Analysis • i
-------
Abstract
A number of techniques are introduced to calculate the hourly indirect emissions benefits
of three types of renewable resources: wind energy, Municipal Solid Waste (MSW)
combustion, and Landfill Gas (LFG) combustion. These techniques are applied to each of
the EPA's eGRID subregions in the continental United States in order to derive hourly,
seasonal, or annual (as appropriate) coefficients for use in evaluating the indirect
emissions benefits of such projects in each region.
For wind power impacts, simulated wind project power profiles are derived using publicly
available wind data scaled to a typical turbine height for new wind projects and
transformed through a power curve for a proxy turbine. The results for one region are
compared to the output of an existing project, although the limitations of this type of
comparison, as well as the limitations of representing large areas of the country with
single proxy curves, are discussed. LFG and MSW are found to have flat, "always on"
profiles based on the limited data available.
The regional, hourly power profiles for each type of resource are combined with the
hourly indirect emissions coefficients to yield annual indirect emissions benefits for each
type of resource for each eGRID subregion.
For each GWh of renewable energy produced each year, the indirect CO2 emissions
benefit is found to be between 600 and 1100 tons of CO2 depending on the region, with
coal-dependent regions having the highest indirect emissions benefit. The indirect NOx
and SO2 emissions benefit also depends on the regional fuel mix, as well as the
stringency of environmental regulation in each region. These indirect emissions benefits
vary between 500 and 5,000 pounds of NOx per GWh, and between 200 and 1,300
pounds of SO2 per GWh. With some exceptions, these results are robust regardless of
whether the renewable resource has a base load profile (such as MSW and LFG) of
output which varies diurnally and seasonally (such as wind). However, as the calculated
benefits do vary in some cases depending on the analytical approach, care must be used
in selecting an appropriate calculation methodology for each application.
This report was submitted in fulfillment of Order Number EP07C000079 under the
sponsorship of the EPA.
Indirect Emissions Analysis • ii
-------
Table of Contents
Notice i
Foreword i
Abstract ii
List of Tables iv
List of Figures v
Acronyms and Abbreviations vii
List of Appendices viii
1. Introduction and Summary of Conclusions 1
2. Operating Data on Renewable Energy Technologies 6
A. Wind Generation: Existing and Synthetic Projects 6
Data Sources for Synthetic Generation 7
Generation of Synthetic Wind Power Data 8
Comparing Empirical and Simulated Wind Generation 10
B. Landfill Gas 14
C. Municipal Solid Waste 14
3. Hourly generation and emissions data 15
Characterization of emission rates 16
Data filtering 16
4. Emissions Displacement Methods 17
A. Limiting Cases 18
B. Hourly Average Emissions Rate (HAER) 19
C. Emissions Slope Factor 22
D. Empirical Incremental Emissions Rate (EIER) 29
E. Load-Following Incremental Emissions Rate (LFIR) 30
F. Flexibility-Weighted Hourly Average Emissions Rate (FW-HAER) 31
G. Comparison of Indirect Emissions Rates 35
5. Indirect Emissions Results 37
6. Conclusions 44
Appendix A: 46
Appendix B: 48
Appendix C: 52
Appendix D: 56
Indirect Emissions Analysis • ill
-------
List of Tables
Table 1. Seasonal and annual emissions slope factors and R-squared
values for C02 24
Table 2. Seasonal and annual emissions slope factors and R-squared
values for S02 25
Table 3. Ozone season and non-ozone season emissions rate
slopes for NOx 26
Table 4. Guide to the summary figures 37
Table 5. Annual indirect emissions reduction rates associated with an
incremental GWh of wind energy in each of the eGRID
subregions for 2005 39
Table 6. Annual indirect emissions reduction rates associated with an
incremental GWh of landfill gas of MSW generation in each of
the eGRID subregions for 2005 40
Table 7. "Shape impact" for indirect emissions calculations for each
pollutant in each eGrid region 40
Indirect Emissions Analysis • iv
-------
List of Figures
Figure 1. Indirect emissions results for C02, NOx, and 862 for wind,
municipal solid waste and landfill gas generation projects 4
Figure 2. Wind power curve for a GE 1.5se wind turbine 9
Figure 3. Proxy wind turbine power curve plotted against histograms of
raw and scaled wind speeds at a sample site in the Midwest 10
Figure 4. Comparison of power output distribution at sample wind
power site in the Midwest with predicted output based on
scaled wind data from a nearby meteorological station 12
Figure 5. Comparison of hourly synthetic and observed wind power
time series at nearby locations in the Midwest 13
Figure 6. Cumulative distribution of the hourly deviation between
observed and synthetic generation time series 13
Figure 7. Distribution of C02 emission rates for fossil generating plants
by reported fuel type 18
Figure 8. Representations of C02 hourly average emissions rate
(HAER) in New England 21
Figure 9. Determination of the regional emissions slope factor in the
RFCE (Reliability First/Central) region 23
Figure 10. Bifurcation of the NOx emissions slope factor for the RFCE
region 23
Figure 11. Regional slope factors for C02 27
Figure 12. Seasonal and regional slope factors for S02 27
Figure 13. Regional slope factors for NOx 28
Figure 14. Flexibility coefficient vs. generator size in New England and
California 33
Figure 15. Hourly dispatch and flexibility coefficient for three plants of
differing capacities in the FRCC subregion 34
Figure 16. Comparison of the distribution of indirect emissions
coefficients for C02, S02, and NOx based on methods
described in the text 36
Indirect Emissions Analysis • v
-------
Figure 17. Annual indirect C02 emissions impact associated with one
incremental GWh of wind energy in each of the eGRID
subregions for 2005 41
Figure 18. Annual indirect NOx emissions impact associated with one
incremental GWh of wind energy in each of the eGRID
subregions for 2005 41
Figure 19. Annual indirect 862 emissions impact associated with one
incremental GWh of wind energy in each of the eGRID
subregions for 2005 42
Figure 20. Annual indirect C02 emissions impact associated with one
incremental GWh of landfill gas or municipal solid waste
energy in each of the eGRID subregions for 2005 42
Figure 21. Annual indirect NOx emissions impact associated with an
incremental GWh of landfill gas or municipal solid waste
energy in each of the eGRID subregions for 2005 43
Figure 22. Annual indirect 862 emissions impact associated with an
incremental GWh of landfill gas or municipal solid waste
energy in each of the eGRID subregions for 2005 43
Indirect Emissions Analysis • vi
-------
Acronyms and Abbreviations
AWEA American Wind Energy Association
BTU British Thermal Unit
CO2 Carbon Dioxide
DOE U.S. Department of Energy
eGRID Emissions & Generation Resource Integrated Database
EIER empirical incremental emission rate (defined in this study)
EPA United States Environmental Protection Agency
FW-HAER flexibility-weighted hourly average emission rate (defined in this study)
GE General Electric
GW gigawatt
GWh gigawatt-hour
HAER hourly average emission rate (defined in this study)
kW kilowatt
kWh kilowatt-hour
LFG landfill gas (generating resource)
LFIR load-following incremental emissions rate (defined in this study)
mmBTU million British thermal units
MSW municipal solid waste (generating resource)
MW megawatt
MWh megawatt-hours
NCDC National Climatic Data Center
NERC North American Electric Reliability Corporation
NOAA National Oceanic and Atmospheric Administration
NOx mono-nitrogen oxides (NO and NO2)
NREL National Renewable Energy Laboratory
SO2 sulfur dioxide
WBAN Weather Bureau Army Navy
Indirect Emissions Analysis • vii
-------
List of Appendices
Appendix A
Identification and map of eGRID subregions
Appendix B
Synthetic wind power time series (hourly percent of capacity) for each eGRID subregion,
represented as color intensity plots
Appendix C
Summary tables of raw and scaled wind speed and synthetic wind power time series
Appendix D
Color intensity maps of hourly indirect CO2, NOx and SO2 emissions rates for each
eGRID subregion, using several alternative calculation methodologies
Indirect Emissions Analysis • viii
-------
1. Introduction and Summary of Conclusions
The purpose of this report is to develop, analyze, and report upon methods of quantifying
the indirect emissions benefit associated with renewable energy generation resources. In
particular, the goal is to conduct a national assessment of indirect emissions reductions
from wind, LFG and MSW by NERC subregion.1 Because conventional electricity
generation resources can differ widely in their emissions characteristics, the
determination of exactly which conventional resource or resources would be running but
for the contribution of the renewable resource would be the ideal first step in calculating
indirect emissions benefits. If we knew which resources were "displaced" by the
renewable resource, we could simply multiply the emission rate of those resources by the
total number of megawatt-hours (MWh) displaced, accounting for line losses, and an
exact answer would be obtained. The answer thus obtained would vary from region to
region, and from hour-to-hour within any given region, depending on current and
expected market conditions.
Unfortunately, in large, interconnected, security-constrained2 electricity markets,
determining which units would be displaced in any given hour would require perfect
information and a complex dispatch model. Even under such ideal circumstances, the
analysis would be prohibitively time-consuming and complex for each hour. Thus our goal
in the current study is to develop and apply proxy methods that, while imperfect, are
applicable to and useful for the general estimation of the indirect emissions benefits for
renewable energy projects anywhere in the continental United States. Over the course of
this report, we develop and apply these methods to estimate the indirect emissions
benefits of three kinds of renewable energy projects, for carbon dioxide (CO2), sulfur
dioxide (SO2) and oxides of nitrogen (NOx), in each of the 22 eGRID3 subregions of the
continental United States.
An "indirect emissions benefit" may be a real reduction in total emissions in a region or,
under cap-and-trade regulation such as that in much of the United States for the
pollutants SO2 and NOx, it may be an opportunity to release emissions allowances for
some other use. The calculation methods explored in this report are appropriate for either
application, but the impacts of these scenarios are quite different in terms of their ultimate
effect on pollutant emissions. In the case of cap-and-trade regulation, displacing air
emissions provides an opportunity for the allowance holder to sell or bank valuable
emissions allowances. There may also be a societal benefit in terms of reducing the
market price of emissions allowances, thereby reducing this component of the cost of
1 EPA RFQ # RFQ-OH-07-00015, as posted at:
https://www.fbo.gov/?s=opportunity&mode=form&id=556118954b3728ae31326t>103396bcf6&tab=core&_cview=0.
2
"Security constrained" refers to the need for system operators to dispatch generating units to meet load while respecting
the loading on transmission lines and interfaces within their operating limits. Because of these security constraints,
generating units cannot simply be dispatched in merit order, from least expensive to most expensive running costs. In
many cases more expensive units must be run out of merit order due to transmission transfer limits.
The Emissions & Generation Resource Integrated Database (eGRID) is a comprehensive inventory of environmental
attributes of electric power systems, (http://www.epa.gov/cleanenergv/energy-resources/egrid/index.html.) Definitions and a
map of the eGRID subregions are shown in Appendix A. eGRID subregions are roughly coincident with NERC (North
American Electric Reliability Corporation) subregions.
Indirect Emissions Analysis • 1
-------
electricity. Finally, there may ultimately be a total emissions reduction if the lower cost
and greater availability of allowances permits regulators to tighten the total emissions
cap.
This report is organized as follows:
• In Section 2, we develop and present operating characteristics of wind, landfill gas,
and municipal solid waste energy resources in each of the eGRID subregions of the
United States. We develop hourly operating profiles for a "typical" such resource in
each region, normalized to produce one GWh4 of electricity per year.
• In Section 3, we collect and present data on conventional power plants in each
eGRID region based on the EPA's Acid Rain database.5 We characterize the data
and discuss our approach to data filtering and error handling, in preparation for
characterizing the indirect emissions benefits associated with changes in dispatch.
• In Section 4, we implement and evaluate a number of approaches to applying the
filtered and cleaned EPA data for the purposes of calculating hourly avoidable
emissions factors. These factors can then be applied to renewable resource
production profiles to calculate annual indirect emissions benefits. In addition, we
present hourly indirect emissions results for each pollutant, region, and calculation
approach under consideration, and we discuss which approaches make sense for
which kinds of analyses.
• In Section 5 we apply the indirect emissions calculation methods to determine
indirect emissions benefits associated with each resource type under consideration
for each region and pollutant.
• Finally, in Section 6, we discuss our results and present suggestions for further
research in this area.
Fundamental to the analysis in this report, all of which is based on retrospective data, is
the primary focus on short-term, operational response of the electric system to the
presence of new renewable energy resources. This analysis is thus most appropriate for
analyzing the impact of a relatively small quantity of new resources during the first few
years of operation, during which the displaced resources are likely to be the most flexible,
load-following units on the system. Over the long term, it is likely that the capacity mix of
an electric system will be altered as a result of the addition of new renewable energy
resources. Specifically, new generating capacity investments may be avoided or
deferred, or existing capacity may be retired, until the system returns to equilibrium in
terms of the balance of base load and load-following resources. Ultimately, it may be
expected that the proportion of flexible units will be restored to what it would have been
without the new resource. In this sense, as it matures and is incorporated in larger
quantities, renewable energy technology will ultimately displace more base load capacity
4
A GWh (gigawatt-hour) is one thousand megawatt-hours, approximately the amount of power that could be produced by
one of the largest power plants in the United States in one hour. For perspective, the largest new single wind turbines
are about 3 MW in capacity but produce, on average, about one MWh of electricity per hour. Thus one of the largest
wind turbines would be expected to produce about nine GWh of electricity each year. In general, wind projects have
more than one wind turbine, and can have up to 200 or more.
Available from the EPA Clean Air Markets data website, at http://camddataandmaps.epa.gov/gdm/
Indirect Emissions Analysis • 2
-------
on the system. While we touch on this important issue in some parts of this report, it is
not a primary focus.6
The indirect emissions benefit calculated for any kind of renewable energy resource
depends on a number of questions:
• Where is the resource located?
• What is the pollutant of interest?
• Is the time period of interest historical, in the near future, or several years in the
future?
• Is the resource base load, dispatchable, or intermittent and nondispatchable?
• If intermittent and nondispatchable, what is the expected hourly and seasonal profile
of the resource?
Because of the wide range of applications reflected in the possible answers to these
questions, the best method to use in calculating indirect emissions benefits can vary with
the application. In this report we present a number of methods, and apply them to three
types of renewable resources and three pollutants in 22 regions of the United States.
Even this level of detail is almost certainly insufficient for precise analysis of the
emissions benefits of individual projects. However, it leads to illuminating and useful
results that will allow for a far more detailed understanding of the issues and a better first
approximation of the indirect emissions benefits than has previously been available.
We find:
• The indirect emissions benefits of renewable energy for all pollutants vary
significantly by region, and that these differences can be quantified and applied in
calculating the indirect emissions reductions.
• For each GWh of renewable energy produced each year, the indirect CO2 emissions
benefit is found to be between 600 and 1000 tons of CO2, with coal-dependent
regions having the highest indirect emissions benefit.
• For each GWh of renewable energy produced each year, the indirect NOx emissions
benefit varies between 500 and 5,000 pounds of NOx.
• For each GWh of renewable energy produced each year, the indirect SO2 emissions
benefit varies between 200 and 1,300 pounds of SO2.
• These results are robust regardless of whether the renewable resource has a base
load profile (such as MSWand LFG) of output which varies diurnally and seasonally
(such as wind).
• Different calculation approaches are appropriate for different types of resources and
different applications.
For a thorough treatment of this issue, see 2005 Synapse report Using Electric System Operating Margins and Build
Margins: Quantification of Carbon Emission Reductions Attributable to Grid Connected COM Projects, available at
http://www.synapse-energy.com/
Indirect Emissions Analysis • 3
-------
(a)
• More research is needed to establish which methods of calculating indirect emissions
changes most accurately reflect real-world dispatch over a range of timescales and
system conditions.
Of the methods introduced here, we find the flexibility-weighted hourly average emission
rate (FW-HAER) to be the most appropriate for near-term (<3 year) projections of indirect
emissions impact from wind resources, because it best represents the response of the
electric system to rapid fluctuations in system load. We find the seasonal slope factor
approach to be most appropriate for calculating the indirect emissions impact from base
load resources such as municipal solid waste and landfill gas electricity generation.
The annual indirect emissions results for these resources based on 2005 data, assuming
one GWh of energy was produced by each over the course of the year, are shown in
^Figure 1T With some notable exceptions, the indirect CO2 and NOx emissions benefits of
the different renewable energy resources is generally similar; this benefit diverges most
dramatically by resource type for SO2. This divergence probably reflects the fact that in
many regions, base load resources like MSW and LFG can displace coal, which emits
SO2, while wind initially primarily replaces gas, which does not emit SO2. However, with
increasing penetration and geographic diversity of wind generation it is likely that this
resource will begin to displace more base load coal, and the indirect emissions impact of
wind generation will be closer to that of MSW and LFG.
1,200 -,
1 ,000
CD
O
O
'o
(A
c
o
T3
8
jn
Q.
800
600
400
200
X
<
0
0
O
0
CC
DWind
BMSW/LFG
Figure 1. Indirect emissions results for (a) COz (b) NOx; and (c) SC>2 for wind, municipal solid
waste (MSW) and landfill gas (LFG) generation projects. (Continued on next page)
Indirect Emissions Analysis • 4
-------
(b)
b,UUU -
_c
O
9. A nnn
X
o
'o
m ^ nnn
c
o
CL
"m 9 nnn
ro
Q.
(A
b
1 nnn
1-1
5
z
N
<
-
flffl
-
i-.
-i
x H o iu <
s °. o O i
r
p.
_
p
r
-i
r
-i
-i
r
'
= LiJCLaziQ-LiJ5§5O>O
)5|o^^^oo|zco||co^>
= z-zz zcccccccecococo^coWco
DWind • MSW/LFG
(C)
I4,UUU -
19 nnn
i
O -in nnn -
0
Q_
6"
co R nnn
c
D c nnn
CL
T3
8
ro ^ nnn -
(A
b
9 nnn -
n -
fln-rll"
•I
-,
[
imr
r
-
r
-i
r
-
[
\
-
r
Iffl
-1
r
-
r
DWind
MSW/LFG
Indirect Emissions Analysis • 5
-------
2. Operating Data on Renewable Energy Technologies
In order to determine the indirect emissions reduction impacts of electricity production
from renewable energy sources, it is necessary to first establish what kinds of
conventional resources they are displacing. As a first step, in this section of the report we
explore the operational characteristics of three types of renewable energy resources:
wind power generation, municipal solid waste combustion, and landfill gas. We develop
profiles for these resources representative of each eGRID subregion of the continental
United States7 based on historic data and operating characteristics of these resources.
(The eGRID subregions are identified and mapped in Appendix A.) Once we have
determined their hourly output profiles over the course of a historic year, we can begin an
assessment of which resources they would have been likely to displace.
In the section that follows, we will explore corresponding operational and dispatch
characteristics of conventional resources. Following that we combine the results of these
research tasks, characterizing renewable energy output and assessing displaceable
conventional resources, to make a final assessment of which fossil resources would have
been likely to be displaced by renewables in each eGRID subregion. Finally, we use the
results of this combined analysis to determine factors representing the indirect emissions
benefit for each subregion.
A. Wind Generation: Existing and Synthetic Projects
The American Wind Energy Association (AWEA) estimates that there are approximately
8.3 gigawatts (GW) of installed wind capacity in the United States as of the 4th quarter of
2007, with another 4.1 GW under construction.8 Texas and California dominate the
current market, but there are underutilized regions throughout the Midwest. Some states
with the greatest wind potential, such as the corridor stretching from North Dakota south
through Kansas, have yet to significantly tap this resource. In general, there are
extensive opportunities to expand the rapid growth in wind generation, resulting in the
further displacement of fossil fuel generation and decreased emissions of greenhouse
gases and other pollutants. This large resource potential motivates the need to
understand in more detail the emissions benefits associated with renewable energy
projects in each region.
Ideally, to characterize the output profiles for new wind generation projects, we would
draw empirical wind power output datasets from a broad sampling of existing projects
around the country. Unfortunately, wind generation projects are not required to publicly
report hourly generation, and thus there are limited opportunities to calculate expected
emissions reductions from existing resources. In addition, "real" wind projects present
numerous hurdles that make them problematic as a source of research data for a broad-
based study of operational characteristics. For example, there is generally no uniformity,
and often no available information, on the technical details of existing resources. With
eGRID subregions are almost coincident with NERC subregions. We chose to use eGRID subregions for this analysis
as this is the system used by EPA for emissions reporting.
American Wind Energy Association, 2007. Updated information can be found at http://www.awea.org/projects/.
Indirect Emissions Analysis • 6
-------
operational data available on few projects, what may appear to be regional differences in
operating characteristics would be heavily influenced by unknown differences in
technology, turbine size, turbine height, vintage, and specific local conditions associated
with each individual project. In addition, the geographical coverage of available wind
power output data is extremely limited. In fact, we were only able to find one source for
such data, from the U.S. Department of Energy's National Renewable Energy Laboratory
(NREL). This dataset covered only a small number of projects with limited geographic
distribution. Even with this source we were asked not to present specific, identifiable
details or operational data to avoid disclosure of proprietary information, limiting its
usefulness for our study.
Our solution to these limitations is to simulate wind power projects based on
meteorological data gathered at locations where wind projects are likely to be built in
each region, determined based on the quality of the wind resource. While wind power
data are scarce, wind speed data are abundant and readily available at reasonably high
resolution throughout the United States. Near-surface hourly wind speed time series data
can be scaled to reflect a typical turbine height, and then converted with the help of a
"power curve"9 into corresponding simulated wind power output time series. We have
attempted to perform a comparison of one of our simulated wind power time series with
the output of a nearby physical wind-power project; however, the results are difficult to
interpret. The wind power and the wind data are not from exactly the same location or
height, and even such small differences can be extremely significant in determining the
output of a resource. The technology and vintage represented by our power curve is also
unlikely to be well matched to the actual technology and vintage of the specific wind
turbines for which we have data. Even so, this approach provides a useful means of
evaluating the potential impacts of wind energy production on emissions. The data
sources and comparison results are described in more detail below.
Data Sources for Synthetic Generation
Over 1700 meteorological stations collect hourly data throughout the United States and
territories for the National Climate Data Center (NCDC) of the National Oceanic and
Atmospheric Administration (NOAA). Our primary source of wind data for generating
synthetic wind power time series was NCDC's Weather Bureau Army Navy (WBAN)
stations, distributed throughout the United States. To estimate the temporal
characteristics of potential wind power generation, and thus expected indirect emissions
reductions for each eGRID subregion, we needed to determine likely locations for new
wind generation projects. We did this by reviewing the highest resolution available 50-
meter wind speed estimates in each eGRID subregion, as cataloged by the NREL,10 to
determine where favorable wind resources were likely to be located based on average
wind speeds at 50-meter hub heights. We assumed that the minimum economic wind
speed threshold for a wind power project would be at Class 4 wind speeds, or winds of
9
A "power curve" is a graph which relates wind speed to expected power production for a specific wind turbine technology
and configuration. Actual power output data from wind turbines generally exhibit considerable scatter around the
expected power curve, but this remains the best means of characterizing this relationship.
10 USDOE NREL Wind and Hydropower Electricity Program, 2007. Available online at:
http://www.eere.energy.gov/windandhydro/windpoweringamerica/wind_maps.asp
Indirect Emissions Analysis • 7
-------
15.7-16.8 miles per hour (mph). Once the area was selected in this manner, the closest
high-quality NCDC station wind data location was selected for each subregion. Twenty-
two NCDC meteorological stations near these locations were selected as proxies to
represent the temporal pattern of wind generation expected at these sites. The locations
of the sites and summary wind characteristics from the selected NCDC WBAN stations
may be found in Appendix C.
It must be noted at the outset that the geographic granularity of the wind data is far from
satisfactory. Wind conditions can change significantly over spatial scales as small as a
few kilometers or less, so representing each eGRID subregion with a single site entails a
considerable simplification. We have tried to choose sites which are reasonably
representative of promising wind power locations for each region, understanding that
considerably more research and analysis would be needed to pick a truly representative
site, if indeed a single site can stand in for a large region. The same can be said for solid
waste and landfill gas projects considered in our analysis: there is simply not enough
information available to confidently predict the behavior of all potential future sites. This is
an inevitable shortcoming of attempting to represent site-specific characteristics on a
national scale, and it should be kept in mind by the reader in interpreting our results.
Generation of Synthetic Wind Power Data
WBAN meteorological stations are typically located at local and regional airports, or near
rural towns, at 10 meters (33 ft) above ground level. Due to surface friction, wind speeds
near the ground are significantly lower than speeds at expected 50-100 m windmill hub
heights. In addition, the WBAN stations collect meteorological data in irregular intervals,
with increments of approximately one hour. To create hourly synthetic time series, we
used the simple average of the two measurements surrounding each hour. Next we
normalized wind speeds to Class 4 speeds by scaling the time series for each station to
an average speed of 16.25 mph, as follows:
we woo 16.25
wst. = wst. * -=
'•> *•> TJ7"O
Where:
WSjj = Adjusted wind speed during hour; and station j
WS°ij= Unadjusted wind speed during hour; and station j
and WSj = Average unadjusted wind speed at station j over all hours.
Wind scales non-linearly with height, increasing in speed as the influence of ground-level
friction decreases. However, we assume that a proximal Class 4 wind site would have
similar temporal characteristics to the meteorological station, and thus our approximation
is to scale the WBAN wind data linearly to reflect Class 4 characteristics. This approach
encompasses a host of decisions, design parameters, and operational characteristics into
a single scalar that represents, roughly, the decisions made by the builder of a wind
power facility to optimally exploit a local resource.
Indirect Emissions Analysis • 8
-------
To convert the scaled wind data into synthetic wind power data, it was necessary to
select a proxy technology with an appropriate wind power conversion function. We used
a wind power curve based on the General Electric (GE) 1.5se (1.5 MW nameplate
capacity) wind turbine to estimate potential generation associated with the wind time
series at each selected meteorological station. The power curve, shown in/igure 2,, is
determined by a number of factors including the area swept by the blades, the "pitch" or
angle of the blades (which can be variable in different wind conditions), and
characteristics of the turbine gearbox. Wind turbines require a minimum amount of wind
to begin generating power, known as the cut-in speed, and are designed to stall, or even
brake, at higher wind velocities (the cut-out speed) to avoid damage.11 Once again, we
do not know a priori what technology or set of parameters would be most appropriate for
a wind turbine in each area, but by scaling the winds to match a proxy turbine we
generate a series that preserves temporal variability while representing a project which is
optimized to local conditions.
Power Curve
1,800
246
-GE 1.5 xle
8 10 12 IH
-GE1.5sl/sle
16 18 20 22 24
— GE 1.5s/se m/s
Figure 2. Wind power curve fora GE 1.5se (1.5 MW) wind turbine.
Source: Http://www.gepower.com/prod_serv/products/wind_turbines/en/downloads/ge_15_brochure.pdf
The GE 1.5se proxy turbine is a common mid-sized turbine used on-shore; larger
turbines are generally used offshore due to high wind conditions. The design tower height
for these turbines is 54.7 meters (m), with 35.25 m blades sweeping an area of 3,904 m2.
The cut-in speed is 4 meters per second, and the cut-out speed is 25 meters per second.
To generate the wind power time series, we relied upon a GE-supplied power curve
which was imported as a look-up table for our analysis; the look-up table indexes the
power curve and returns an estimated generation value for each hour given the
calculated wind speed at turbine height.
Tony Burton, David Sharpe, Nick Jenkins, Ervin Bossanyi, Wind Energy Handbook, John Wiley & Sons, 2001.
Indirect Emissions Analysis • 9
-------
For purposes of our analysis, we approximated the GE wind power curve with a sigmoid
function:
Q =
a
l + e
b-cWS
(2)
Where Q is output (kW) and l/l/S is wind speed (m-s~). The best fit to the GE curve in
these units was obtained with parameter values a=1500, b=8.1, and c=0.9. The curve
was truncated to zero at the cut-out speed of 25 m-s"1 The resulting curve is shown in
Figure 3. along with reported and scaled wind data.
20% n
18%
16%
14%
0 12%
.n
o 10%
§ 8%
1
6%
4%
2%
0%
n
1
ll . ^
1 234567
r
r n
f
[4t J.
/
[
1
/
fl
X
•
i 1 Raw Wind Speed
l=l Scaled Wind Speed
— Power Curve
J J Lfl 0 H n n „ „
8 9 10111213141516171819202122232425262728
- 1800
- 1600
o
- 1400 2
01
c
Ol
- 1200 o>
- 1000 m _
-800 | ~
II—
- 600 i.
-400 %
- 200 £
- 0
Wind speed (mph)
Figure 3. Proxy wind turbine power curve plotted against histograms of raw and scaled wind speeds at a
sample site in the Midwest. "Scaled" wind speeds are adjusted to reflect class 4 winds at a turbine height of
50-100 meters using Equation 1.
Comparing Empirical and Simulated Wind Generation
As noted earlier, we were able to find one source of wind power data for comparison with
our synthetic power profiles. The wind power time series that we were able to obtain
represent twelve operational projects associated with the NREL wind farm monitoring
program. The NREL data cover twelve sites including two in Minnesota, one in Iowa, one
in Oklahoma, four in Texas, and four in Oregon. Total capacities ranged from 25 MWto
230 MW, and the turbines ranged from 0.6 to1.6 MW each. While NREL was able to
make 2005 production records from these projects available for research purposes, site-
identifying information remains confidential. The turbines represent primarily plain and
low-ridge wind farms; the farms in Iowa and Minnesota are distributed throughout active
agricultural land, while the Oregon, Oklahoma, and Texas sites partially overlap grazing
lands.
Indirect Emissions Analysis • 10
-------
To test whether our simulated wind generation time series were realistic and
representative of the output profile for a physical wind project at the same location, we
selected an NREL wind project from a midwestern U.S. location and compared the
recorded output with the simulated production time series generated using wind data
from a nearby location. The test project is in a region with Class 4 and Class 5 wind
speeds, running 145 turbines of 750 KWeach, fora total capacity of 108 MW. A nearby
WBAN meteorological station was located, and wind records for this location extracted
from the NCDC data. The wind data were transformed through the power curve (Equation
2) to generate hourly power output.
Such a comparison has a number of complications, including:
• Gaps in wind data. In the case of the selected series, there are 84 gaps larger than
2 hours in duration, and 9 gaps longer than 24 hours in duration. The longest gap is
nearly a week in duration. We have no way of knowing whether these were neglected
because they were zero, or they were very high, or they simply represent random
data errors, Thus it is impossible to determine what effect their omission may have on
our analysis.
• Gaps in power output data. The output time series for the installed project has 552
missing hours in the sample year (6% of all hours) which were removed from the
analysis. As with the missing wind data, it is impossible to determine what effect their
omission may have on our analysis.
• Granularity of wind data. Wind speed data are taken approximately once per hour,
meaning that variations in wind speed within each hour are unknown. In addition, the
wind data show recording artifacts, such as certain values that rarely appear in the
record including any measurements between approximately 0.5 and 2.75 mph. We
have not attempted to perform any smoothing to compensate for this.
• Representativeness of wind site. While the wind data are taken from a
meteorological station in proximity to the turbine, the station is clearly not measuring
winds identical to those that drive the turbine. This lack of correspondence to actual
values is exemplified by the average wind speed in the meteorological record of only
9.71 mph; to scale to the Class 4 average of 16.25 mph winds, we multiply all values
by 1.67, consistent with Equation 1. In addition, a wind power installation can contain
hundreds of turbines, each responding to highly localized conditions and each
subject to maintenance schedules and other operational constraints which are not
captured in the power curve.
• Idealization of power curve transformation. Each hour's scaled wind speed is
transformed into an hour of power output assuming a perfect generator response to
steady winds, neglecting any effects such as generator inertia which may dampen
the response. Wind speeds are implicitly assumed to hold steady for exactly one hour
before instantly changing to the next hour's value, again with an instantaneous
response from the turbine.
The comparison of simulated and observed power output is shown in ^Figure \ In this
analysis we take the maximum recorded output for the year as the maximum output for
the station, although this value is only about 90% of the nameplate capacity of the
Indirect Emissions Analysis • 11
-------
installation. This assumption presumably corrects for scheduled maintenance and forced
outages which would affect some subset of turbines at all times.
The simulated output curve shows an exaggerated peak at 5%-10% of capacity relative
to the recorded power output series, and a peak at the 100% output level which is absent
in the observed data. These peaks likely reflect the granularity issues identified above for
the wind records, which can reflect just one or two observations for each hour while the
measured power output reflects inevitable variations in wind speed and output over the
course of each hour. For the simulated project the overall capacity factor over the course
of the study period is 32.8%, while for the observed project the overall capacity factor is
30.3%.
30% -,
25%
20%
° 15% -
I
fe 10% -
Q.
5% -
0%
D Simulated Generation
• Obserwd Generation
O\O ,O\O
Percent of maximum output
O\O ,O\O
Figure A. Comparison of power output distribution at sample wind power site in the Midwest with predicted
output based on scaled wind data from a nearby meteorological station. Output levels for observed
generation are relative to the maximum recorded output for the year.
The simulated and observed generation, as well as the difference between the two, are
shown on color charts in/igure 5^ These charts represent each hour of the year as a cell
on a 24-hour by 365-day matrix, with each cell color-coded according the scheme shown.
^Figure e^shows a cumulative distribution of the hourly deviation between the two, in
percentage points. The simulated output exhibits a pronounced diurnal cycle which is not
present in the project data. This discrepancy could be due to any number of factors,
including light winds which blow through the evening at hub-heights which do not appear
in the NCDC dataset (NCDC does not record wind speeds less than 3 mph at ground
level) or more sensitive wind turbines with lower stalling speeds than the proxy turbine,
which would produce a generally flatter distribution of power generation throughout the
day. This discrepancy may also simply reflect the difference in the location of the two
sites. The two time series are within 20 percentage points of each other during
approximately 65% of the 6,434 hours recorded in both datasets.
Indirect Emissions Analysis • 12
-------
A) Project generation (% of capacity)
B) Expected generation from scaled wind speed (% of maximum)
I 15
20
Jai Feb Mar Apr May Jun JiJ Aug
Percent scale (A-B)
Sep
Oct
Nov
Dec
10
20
30
40
50
60
70
80
90
100
C) Difference: expected generation (% of maximum) minus observed project power (% of capacity)
5
10
15
20
Jai
Feb Mar Apr May
Jun Jul Aug
Percent scale (C)
Sep
Oct
Nov
Dec
-100 -80 -60 -40 -20 0 20 40 60 80 100
Figure 5. Comparison of hourly synthetic and observed wind power time series at nearby locations in the
Midwest. Each plot shows 8,760 hours in 2005, with hour of the day on the vertical axis and day of the year
on the horizontal. Missing data shown in black. Top: Observed wind power output (% of maximum); Middle:
Synthetic wind power output (% of capacity) based on scaled wind speed data; Bottom: Hourly deviation of
synthetic from observed time series.
10%
20%
30% 40% 50% 60% 70%
Non-Exceedence Hours (%)
90% 100%
Figure 6. Cumulative distribution of the hourly deviation (in absolute percentage points relative to maximum
output) between observed and synthetic generation time series.
Indirect Emissions Analysis • 13
-------
While there are clearly significant discrepancies between these two datasets, we feel that
the NCDC meteorological dataset is a reasonable, if by no means ideal, resource for
estimating temporal patterns and potential offsets from projects which might be built in
each eGRID region.
The hourly synthetic output generated for each of the eGRID subregions is shown in
Appendix B as 24-hour by 365-day color intensity plots. Monthly summary tables of raw
and scaled wind speeds, as well as synthetic wind power, may be found in Appendix C.
The hourly data are available in electronic format upon request.
B. Landfill Gas
To determine the operational and dispatch characteristics of landfill-gas-fired generating
facilities, we contacted several landfill gas (LFG) project developers.12 We received
hourly generation data for several landfill gas projects from Granger Electric Company
and Granger Energy, LLC, ("Granger") which together own and operate six landfill gas
projects in Michigan, and 10 LFG projects in Indiana, Ohio, and Pennsylvania.
According to Granger, the generation profile of LFG projects generally does not differ
significantly by time or season.13 The primary factors that impact the energy output of
LFG projects have to do with ambient conditions. The warmer and the wetter the climate,
the higher the methane content of the landfill gas, such that more power can be produced
from this resource.
High temperatures also affect the plant's output by generally reducing the temperature
gradient that drives the turbine. This temperature effect is minimal for the reciprocating
engines which are used by Granger. Gas turbines appear to be the predominant choice
for LFG projects by other companies. Gas turbine outputs are more susceptible to
variation in temperature. This characteristic largely mirrors the behavior of base load
generating plants, and system operators can reasonably expect a constant and
predictable stream of energy output from these facilities. Thus we conclude that the
fossil-fuel-fired generation displaced by landfill gas in any region of the country will be
similar in operational and emissions characteristics to typical base load resources in that
same region.
C. Municipal Solid Waste
Municipal Solid Waste (MSW) generators, also known as waste-to-energy facilities, are
currently operating in 27 states and burning roughly 95,000 tons of garbage each day,
generating roughly 2.5 GW of electricity.14'15 Even with this large volume of output, the
Landfill gas project operators were identified by Rachel Goldstein of the EPA Landfill Methane Outreach Program:
http://www.epa.gov/lmop/.
Granger does have one project which operates as peaking generation, but this appears to be an anomaly and does not
affect our conclusions about general LFG generation characteristics.
14
New York State Department of Environmental Conservation, http://www.dec.nv.aov/chemical/8979.html.
Because some portion of the fuel supply for MSW generators is generally fossil-based, there is some dispute over
whether this resource should be considered "renewable" or not. However, as a number of state programs include MSW
as a renewable energy source (see dsireusa.org for updated details), we treat it as a renewable resource for the
purposes of this report.
Indirect Emissions Analysis • 14
-------
facilities are not fuel-limited; our research and discussions with facility operators reveals
that waste-to-energy plants run at high capacity factors of 85% and above. As with landfill
gas projects, these characteristics suggest that they serve as base load capacity, and
that the fossil generation they are likely to displace is also base load capacity. Thus we
conclude that the fossil generation displaced by MSWin any region of the country will be
similar in operational and emissions characteristics to typical base load resources in that
same region.
3. Hourly generation and emissions data
In this section we characterize the operational, dispatch, and emissions characteristics of
conventional power plants in order to predict the change in dispatch and emissions
resulting from the presence of new renewable energy resources.
Electricity grids are often characterized as having a range of resource types including
base load, intermediate, and peaking resources. These resources respond differently to
variations in load over different timescales: put simply, base load resources run most
hours regardless of load levels, intermediate resources ramp up and down frequently in
response to changes in system load, while peaking resources run only during the highest
load periods. However, real-world electric grids are more complicated than this, and real-
world resources have complex operating constraints and practices depending on factors
including heat rate, ramping capability, demand and price in the energy and ancillary
service markets, availability of competing resources, local transmission constraints,
environmental constraints, operator or dispatcher discretion, and even the warrantee
and/or service contracts on specific pieces of generating equipment. It is thus impossible
to predict with specificity which resources would respond to perturbations in load or
available low-cost energy. However, historical data on unit operations can lend insight
into which units are more or less likely to be displaceable at any point in time.16 To the
extent that these operational data are coupled with emissions characteristics, they can be
used to estimate the likely displaceable emissions at any point during a historical year.
Such an approach would be more refined than using a simple average emission rate at
any point in time to represent displaceable emissions, because it would be more sensitive
to the actual dispatch characteristics of individual generating plants.
To implement this approach, we obtained generation and emissions data for United
States fossil-fuel powered generating plants for 2005 from the Hourly Emissions
database of the EPA Clean Air Markets Programs.17 The database consists of hourly
reported generation, heat rate, and total emissions of CO2, SO2, NOx, and mercury for all
fossil fuel-fired generators with nameplate capacities above 15 MW. These data form the
basis of all of the displaceable emissions analysis reported here. We use the hourly
It is common to think of generating resources as responding only to changes in electricity demand; however, certain
intermittent renewable resources, such as wind generation, change the load on fossil resources in a very similar manner
to variations in load. Thus we treat the availability of energy from such resources in the same manner that we would a
reduction in energy demand from consumers.
EPA, 2007. Data Sets and Published Reports: Emissions Data Prepackaged Data Sets. Online at
http://camddataandmaps.epa. gov/gdm/index.cfm?fuseaction=prepackaged. select
Indirect Emissions Analysis • 15
-------
generation data to characterize the operational characteristics of generating units in each
NERC subregion, and we use the associated emissions data to characterize the indirect
emissions benefit that would be associated with displacing the output of each generating
unit.
Characterization of emission rates
Hourly emissions are reported in the Clean Air Markets database for CO2, NOx, and SO2.
Within the database, the data records include fields to indicate whether the reported
emissions are measured or calculated. Calculated emissions are based on the
composition of the fuel, the unit's heat rate, and the output of the unit. In general, it
appears that most of the emissions data in the database for 2005 are measured.
Pollutant emission rates are reported in the hourly emissions database as tons per million
British thermal units (mmBtu) for CO2, and pounds per mmBtu for NOx and SO2. For the
purposes of this analysis it is necessary to consider emissions in units of mass per MWh
output. However, while emission rates per unit of heat input may be fairly constant,
emission rates per unit of energy output can vary considerably—this variation is due to
changes in the efficiency (or "heat rate") of a generating unit with different output levels.
For purposes of this analysis, we will use the overall average emissions rate for each unit
as derived from the EPA hourly emissions database, unless otherwise noted.
Data filtering
The EPA emissions data contain anomalous features which are indicative of either
reporting or calculation errors. While the hourly generation data are used to characterize
generator behavior and identify times at which the generator is operational, emission
rates could be misleading if anomalous data were included. Therefore, data are removed
from the emission rate analysis according to the following rules:
• Generation, CO2, or heat rate is reported as zero.
• The reported heat rate is in the top 99.5th or bottom 0.5th percentile (for a record with
8,760 non-zero hourly values, this would remove the 44 highest and 44 lowest
values).
• CO2 emission rate is above 2.5 tons of CO2 per megawatt-hour (tCO2 MWh"1) or less
than zero.
We did not filter on the basis of NOx or SO2 measurements as we had no basis for
judging reasonable emission rate limits.
Indirect Emissions Analysis • 16
-------
4. Emissions Displacement Methods
In order to calculate the indirect emissions benefits associated with power production
from renewable resources, it is necessary to determine which power plants will have their
output curtailed as a result of the availability of the new resource. This need arises from
the recognition that the emission rates of power plants which will be running in either
case (with and without the new resource) do not affect the calculation; nor do those of
power plants that would not be running in either case. In fact, it is only a small number of
plants, as few as one marginal resource in any given dispatch interval, that will determine
the relevant emission rate. The difficulty comes in identifying which plant or plants to
consider, and thus in identifying the appropriate emissions rate or rates for each interval.
There are numerous factors that affect which power plants operate, are committed to
operate, or are held in reserve at any point in time. These factors include:
• Offer price or short-run marginal cost;
• Constraints such as ramp rates, minimum up and down times, and other operating
restrictions;
• Transmission or other dispatch constraints;
• Maintenance schedules;
• Unplanned outages and/or deratings; and
• Environmental constraints.
Unfortunately, most of this information is not available from public sources for most
resources. Using only the information which we have available from the EPA database
(i.e., hourly generation, heat rate, and emissions), it is impossible to reconstruct all of the
factors that guide dispatch and to determine the precise marginal emissions rate in effect
for any particular hour.
Our solution is to infer operational characteristics from the suite of units in each region
based on their patterns of operation and emissions over the course of the study year.
There is a broad range of possible approaches to interpreting the data in this manner.
Some of these approaches use emissions characteristics aggregated over a season or
year, and some are more directly focused on hourly analysis.18 Each approach carries its
own implicit model about system operation and response to new resources, and each
may be more appropriate than others to address the impact of certain kinds of resources
or over specific time scales. In this analysis, we present five methods, ranging in
simplicity, design, and conceptual model, and explore the implications of each. These
methods are:
Earlier investigations based on extracting hourly performance information from the same EPA database used here:
• Connors, S., K. Martin, M. Adams, E. Kern, and B Asiamah-Adjei. 2005. "Emissions Reductions from
Solar Photovoltaic (PV) Systems" Publication MIT LFEE 2004-003 Report.
• Berlinski, M. and S. Connors, "Economic and Environmental Performance of Potential Northeast
Offshore Wind Energy Resources: Final Report", Report to the Offshore Wind Collaborative Pilot
Research Projects, January 2006.
Indirect Emissions Analysis • 17
-------
• Hourly Average Emissions Rate (HAER);
• Annual (and seasonal) emissions slope factor (Slope Factor);
• Empirical Incremental Emissions Rate (EIER);
• Load-Following Rate Incremental Emissions Rate (LFIR); and
• Flexibility-Weighted Hourly Average Emissions Rate (FW-HAER).
A. Limiting Cases
Before delving into each method in detail, it is useful to characterize the most extreme
emission scenarios to bound the results of our analysis, particularly with respect to CO2.
In producing a single MWh of electricity, power plants fueled by coal (the most carbon-
intensive fossil fuel) typically produce somewhat over one ton of CO2; modern plants
fueled by natural gas produce about 0.6 tons of CO2 per MWh.19 The exact emission rate
for each plant depends on technology, age, maintenance, and specific fuel (i.e., some
coals have a higher carbon content than others for the same energy output.) The ranges
of reported hourly emission rates for gas, coal, and oil-fired plants in the EPA database
are shown in/igure 7r While this figure highlights some of the anomalous data in the
database, such as coal plants which report unrealistically low CO2 emissions, it does
show that the reported data are generally consistent with the expected range of emission
rates by fuel type.
35%
30%
0%
0.4
0.6
0.8 1 1.2
Tons per MWh
1.4
1.6
1.8
Figure 7. Distribution of CO2 emission rates for fossil generating plants by reported fuel type.
Note that reported primary fuel is not a precise categorization, as some units are capable of
switching fuels and some plants may actually have units which burn different primary fuels.
The difference between emission rates from coal and gas essentially bounds the range of
outcomes in accounting for the displaced emissions associated with renewable
19
http://www.eia.doe.gov/cneaf/electricitv/page/co2 report/co2emiss.pdf:. Values in the EPA database confirm these numbers:
the median emission rate for a coal plant is 1.024 tCO2 MW1, while for a gas plant it is 0.697 tCO2 MW1
Indirect Emissions Analysis • 18
-------
generation. Renewable resources cannot displace more carbon emissions than the
emission rate of an inefficient coal plant on a per-MWh basis; nor is it likely that they will
displace less CO2 for each MWh generated than the amount emitted by an efficient
combined cycle gas plant. This is because other low-carbon sources of power (such as
nuclear and hydro) have low running costs and are unlikely to be displaced.2021 In
general, the result will lie somewhere between these two extremes.
For NOx, the story is somewhat more complicated. NOx emissions tend to be bifurcated
into ozone-season emissions (May through September), when emissions are regulated
and many power plants are operating emissions control technology; and non-ozone
season emissions, when they are not. As will be shown below, in some areas this
bifurcation is more pronounced than in others. However, NOX emission rates generally
range from zero to perhaps five or six pounds per MWh, with ozone season rates
generally one or two pounds below non-ozone season rates.
SO2 emission rates depend upon the fuel used for generation, with natural gas having no
sulfur (and therefore producing no SO2) and certain types of coal having the most. Thus if
only gas generation is displaced, there is no displacement of SO2 emissions. For this
reason, we would expect SO2to have perhaps the largest regional displaceable
emissions differences of all of the pollutants considered, with ranges from zero up to
about eight pounds per MWh for regions that rely on coal resources for load following
capacity.
To narrow these ranges, it is necessary to find a means to identify which types of plants
are more likely to be displaced in any given region. All of the methods described in this
section are designed to refine the estimate of displaced emissions based on the
observed operational characteristics of the system as deduced from the EPA database.
For clarity, we will develop the displacement methodologies with primary reference to
carbon dioxide emissions since it is the primary anthropogenic cause of global warming.
However, results will be presented for all three of the pollutants considered here.
B. Hourly Average Emissions Rate (HAER)
One of the most conceptually straightforward methods for estimating displaced emissions
is to use a simple hourly average. This method calls for adding the aggregate emissions
from an eGRID subregion (for example, in tCO2) for each hour of the study period and
dividing this by aggregate fossil generation (MWh) for the same hour:
HAER, = - — (3)
As noted earlier, displaced emissions are not necessarily equivalent to avoided emissions, as emissions policies and
other factors can come into play in determining total emissions in any given area.
21
There are exceptions to this statement - in some cases storage hydropower may be the most flexible resource
available to follow load. However, storage hydro is energy-limited resource, so total generation from such a facility will
not be affected by the presence of a new renewable resource.
Indirect Emissions Analysis • 19
-------
Where t denotes a particular hour, and the summations are over all of the generating
units in the region of interest.
Equation 3 produces the average emissions rate across all fossil units for a specific
region and hour, with the implicit assumption that every fossil plant running during that
hour has an equal probability of being displaced. This is clearly not the case on an hour-
to-hour basis: certain units have limited ramping capability and other operational and
economic constraints which dictate that they only be used as base load resources.
However, this method is fairly straightforward to understand and apply, and as noted
earlier we cannot know for sure which units would be displaced in any hour. In addition,
as the market matures with higher levels of renewables penetration, it is reasonable to
imagine that units throughout the supply curve may ultimately be displaced. Finally, this
method may be the most appropriate for base load renewables such as landfill gas or
municipal solid waste combustion, for which output is stable and predictable. Thus it may
be reasonable to quantify displaced emissions using this approach if more of a long-term
or base load impact is under consideration for any of these reasons.
^Figure 8,.shows selected representations of the variations in HAER for CO2 in the New
England (NEWE) eGRID subregion. The first representation is a timeline of hourly
variations, demonstrating that the CO2 emissions rate fluctuates between about 0.65 and
0.95 tCO2 per MWh, with an average of 0.73 tCO2/MWh and a standard deviation of 0.05
tCO2/MWh. the second panel from the top shows a three-week period in the summer in
more detail. The second representation (third from top) is a histogram of hourly HAER
values. The final representation (bottom panel) recasts the dataset as a color intensity
image of HAER, with day of year (DOY) in 2005 along the horizontal and the hour of the
day on the vertical. The green / yellow band across the top and bottom of the image
indicates that the evening hours have higher average emission rates, probably reflecting
the lower contribution of natural gas units during these hours.
We can see from these graphs that there is significant diurnal variability to the data over
a fairly narrow range, with higher hourly average emission rates at night. This reflects the
low position in merit order for coal plants with higher CO2 rates, which dominate the
lower-usage hours at night, but also the fact that these same large plants strongly
influence the overall average, in combination with lower-emitting gas plants, even during
peak times.
Indirect Emissions Analysis • 20
-------
>
Hourly Average Emissions Rate (HAER) in New England (NEWE)
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan
0.9
8 0.8
QL 0.7
LU
I 0.6
07/18
07/20
07/22
07/24
07/26
07/28
07/30
08/01
08/03
08/05
08/07
IOUU
1ROO -
i4nn
o
o -lonn
C
(A
"- 1000 -
o
"s snn
o
^2
ERnn
z
400 -
9nn
n -
1
ll..
Emissions Rate Bin (tCO2/ MWh)
Hourly Average Emissions Rate (HAER) in New England (NEWE)
150 200
Day of Year (2005)
Figure 8. Representations of CC>2 Hourly Average Emissions Rate (HAER) in New England. Top: full record,
with detail shown for three weeks in the summer of 2005; Middle: Histogram of hourly emissions rates
showing distribution throughout the year; Bottom: 24x365 color representation of emission rates, showing
diurnal cycle and seasonal variations.
Indirect Emissions Analysis • 21
-------
C. Emissions Slope Factor
The emissions slope factor is an empirical estimate of the marginal emissions rate based
on the assumption that the observed linear relationship between emissions and electricity
output from fossil fuel units is the best indicator of the system's response to a change in
load. This factor can be calculated as a simple regression relationship over an entire
year, or on a seasonal basis. Because of seasonal differences in dispatch and operating
constraints, there may be additional value in performing this analysis on a seasonal
basis. For the time period under consideration, total emissions and total fossil generation
for each subregion are summed for each hour, based on the EPA database. When the
hourly pairs of emissions/output data points for a subregion are plotted on a scatter plot,
the slope of the regression line represents an empirical measure of the dependence of
emissions on total electricity output in units of mass of pollutant emitted per MWh
produced. If the correlation is high, this slope provides a reasonable estimate of the
regional, seasonal avoidable emissions factor for a change in load or for the addition of
emissions-free energy. This metric makes no specific assumptions about which generator
will be displaced by a change in system load, but instead relies on a robust diagnostic of
the overall system response to variations in load.
One interesting feature of this approach is that the line of best fit rarely has a y-intercept
of zero. This may seem counterintuitive, as zero generation would surely be associated
with zero emissions. However, zero generation is clearly a domain which is not of interest
for this analysis, and the very lowest generating units on the supply curve are likely to
have emissions characteristics that diverge from those of units that are more likely to be
load-following. Were we to force the regression lines to cross through the origin, we
would obtain a much poorer estimate of the emission factor in each region.
^Figure ^illustrates the application of this method for emissions in the Reliability
First/Central (RFCE) subregion. All three pollutants of interest are shown. CO2 and SO2
emissions can be characterized as having a linear dependence on total MW output, with
R2 values of 0.98 and 0.81, respectively.
The NOx emissions data (Figure 10) do not lend themselves to this simple relationship.
Readily apparent from the Figure are two different trends. These trends represent the
emissions in the ozone season, when most generating units are required to operate
ozone control technology, and the non-ozone season when they are not.22 Finally, there
appears to be an up-tick in NOx emissions during very high load hours during the ozone
season. This up-tick may represent units that disengage their NOx controls to increase
output during very high load hours.
22
Complicating this analysis is the fact that many units were not required to report NOx output during the non-ozone
season in 2005.
Indirect Emissions Analysis • 22
-------
450,000
400,000
— 350,000
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000
MW Output
Figure 9. Determination of the regional Emissions Slope Factor in the RFCE (Reliability First/Central) region.
Total emissions of CC>2, NOx and SC>2 vs. MW output for each hour of 2005 are shown. A linear line of best
fit is calculated for each pollutant, and the slope of this fit determines the annual slope factor for the region.
The bifurcation of the NOx data reflects the differential operations of pollution control equipment during the
ozone (summer) vs. non-ozone seasons.
120,000
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000
MW Output
Figure 10. Bifurcation of the NOx emissions slope factor (pounds of NOX per MW output) for the RFCE
region. Slopes for the ozone season (May through September) and non-ozone season hours are shown.
The difference reflects the required operation of ozone control technology during the ozone season. Also
apparent are excess NOx emissions during high-load hours during the ozone season.
Indirect Emissions Analysis • 23
-------
These figures and statistics suggest that the empirical slope factor approach is a well-
defined and reasonable way to characterize the variation of emissions with load in this
region, especially if some care is taken in subdividing the load to isolate meaningful
trends. Unfortunately, the regression relationships are not as constrained for all pollutants
and all regions. Figure 1 shows the seasonal and annual slope factors for CO2, and Table
2,for SO2, for each eGRID region. Table ^shows analogous results for NOx separated
into the ozone and non-ozone seasons. Also shown are the corresponding R-squared
metrics indicating the strength of the regression relationship underlying each slope factor.
The R-squared values for CO2 are uniformly quite high, indicating strong confidence in
these calculated slopes. For NOx the relationships are also generally quite strong, but for
SO2 the relationships range from very strong to extremely weak. This diagnostic should
be taken into account in applying this approach.
In most cases, the differences between the seasons are small, and some of these
relationships require significant error bars. As a rule we apply seasonal slopes in
performing the emissions analysis, but it is important to keep in mind that we do so with
more confidence in some areas and for some pollutants than for others.
Summary results for the Slope Factor method for all eGRID subregions, including factors
calculated forCO2 (tons/MWh), SO2 (Ibs/MWh), and NOx (Ibs/MWh) are displayed in
^Figure 11rfigure 12, andfiqure 13, respectively.
Table 1. Seasonal and annual emissions slope factors (left) and R-squared values (right) forCO2. Spring =
March-May; Summer = July-August; Fall = September-November; Winter = January, February, and
December.
Region
AZNM
CAMX
ERCT
FRCC
MROE
MROW
NEWE
NWPP
NYCW
NYLI
NYUP
RFCE
RFCM
RFCW
RMPA
SPNO
SPSO
SRMV
SRMW
SRSO
SRTV
SRVC
CO2 Slope Factor
Spring
0.59
0.56
0.69
0.73
0.80
0.97
0.63
0.82
0.66
0.55
0.69
0.78
0.81
0.95
0.74
0.93
0.70
0.66
0.93
0.83
0.81
0.94
Summer
0.53
0.50
0.59
0.69
0.77
0.88
0.65
0.71
0.55
0.48
0.66
0.69
0.67
0.85
0.70
0.93
0.60
0.60
0.86
0.70
0.76
0.79
Fall
0.56
0.53
0.64
0.73
0.83
0.93
0.61
0.78
0.63
0.61
0.60
0.75
0.75
0.89
0.86
0.96
0.66
0.64
0.95
0.79
0.87
0.88
Winter
0.55
0.48
0.67
0.72
0.96
0.91
0.68
0.70
0.64
0.51
0.76
0.81
0.93
0.92
0.89
1.08
0.65
0.64
0.99
0.80
0.88
0.90
Annual
0.58
0.52
0.66
0.74
0.83
0.95
0.65
0.82
0.56
0.53
0.66
0.76
0.71
0.89
0.79
0.97
0.66
0.61
0.94
0.80
0.84
0.85
R-Squared
Spring
0.95
0.91
0.94
0.98
0.94
0.98
0.94
0.97
0.83
0.84
0.95
0.99
0.81
0.98
0.85
0.94
0.87
0.94
0.99
0.99
0.97
0.97
Summer
0.97
0.99
0.99
0.99
0.97
0.99
0.98
0.95
0.97
0.90
0.97
0.99
0.97
0.99
0.92
0.97
0.98
0.97
0.99
0.98
0.97
0.99
Fall Winter
0.94
0.95
0.99
0.99
0.95
0.98
0.97
0.90
0.98
0.97
0.92
0.98
0.91
0.99
0.90
0.99
0.97
0.94
0.99
0.97
0.98
0.98
0.91
0.92
0.97
0.99
0.94
0.94
0.88
0.87
0.91
0.82
0.95
0.98
0.94
0.99
0.93
0.97
0.89
0.89
0.98
0.96
0.97
0.99
Annual
0.96
0.97
0.97
0.99
0.95
0.98
0.95
0.95
0.93
0.92
0.95
0.98
0.93
0.99
0.90
0.97
0.94
0.96
0.99
0.98
0.97
0.98
Indirect Emissions Analysis • 24
-------
Table 2. Seasonal and annual emissions slope factors (left) and R-squared values (right) for SO2.
Region
AZNM
CAMX
ERCT
FRCC
MROE
MROW
NEWE
NWPP
NYCW
NYLI
NYUP
RFCE
RFCM
RFCW
RMPA
SPNO
SPSO
SRMV
SRMW
SRSO
SRTV
SRVC
SO2 Slope Factor
Spring
0.29
0.31
1.49
5.68
4.65
6.59
3.04
2.15
1.51
2.94
6.11
12.99
7.40
13.66
0.06
5.91
1.52
1.41
6.69
8.75
6.56
10.84
Summer
0.19
0.03
0.51
5.08
3.59
4.36
3.26
1.07
1.01
1.59
5.98
6.39
4.77
10.38
1.00
7.02
1.03
1.53
6.58
5.26
6.36
7.29
Fall
0.34
0.33
0.80
5.30
4.18
5.06
1.87
1.25
1.35
3.01
4.93
8.58
6.85
12.02
2.03
7.84
1.52
1.53
6.64
7.28
7.19
9.76
Winter
0.07
0.16
1.52
5.70
6.20
5.22
4.33
0.97
2.07
3.05
7.00
9.73
9.68
14.73
0.82
7.26
2.13
1.91
9.02
9.14
9.23
10.24
Annual
0.06
0.18
0.86
5.19
4.15
4.95
2.98
1.85
0.99
2.07
6.14
8.58
5.87
11.00
1.11
6.59
1.43
1.52
7.08
7.57
7.07
8.60
R-Squared
Spring
0.06
0.05
0.32
0.94
0.70
0.82
0.38
0.74
0.47
0.33
0.84
0.87
0.72
0.76
0.00
0.45
0.21
0.35
0.72
0.81
0.45
0.90
Summer
0.10
0.01
0.16
0.92
0.68
0.82
0.78
0.28
0.70
0.54
0.88
0.79
0.78
0.90
0.19
0.85
0.43
0.62
0.83
0.64
0.70
0.89
Fall Winter
0.11
0.10
0.26
0.93
0.34
0.75
0.27
0.09
0.77
0.75
0.76
0.78
0.79
0.92
0.49
0.90
0.53
0.32
0.89
0.67
0.72
0.90
0.00
0.03
0.27
0.79
0.79
0.74
0.63
0.05
0.83
0.55
0.79
0.85
0.72
0.93
0.15
0.79
0.41
0.36
0.70
0.75
0.72
0.92
Annual
0.00
0.07
0.25
0.91
0.47
0.80
0.40
0.47
0.51
0.50
0.85
0.81
0.72
0.84
0.19
0.79
0.38
0.47
0.79
0.77
0.66
0.89
Indirect Emissions Analysis • 25
-------
Table 3. Ozone season and Non-Ozone season emissions rate slopes for NOx. Ozone season = May
through September.
Region
AZNM
CAMX
ERCT
FRCC
MORE
MROW
NEWE
NWPP
NYCW
NYLI
NYUP
RFCE
RFCM
RFCW
RMPA
SPNO
SPSO
SRMV
SRMW
SRSO
SRTV
SRVC
NOx Slope
Factor
Ozone Non-Ozone
season
0.79
0.20
0.83
2.56
1.37
3.26
1.10
2.17
2.08
1.88
1.61
1.86
2.06
2.10
1.73
4.06
2.21
1.68
1.17
1.40
1.80
1.61
Season
0.92
0.62
0.76
2.63
1.93
4.43
1.34
2.37
1.70
2.24
2.15
3.50
3.29
5.18
2.61
5.56
2.58
1.51
3.55
2.26
4.38
3.60
R-Squared
Ozone
season
0.54
0.21
0.88
0.95
0.77
0.90
0.83
0.88
0.89
0.82
0.93
0.96
0.87
0.95
0.55
0.88
0.94
0.71
0.50
0.80
0.77
0.96
Non-Ozone
Season
0.28
0.42
0.59
0.90
0.64
0.85
0.52
0.61
0.52
0.74
0.82
0.93
0.71
0.87
0.67
0.85
0.74
0.66
0.67
0.67
0.73
0.90
Indirect Emissions Analysis • 26
-------
CN
ai ^
Q.
o
s °
ra *~
o:
W
C O5
.2 CD
W
§0>0
poo2iZw2§wi->
^LLiiSQ-Q-cejSceceK
"-cccecewwc/j^www
Figure 11. Regional slope factors forCC>2 (tons/MWh)
CD
•* _
Q)
Q.
0
55 ?!
0)
ra
« 0
-
c
o
'in
.i2 °°
u
8 w
ro
c
o •*
in
ro
5
CM
Q
X Spring (MAM)
n Summer (JJA) |
+ Fall (SON) x
• Wnter (DJF) ^
o Annual
0 V
D i
• • _ _ ™
B • •
B X ± A * * n
_-x HQt x ft «
B^g. + °
pt.
H *
^•> H~ ^^ ^5 rn 1 ^P IK
K i p$T IQ rn w
Ur fm 1 — 1 vv
i'i'l'g'g'i'si'g^'s's'g'g'i'i'si'i's1^1^1
|c1-£l|zl|^z^^-l8585glggg
Figure 12. Seasonal and regional slope factors for SO2 (Ibs/MWh)
Indirect Emissions Analysis • 27
-------
SL
£
OT
Si
13
o:
o
'in
m
x
O
"ra
o
in
ro
01
V)
X Ozone Season
n Not Ozone Season
D
X
D D
D
9 n a
n
x
X
X
Figure 13. Regional slope factors for NOx (Ibs/MWh)
Indirect Emissions Analysis • 28
-------
D. Empirical Incremental Emissions Rate (EIER)
The previously discussed methods for estimating emissions factors rely upon the
empirical dependence of total emissions on total output over a predefined period, which
could be a year, a season, or any other time period. As a result, they produce emissions
factors characteristic of the selected time period, but not indirect emissions factors at an
hourly level. However, it is possible to take this analysis to its limit and assess the hourly
slope factor by dividing the change in total emissions from one hour to the next by the
change in total output for the same hour. This approach results in the empirical
incremental emissions rate (EIER), which may be expressed mathematically as follows:
EIER _C02(t)-C02(t-l}
' MW(t)-MW(t-V)
The implicit model underlying this approach is that units which are not marginal or
displaceable in any hour will have constant output and emissions, so that the metric will
be sensitive only to units which change their output from hour to hour. This theoretically
weights the metric by the change in output for these "marginal" plants. In contrast, the
hourly average emission rate approach was weighted by the total output of each unit,
even for units whose output is insensitive to changes in load.
While the EIER has the attractive feature that it is a direct estimator of the dependence of
change in emissions on change in total generation for each hour, there are certain
considerations which must be made in its application. First, the estimator does not
distinguish between units that change their output because they are in a normal ramping
cycle, for example in anticipation of daily changes in load from those that change their
output in response to shorter term load fluctuations. This may be problematic as the
former may be large units and exhibit large changes in load, so they are likely to
dominate the metric during the hours when their output varies. Next is a numerical
stability issue: if the change in total generation is small, the calculated EIER could be
dominated by reporting and round-off errors in the emissions data. Thus we choose
(somewhat arbitrarily) to eliminate hours from consideration if the change in total
generation in the region from the previous hour is less than 100 MW. In this case we
carry forward the EIER from the previous hour.
Finally, the EIER can be less than zero in some hours. This phenomenon occurs when
total reported generation output in some region rises or falls during an hour by an amount
greater than the 100 MW threshold, but total reported emissions changes in the other
direction. This is not necessarily reporting error; it may well be that changes in dispatch,
due to transmission constraints or some other consideration, have enough of an impact
on emission rates to offset the effect of changing total output. This instability, and the
wide scatter in the hourly results for this approach, cast some doubt on the value of this
approach for predicting hour-to-hour indirect emissions benefits associated with
renewable energy resources.
Indirect Emissions Analysis • 29
-------
E. Load-Following Incremental Emissions Rate (LFIR)
The load-following incremental emissions rate is our name for a hybrid of the hourly
average emissions rate and the empirical incremental emissions rate introduced by
Stephen Connors in a number of studies of avoided emissions in the United States.2324
The implicit model underlying the LFIR is that units which change their output in the same
direction that load is changing during any dispatch period can be treated as load
following. Thus the LFIR approach is designed to yield the average emission rate of the
units that change output in the direction of changing load during each hour under
consideration. However, as Connors notes,25 units which are held for spinning reserves
or automatic generation control may actually be load following units, and would be
missed by the first criterion. Thus the list of "load following" units must include any unit
whose last change in output has been in the direction of the change in total load, even if
subsequent hours have produced only small changes in output (less than 2.5% of their
maximum.) Finally, Connors examines the range of unit behavior and finds that any unit
whose output is between 55% and 90% of its capacity is likely to be providing spinning
reserve service, and designates these units as load following as well.26
Following this logic, we find that a large proportion of generating units on the system
most of the time would be designated as "load following" in the 2005 EPA database; in
many regions, one-third to one-half of all resources earn this designation more than 50%
of the time. This brings the effect of this approach close to that of the HAER approach
based on all units running, despite the significantly more complicated analysis required.
At the same time, because the units (and hence emission rates) may change from hour-
to-hour even when total load is changing only slightly, this method suffers from even
greater numerical instability than the EIER approach.
Connors describes the logic for selecting load following units as "excessively inclusive"27
in designating too large a proportion of available units in each hour. To remedy this, he
weights each unit's contribution for each hour by its MW change in output from the
previous hour. This method may serve to mitigate some of the distortion from large base
load resources that happen to make it into the list of "load-following" units by
coincidentally ramping along with load, but only for the hours in which they do not happen
to be ramping. For the hours when they are, they will clearly dominate the calculated
average, as their changes in load will be much greater than that for smaller, truly load-
following units.
In summary, we find that the LFIR approach suffers from many of the same distortions as
both the EIER approach, and the HAER approach. The LFIR is numerically unstable,
producing emissions rates which are highly and unrealistically variable throughout the
year, while at the same time it gives undue weight to large units which are likely to have
23 Connors et al. (2005).
24
Berlinski, and Connors (2006).
25 Ibid., p. 1-4.
26 Connors et al., 2005, Table 1.1, p. 1 -6.
27 Ibid, p. 1-10.
Indirect Emissions Analysis • 30
-------
larger changes in output from hour to hour but are not necessarily likely to be load-
following. Although we have calculated and tabulated the estimated indirect emissions
benefits using the LFIR for comparison, we conclude that it does not produce a useful
metric for estimating displaced emission factors on electricity grids.
F. Flexibility-Weighted Hourly Average Emissions Rate (FW-HAER)
The final method considered, FW-HAER is also based on the premise that only certain
units are likely to be displaced by variable output renewable energy resources. This
approach attempts to identify load following units based on their operational behavior
over the course of the entire study period, rather than during individual hours as with the
LFIR method. The FW-HAER approach assigns to each unit a score based on how
readily it appears to shift output; this score is then used as a weighting coefficient in
calculating the indirect emissions coefficient for each hour.
While it is impossible to determine exactly which units would be load following at any
given time given the complexities of dispatch in real electric systems, it is reasonably
straightforward to determine which tend to behave in a more or less flexible manner.
Large base load units are generally scored lower than smaller units. Units that appear to
spend a lot of time ramping in response to load during the year, or that are often found
partially dispatched, are ranked as highly displaceable whenever they are running.
To capture these dynamics, the FW-HAER approach tracks how often a plant is in the
process of ramping relative to the number of hours it operates during the year, as shown
in Equation 5:
77 _ ramping,i
' ~ N
operating,!
Where F is the flexibility coefficient and i refers to an individual unit. A "ramping" hour is
defined as any hour in which the change in the unit's output is greater than or equal to
2.5% of its maximum capacity; by dividing the number of ramping hours by the total
number of hours in which the unit operates, we obtain a unit flexibility coefficient which
represents the proportion of its operating hours which appear to be in "ramping" mode.
Once the flexibility coefficient of each unit on the system is calculated, the indirect
emissions rate for each hour of operations is calculated as shown in Equation 6:
(6)
Where ERit is the emission rate for unit i during hourt and the summations are overall
units operating during that hour, and F, is as in Equation 5. Again, this method does not
weight by either unit size or by the change in output in any given hour. Only units which
appear to be more flexible in their output over the year have increased influence on the
displaceable emissions rate. This approach avoids giving undue weight to large, inflexible
Indirect Emissions Analysis • 31
-------
units, and also exhibits substantially less volatility than methods that are based on hourly
changes in output.
The FW-HAER approach has several advantages for estimating the emissions impacts of
variable output resources. It does a much better job than any other approach in
narrowing the list of candidate units that dominate the indirect emissions calculation.
Unlike the other approaches considered here, the FW-HAER avoids giving undue weight
to large, inflexible units. FW-HAER is much more numerically stable than either the EIER
approach or the LFIR approach, yielding a much smaller range of displaceable emissions
rates through the year. Finally, because of this stability and more refined identification of
the most flexible units in any region, it does the best job of differentiating regions by their
true indirect emissions rates. These features are demonstrated in the figures below.
^Figure l^shows the relationships among flexibility coefficient, generator size, and
number of operating hours for (a) New England and (b) California. The size of each circle
represents the observed number of operating hours in 2005. In both regions, we see that
the very highest flexibility coefficients are attributed to small units which run infrequently;
when running, these may also be the first units displaced. In California, it is apparent that
the largest units are not displaceable; in New England, there does not appear to be much
of a relationship between unit size and flexibility.
^Figure l^shows hourly generation profiles for three sample plants in Florida, exhibiting
typical profiles for low, medium, and high flexibility coefficients.
Indirect Emissions Analysis • 32
-------
(a)
NEWE
1
0.9
O.B
0.7
C
| 06
:& 0.5
'x
il 04
0,3
02
0.1
n
o
8
D
o O
-. o o
V^/ r~\ ( }
o /T^.
O o oO ~O
o ^ O r
CD ' s*$i t-C^~^ C^)
o W Q~/O °
o o
.
o o ^
QO OQ Q o
1 © o o o
0 100 200 300 400 500 600 700
Hours in operation
• <100
O 8^60
(b)
CAMX
1
0.9
n A
U.o
0.7
c
1 °6
3? 0 5
j=
'x
~ 04
03
0,1
n
Hours in operation
• <100
O 8760
% „
" 1 °
o
- ftO o O
8 ° ° cP
" |° °$> o
' 0° 3 Q°^
p\ ^o yi) °
- c9 o o (t[v^ yfj o
o ° rt3\ ° O ^ o O
-0° o^^^ 0
r>9 (C'
O° ^ 0
0
0
800 900 1000 0 100 200 300 400 500 600 700 800 900 1000
Generator maximum capacity (MW)
Generator maximum capacity (MW)
Figure 14. Flexibility coefficient vs. generator size in New England (NEWE) and California (CAMX). Each circle represents a single generating plant in the indicated
region, and the size reflects total operating hours in 2005. Maximum size is 8760 hours.
Indirect Emissions Analysis • 33
-------
(a)
200
.1 100
£
0)
I „
Jan
200 r
o 100
07/28
^h
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
07/31
08/03
08/06
08/09
08/12
(b)
1000
.2 500
Jan
1000
?
1,
I 500
0)
I „
07/28
Feb
Mar
Apr May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
T T
I \ I
07/31
08/03
08/06
08/09
08/12
(C)
100
° 50
100
o 50 -
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Jan
- ft f I
! 1 J
/I A
,
f
ft
i
\
i
!l fi
1
\
07/28
07/31
08/03
08/06
08/09
08/12
Figure 15. Hourly dispatch and flexibility coefficient (F) for three plants of differing capacities in the FRCC
subregion. Each lower graph is a summer detail of the 2005 profile shown above, (a) Polk power plant, F =
0.08; (b) Manatee gas plant, F = 0.55; (c) Debray peaker, F = 0.86.
Indirect Emissions Analysis • 34
-------
G. Comparison of Indirect Emissions Rates
It is instructive to compare the implications for indirect emission rates from each
estimation approach. ^Figure 1 ^displays these results in three panels, for CO2, SO2, and
NOx, respectively. Each panel is organized vertically by region, with indirect emission
rates on the horizontal axis. For each region, the range (single standard deviation around
the mean) of the hourly indirect emission rates may be read and compared for each of
the calculation approaches considered. Methods depicted are the empirical incremental
emission rate (EIER, solid line), the hourly average emission rate (HAER, angle
brackets), and the flexibility-weighted average emission rate (FW-HAER, shaded
po
rectangle). Also shown is the annual (CO2, SO2) or seasonal (NOx) slope factor for
each region, which is shown as a point as it is not an hourly value but a metric calculated
from the annual or seasonal data.
The graphs infigure 16^ illustrate the variations among regions, and within each region
between the different calculation approaches. For example, the leftmost panel illustrates
that for CO2, there is general agreement among the different calculation approaches in all
regions, although the slope factor tends to suggest a lower indirect emissions rate than
other methods. The HAER and the FW-HAER approaches both yield values in a fairly
tight range and are generally in agreement with each other, while the EIER approach
yields a much broader range of hourly values reflecting its numerical instability. It is also
clear from/igure 1^,that certain regions, such as the Midwest areas that rely heavily on
coal generation, have indirect emissions rates of over 1.0 to 1.2 tCO2 per MWh, while
regions such as California and New England, which rely more heavily on gas, have
indirect emissions rates of perhaps 0.6 to 0.8 tCO2 per MWh.
Color intensity plots for each hourly method for each pollutant, and for each region are
provided in Appendix D. These plots display the diurnal and seasonal variations in each
displacement factor derived using each approach. The data are also available in
electronic format.
We have not plotted the single-standard-deviation range based on the LFIR approach in Figures 14-16 because it is far
too broad for this scale, reflecting the numerical instability of this approach.
Indirect Emissions Analysis • 35
-------
RFCE
RFCM
SRVC
[
(a)
• 1 — a *
*rr — i -
Bs .,
S~~cl •-.
]_< — >j
*S J
\ 1* < >
* 'LT^J.
1 A f i
fr ^ s i
£ 1 •?
*i —
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Indirect emissions coefficients (t C02 / MWh)
AZNM
RFCE
RFCM
SRVC
(b)
• n f •-.
Bs
\ • I
*• >
f~< *
•i — ^j >
A.-- --.i
r*i < i
a,- -.
• I < > I
* i< >
0 2 4 6 3 10 12 14
Indirect emissions coefficients (Ibs S02 / MWh)
AZNM
RFCE
RFCM
SRVC
-1
(c)
r— i
Brf-VH
1 <*->
• t Ql>
r~* — i rf **
r~s^ .-"-. -.
FTrSi
K P
2 and 862) or seasonal (for NOx; dark = ozone season) emissions slope
factors for each region.
Indirect Emissions Analysis • 36
-------
5. Indirect Emissions Results
In this section we present our assessment of what the annual indirect emissions
reductions of NOx, SO2, and CO2 would have been in each of the 22 eGRID subregions
in 2005, had an incremental renewable energy project (wind, landfill gas, or municipal
solid waste) produced one gigawatt-hour (GWh) of energy over the course of the year.
As discussed earlier, we find that landfill gas and municipal solid waste have identical
base load energy profiles. Thus we have combined the analyses for these two resources
into a single set of results.
All of the results presented here are based on a one-to-one displacement of fossil
generation, assuming emissions-free generation from the renewable resource. While
beyond the current scope, it would improve precision to incorporate line losses, as well
as consideration of internal energy use at power plants. In general, renewable generation
sites are remote from load centers, meaning that more than one MW of generation is
required to serve one MW of load. Fossil generation sites may be near to or remote from
load. However, all fossil units have some parasitic load—again, more than one MWmust
be produced in order to deliver one MW onto the system. The extent to which these
cancel out in any given region, or the adjustments that should be made to accommodate
them, is beyond the scope of this study.
The calculated indirect emissions results are presented graphically as outlined injable 4,
The results are also tabulated in/able 5, (wind projects) and/able 6,.(landfill gas and
municipal solid waste.)
Table 4. Guide to the summary figures.
Figure
figure 17.
figure 161
figure 19.
figure 20.
Figure 21.
Figure 22^
Resource Type
Wind
Wind
Wind
MSW, LFG
MSW, LFG
MSW, LFG
Pollutant
CO2
NOx
SO2
C02
NOx
S02
Methods
HAER, EIER, FW-
HAER
HAER, EIER,
FW-HAER
HAER, EIER,
FW-HAER
HAER, seasonal and
annual slope factors
HAER, Seasonal Slope
Factor
HAER, Seasonal and
Annual Slope Factors
Regions
All eGRID regions
All eGRID regions
All eGRID regions
All eGRID regions
All eGRID regions
All eGRID regions
The results show significant regional differences in the indirect emissions of all pollutants,
no matter what calculation method is used. They also show that the calculation method
applied makes a large difference in some regions, while in other regions these
differences are less significant. This leads to the question of which method is most
appropriate to use under what conditions.
In our judgment, the FW-HAER best captures the short-term operational behavior of
power plants because it reflects observations of short-term changes in output for
individual plants which respond flexibly to system perturbations. We recommend this
approach in particular for shorter-term indirect emissions benefits forCO2, in response to
Indirect Emissions Analysis • 37
-------
variable output, non-dispatchable resources such as wind power. The regional FW-HAER
values presented in this paper can be applied for this purpose, where the definition of
short-term depends on the application but may be the first two or three years of a new
wind resource lifetime.
Over the longer term, and with greater penetration of wind resources, it is reasonable to
assume that the system will adjust by displacing resources lower down in the dispatch
order, ultimately restoring the utilization rate of load-following capacity to accommodate
fluctuations in both load and wind resource output. In addition, NOxand SO2 are both
regulated under regional or national cap-and-trade regimes which make hourly indirect
emissions analysis less relevant for these pollutants. Thus for longer-term indirect CO2
analysis, or for NOx and SO2, we recommend using either HAER or the emissions slope
factor. These approaches are based on the longer-term response to perturbations over
the full resource portfolio, as opposed to reflecting primarily the marginal unit or units.
These two approaches (HAER and slope factor) give generally similar results. The
primary difference between them is that the HAER size-weights base load fossil
resources running at any given hour, giving the most weight to large units regardless of
their flexibility in response to changes in output. Conversely, the slope factor method
tends to deemphasize continually operational resources regardless of size. As we find it
unlikely that the largest and lowest running cost resources would ever be displaced by
renewable energy projects, we prefer the slope factor approach for those regions and
pollutants for which the slope factor is well defined.
For landfill gas and municipal solid waste resources, the most likely type of generators to
be displaced are intermediate units, which are more costly to run than base load but
which do not have the flexibility of load-following resources. In this case, the best method
would again be either HAER or the slope factor approach, and we would again give slight
preference to the slope factor approach if the slope is well-defined.
Finally we note that most of our results, especially those for CO2, are relatively insensitive
to hourly or seasonal variations in wind power generation. Replacing the calculated wind
profiles with a flat power output profile (such as that characteristic of landfill gas or
municipal solid waste) results in small changes in the calculated indirect emissions
impact over the course of a year. Table 8 shows this result as the percent change in the
total annual indirect emissions impact for wind resources if the hourly profile is replaced
by a flat profile, typical of an LFG or MSW facility. The small changes suggest that
indirect emission factors and wind output profiles are generally poorly correlated, and in
many cases little is gained from the effort to apply realistic profiles.
However, some of these differences are larger than others and probably important—for
example, in some regions the impact of using a realistic wind-based shape on indirect
NOx emissions is as high as 10%. This different probably reflects seasonal differences in
wind strength which are correlated with seasonal differences in the operation of NOx
control equipment. Whether the effort required to establish realistic wind profiles pays
sufficient dividends in this type of analysis to be justified probably depends on the specific
application.
Indirect Emissions Analysis • 38
-------
Table 5. Annual indirect emissions reduction rates associated with an incremental GWh of wind energy in each of the eGRID subregions for 2005. Calculation
methods shown are: HAER: Hourly average emission rate; FW-HAER: Flexibility-weighted hourly average emission rate; EIER: Empirical incremental
emissions rate; LFIR: Load-following incremental rate; Slope factor Indirect emission rate based on seasonal relationship between emissions and output
across all hours in each season. For CO2 and SO2 the seasonal slope factors are based on conventional 3-month seasons; for NOx the slopes are defined for
the ozone and non-ozone regulation seasons.
eGrid
Region
AZNM
CAMX
ERCT
FRCC
MROE
MROW
NEWE
NWPP
NYCW
NYLI
NYUP
RFCE
RFCM
RFCW
RMPA
SPNO
SPSO
SRMV
SRMW
SRSO
SRTV
SRVC
2005 Avoided CO2 (tons/GWh)
Slope
HAER EIER LFIR FW-HAER Factor
844 573 591 738 943
670 497 496 619 892
845 606 596 684 1,040
845 694 689 757 994
1,071 803 807 1,054 1,114
1,094 885 884 1,060 1,079
728 588 637 711 780
981 716 657 741 1,076
800 616 657 717 742
662 530 621 687 744
731 664 698 875 816
854 724 731 911 896
1,059 835 773 1,072 959
954 885 854 948 906
980 661 612 734 1,056
1,066 912 906 1,040 1,074
905 618 629 735 1,062
765 659 653 686 1,027
996 905 900 1,001 966
960 818 826 914 997
1,005 856 882 987 974
1,035 955 837 974 933
2005 Avoided NOx (Ibs/GWh)
Slope
HAER EIER LFIR FW-HAER Factor
2,768 1,335 1,260 1,641 927
1,238 259 287 411 508
1,058 863 877 785 775
2,645 2,811 2,757 1,675 2,584
2,981 1,902 2,346 4,380 1,810
4,456 3,407 3,388 4,461 4,101
1,177 1,004 944 913 1,253
3,267 2,400 2,101 2,310 2,547
1,821 2,073 957 1,206 1,846
1,623 2,277 1,160 1,317 2,158
2,016 1,991 1,864 2,172 1,967
3,062 2,691 2,485 3,039 3,009
3,194 3,277 3,057 2,988 2,912
3,579 3,852 3,496 3,435 4,219
2,963 1,675 1,011 1,805 2,377
4,060 4,019 3,203 3,955 4,966
2,642 2,662 2,610 2,151 2,466
1,638 1,912 1,900 1,407 1,616
2,664 2,415 2,489 2,717 2,839
2,769 2,300 2,504 2,919 2,089
3,755 3,270 3,252 3,708 3,499
3,349 3,177 2,685 3,405 3,029
2005 Avoided SO2 (Ibs/GWh)
Slope
HAER EIER LFIR FW-HAER Factor
1,782 1,335 1,049 801 218
1,138 259 202 139 216
3,920 863 1,300 1,074 1,116
4,857 2,811 5,254 3,519 5,454
7,678 1,902 5,147 9,310 4,738
6,887 3,407 5,075 7,850 5,451
4,177 1,004 2,691 2,737 3,126
2,546 2,400 (1,085) 1,493 1,350
1,452 2,073 1,353 1,251 1,545
4,204 2,277 2,804 3,513 2,773
6,277 1,991 6,367 9,737 6,015
13,596 2,691 9,117 12,842 9,881
9,363 3,277 6,448 7,511 7,420
12,250 3,852 11,037 13,443 12,994
1,848 1,675 708 1,699 895
6,949 4,019 6,910 8,953 6,909
3,912 2,662 1,454 1,331 1,531
2,693 1,912 2,148 1,404 1,611
6,739 2,415 6,348 7,941 7,332
11,308 2,300 9,293 10,796 7,909
9,004 3,270 9,463 12,374 7,469
11,289 3,177 9,302 12,135 9,902
Indirect Emissions Analysis • 39
-------
Table 6. Annual indirect emissions reduction rates associated with an
incremental GWh of landfill gas of MSW generation in each of the eGRID
subregions for 2005. Calculation methods shown are: HAER: Hourly average
emission rate; Slope factor: Indirect emission rate based on seasonal
relationship between emissions and output across all hours in each season.
For CC>2 and SC>2 the seasonal slope factors are based on conventional 3-
month seasons; for NOx the slopes are defined for the ozone and non-ozone
regulation seasons.
eGrid
Region
AZNM
CAMX
ERCT
FRCC
MROE
MROW
NEWE
NWPP
NYCW
NYLI
NYUP
RFCE
RFCM
RFCW
RMPA
SPNO
SPSO
SRMV
SRMW
SRSO
SRTV
SRVC
2005 Avoided
C02 (tons/GWh)
Slope
HAER Factor
838 943
667 893
852 1 ,042
859 995
1,086 1,117
1,098 1,080
732 777
982 1 ,078
795 740
660 740
736 812
862 898
1,058 958
956 905
980 1 ,055
1,069 1,075
909 1 ,062
764 1 ,025
999 970
961 997
1,004 975
1,032 932
2005 Avoided
NOx (I bs/ GWh)
Slope
HAER Factor
2,718 901
1,226 476
1 ,069 775
2,632 2,582
2,996 1,764
4,459 4,022
1,138 1,232
3,274 2,472
1 ,739 1 ,887
1,549 2,110
2,004 1,917
2,904 2,829
3,103 2,787
3,303 3,900
2,952 2,278
4,088 5,012
2,653 2,464
1 ,660 1 ,628
2,458 2,482
2,654 1,976
3,366 3,175
3,096 2,790
2005 Avoided
S02 (Ibs/GWh)
Slope
HAER Factor
1,710 223
1,129 206
3,981 1,080
4,806 5,442
7,834 4,649
6,891 5,307
4,140 3,119
2,553 1,362
1 ,339 1 ,482
4,103 2,645
6,306 6,003
13,720 9,424
9,380 7,164
12,187 12,688
1 ,830 974
6,949 7,005
3,958 1 ,549
2,665 1,593
6,773 7,226
11,269 7,601
8,927 7,327
11,168 9,529
Table 7. "Shape impact" for indirect emissions calculations for each pollutant
in each eGrid region. The shape impact is the change in calculated indirect
emissions if an hourly wind profile is replaced with a constant generation
profile with the same total output.
eGrid
Region
AZNM
CAMX
ERCT
FRCC
MROE
MROW
NEWE
NWPP
NYCW
NYLI
NYUP
RFCE
RFCM
RFCW
RMPA
SPNO
SPSO
SRMV
SRMW
SRSO
SRTV
SRVC
CO2 Shape
Impact
Slope
HAER Factor
-0.7% -0.1%
-0.4% 0.0%
0.9% 0.2%
1.7% 0.1%
1.5% 0.3%
0.3% 0.1%
0.5% -0.4%
0.2% 0.2%
-0.7% -0.2%
-0.3% -0.5%
0.6% -0.4%
0.9% 0.2%
-0.1% -0.1%
0.1% -0.2%
0.0% -0.1%
0.3% 0.1%
0.4% 0.1%
-0.1% -0.2%
0.3% 0.5%
0.1% 0.0%
-0.1% 0.1%
-0.3% 0.0%
NOx Shape
Impact
Slope
HAER Factor
-1.8% -2.8%
-0.9% -6.3%
1.0% -0.1%
-0.5% -0.1%
0.5% -2.5%
0.1% -1.9%
-3.3% -1.7%
0.2% -2.9%
-4.5% 2.2%
-4.6% -2.2%
-0.6% -2.5%
-5.1% -6.0%
-2.8% -4.3%
-7.7% -7.6%
-0.4% -4.2%
0.7% 0.9%
0.4% -0.1%
1.3% 0.7%
-7.7% -12.6%
-4.2% -5.4%
-10.4% -9.3%
-7.6% -7.9%
SO2 Shape
Impact
Slope
HAER Factor
-4.0% 2.2%
-0.8% -4.6%
1.5% -3.2%
-1.0% -0.2%
2.0% -1.9%
0.1% -2.6%
-0.9% -0.2%
0.3% 0.8%
-7.8% -4.1%
-2.4% -4.6%
0.5% -0.2%
0.9% -4.6%
0.2% -3.4%
-0.5% -2.4%
-1.0% 8.8%
0.0% 1.4%
1.2% 1.1%
-1.0% -1.1%
0.5% -1.5%
-0.3% -3.9%
-0.9% -1.9%
-1.1% -3.8%
Indirect Emissions Analysis • 40
-------
1,200 -,
1,000
800
O
O
600
400
200
HAER
DEIER
DFW-HAER
Figure 17. Annual indirect CC>2 emissions impact associated with one incremental GWh of wind energy in each of the
eGRID subregions for 2005.
5,000 -,
4 000
O 3 QQQ
0
Q_
X
z
{/)
c 2 000
o
Q_
1 nnn
500
n
-
...
•
5
1
"i
n
~|
ri
1-
|-|
n
•
-l 1
1
•
-1
r
_
,
_
n r
xi-OLuaiucLaziQ-^J555o?
|I^MIil^lfe^lil«Mis
• HAER DEIER OFW-HAER
n
-
\ 1
Figure 18. Annual indirect NOx emissions impact associated with one incremental GWh of wind energy in each of the
eGRID subregions for 2005.
Indirect Emissions Analysis • 41
-------
16,000 -i
14 000
12 000
5 1 n nnn
0
0
Q_
j^ Q nnn
CO
V)
•Q
C
^ c nnn
Q_
4 000
2 000
n _
Ilk
Tl
1
1
-
-
n
[
1
1
n
r
n
-I ]
-i -
1
n
-l
•
-r
X H O LU
Q-gpQ,LiJ5ggO
IHAER
OEIER
DFW-HAER
Figure 19. Annual indirect SC>2 emissions impact associated with one incremental GWh of wind energy in each of the
eGRID subregions for 2005.
1,200 -,
1,000
800
600
400
200
o
o
HAER
Seasonal Slope Factor
D Annual Slope Factor
Figure 20. Annual indirect CC>2 emissions impact associated with one incremental GWh of landfill gas or municipal
solid waste energy in each of the eGRID subregions for 2005.
Indirect Emissions Analysis • 42
-------
6,000
5,000
4,000
O 3,000
°- 2,000
1,000
I
X H O LU
8-g=]9;W5g<00>gO>0
IHAER
• Seasonal Slope Factor
Figure 21. Annual indirect NOx emissions impact associated with an incremental GWh of landfill gas or municipal
solid waste energy in each of the eGRID subregions for 2005.
16,000 -i
1 4 000
1 2 000
5 10 000
0
Q_
j^ Q nnn
CO
V)
•Q
C
^ c nnn
Q_
4 nnn
2 000
-
1 L
5 X
y <
< o
1
k.
-,
•
-,
•-i
"llfti
^ § § o § | § ^
S uL i K ^ 1 ^ '
"-2 5 z z z
IHAER
1
n
-I
-I
^ >~ LL- L_ n
- z a: it ^
• Seasonal Slope Factor
b
r
D- Z
5 CL
o: co
11
O
73
|
1 n
r
-,
> g O > O
/> ^ CO "5 CO
D Annual Slope Factor
Figure 22. Annual indirect SO2 emissions impact associated with an incremental GWh of landfill gas or municipal solid
waste energy in each of the eGRID subregions for 2005.
Indirect Emissions Analysis • 43
-------
6. Conclusions
In this analysis we have produced and applied a number of approaches for calculating
indirect emissions impacts and provided a general discussion of which might be more
appropriate under different circumstances. We have calculated indirect emissions
benefits associated with CO2, NOx, and SO2 for wind, landfill gas, and municipal solid
waste electricity generating resources, in each of the 22 eGRID subregions of the
continental United States.
The indirect emissions benefit calculated for any kind of renewable energy resource
depends on a number of questions:
• Where is the resource located?
• What is the pollutant of interest? Is there a cap and trade system in place for this
pollutant?
• Is the time period of interest historical, in the near future, or several years in the
future?
• Is the resource base load, dispatchable, or intermittent and nondispatchable?
• If intermittent and nondispatchable, what is the expected hourly and seasonal profile
of the resource?
• We have found that there are important regional differences in indirect emissions
which can be quantified in ways that will be useful for estimating the emissions
benefits of renewable energy projects in each region.
• Because of the wide range of applications reflected in the possible answers to the
questions posed above, the best method to use in calculating indirect emissions is
almost certainly application-specific. We believe that these methodologies provide
useful guidance for many applications, and a better understanding of the issues for
approximation of indirect emissions benefits than has previously been available.
• The results presented here could be validated by a modeling analysis, using a full
electric system dispatch model to simulate actual indirect emissions under realistic
unit commitment and dispatch conditions. The results could then be compared to
those that would be derived using each of the methods applied here. This
comparison would provide an empirical basis for establishing which approach is most
consistent and accurate in predicting indirect emissions in any particular region or
under specific circumstances, at least in the context of the model.
• In summary, we have found that:
• The indirect emissions benefits of renewable energy for all pollutants vary
significantly by region, and these differences can be quantified and applied in
calculating indirect emissions impacts;
• Different calculation approaches are appropriate for different types of resources and
different applications;
Indirect Emissions Analysis • 44
-------
In general, hourly profiles of renewable energy resources make only a modest
difference in quantifying annual indirect emissions benefits;
More research is needed to establish which methods of calculating indirect emissions
benefits most accurately reflect real-world dispatch over a range of timescales and
system conditions.
Finally, we note again that the geographic coverage of this analysis was by necessity
coarse, in the interest of presenting results for the entire continental United States,
particularly with respect to wind resources. For the analysis to be applied to calculate
or predict indirect emissions impacts for specific projects, the tables of hourly
avoidable emissions data in Appendix D could be applied directly to site-specific
operational and meteorological data. However, given the modest impact of
differences in hourly profiles noted above, the annual results presented here are
likely to be sufficient for most purposes.
Indirect Emissions Analysis • 45
-------
Appendix A:
Identification and map of eGRID subregions
Indirect Emissions Analysis • 46
Appendix A
-------
Figure A-1. Map of eGRID subregions. Subregions in Alaska and Hawaii are not considered in this analysis.
Acronyms are defined below. Source of figure: https://www.energvstar.gov/istar/pmpam/help/eGRID_Subregion_Map.htm.
AZNM Arizona and New Mexico
CAMX California
ERCT ERGOT (Texas)
FRCC FRCC (Florida)
MROE Midwest Reliability - East
MROW Midwest Reliability - West
NEWE New England
NWPP Northwest
NYCW New York City/ Westchester
NYLI New York - Long Island
NYUP New York - Upstate
RFCE Reliability First - East
RFCM Reliability First - Michigan
RFCW Reliability First - West
RMPA Rocky Mountain region
SPNO Southwest Power Pool - North
SPSO Southwest Power Pool - South
SRMV SERC - Mississippi Valley
SRMW SERC-Midwest
SRSO SERC - South
SRTV SERC - Tennessee Valley
SRVC SERC - Virginia/Carolina
Indirect Emissions Analysis'
Appendix A
47
-------
Appendix B:
Color intensity plots of synthetic wind power time series (hourly percent of
capacity) for each eGRID subregion. Each grid shows hour of the day on
the vertical axis and days of the year on the horizontal axis. Color scale
for all charts is shown below and at the bottom of the last page.
These data were obtained by scaling publicly available ground-level
meteorological data to a typical turbine height and class 4 wind strength
(Equation 1 in the text) and then transforming into wind power output
using Equation 2.
The hourly data are available in electronic format upon request.
Percent Capacity (%)
10
20
30
40
50
60
70
80
90
100
Indirect Emissions Analysis • 48
Appendix B
-------
AZNM
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
CAMX
Jan Feb Mar
Apr May Jun Jul
ERCT
Aug Sep
Nov Dec
Jan Feb Mar
Apr May Jun Jul
FRCC
Aug Sep Oct Nov Dec
Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
MROE
5
10
15
20
Jan Feb Mar Apr May
Jun Jul Aug Sep
MROW
Oct Nov Dec
Jan Feb Mar Apr May
Jun Jul Aug Sep
NEWE
Nov Dec
18
Jan Feb Mar Apr May
Jun Jul Aug Sep
NWPP
Oct Nov Dec
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Indirect Emissions Analysis • 49
Appendix B
-------
NYCW
Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
NYUP
Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Feb Mar
Apr May Jun Jul Aug Sep
RFCM
Oct Nov Dec
Feb Mar
Apr May Jun Jul Aug Sep
RFCW
Oct Nov Dec
Jan Feb Mar Apr May
Jun Jul Aug Sep
RMPA
Nov Dec
Jan Feb Mar Apr May
Jun Jul Aug Sep
SPNO
Oct Nov Dec
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Indirect Emissions Analysis • 50
Appendix B
-------
SPSO
Jan Feb Mar
Apr May Jun Jul Aug Sep Oct Nov Dec
SRMV
Jan Feb Mar Apr May
Jun Jul Aug Sep
SRMW
Nov Dec
5
10
15
20
Jan Feb Mar Apr May
Jun Jul Aug Sep
SRSO
Oct Nov Dec
Feb Mar
Apr May Jun Jul Aug Sep Oct Nov Dec
SRTV
Feb Mar
Apr May Jun Jul Aug Sep Oct Nov Dec
SRVC
Jan Feb Mar
Apr May Jun Jul Aug Sep Oct Nov Dec
Percent Capacity (%)
10 20 30 40 50 60 70 80
90 100
Indirect Emissions Analysis • 51
Appendix B
-------
Appendix C:
Summary tables of raw and scaled wind speed and synthetic wind power
time series. Monthly summaries are shown for each eGRID subregion.
The hourly data are available in electronic format upon request
The proxy wind turbine locations are shown below, overlaid on map of all
U.S. WBAN stations.
• ...T . ^/^^"^Htef^a^
% .fit.VTJe: «3I*?**
. • •».%... I*."**.**!*;* • • V_
•{• ••!••• ••<••« •••• •!»• *•• *t*»H*
^5J®^^S§5
o -j^t^ •••• ••fOwt *
•«•!•!••••••• «N«.%*
• i» • ••
• ••••«./•
*•
i
•9
Index
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
NERC
Subregion
AZNM
CAMX
ERCT
FRCC
MROE
MROW
NEWE
NWPP
NYCW
NYLI
NYUP
RFCE
RFCM
RFCW
RMPA
SPNO
SPSO
SRMV
SRMW
SRSO
SRTV
SRVC
WBAN
Number
23055
23187
23042
12844
14898
24012
14739
24150
94789
94789
4725
93730
94860
14848
24018
23065
23047
12916
93822
13874
13893
93729
Location
GUADALUPE PASS, TX
SANDBERG, CA
LUBBOCK, TX
WEST PALM BEACH, FL
GREEN BAY, Wl
DICKINSON, ND
BOSTON, MA
LIVINGSTON, MT
NEW YORK, NY
NEW YORK, NY
BINGHAMTON, NY
ATLANTIC CITY, NJ
GRAND RAPIDS, Ml
SOUTH BEND, IN
CHEYENNE, WY
GOODLAND, KS
AMARILLO, TX
NEW ORLEANS, LA
SPRINGFIELD, IL
ATLANTA, GA
MEMPHIS, TN
CAPE HATTERAS, NC
Latitude
31.5
34.44
33.4
26.41
44.31
46.48
42.22
45.42
40.4
40.4
42.13
39.28
42.53
41.43
41.1
39.22
35.13
29.59
39.51
33.38
35.04
35.14
Longitude
-104.49
-118.43
-101.49
-80.06
-88.07
-102.48
-71.01
-110.27
-73.48
-73.48
-75.59
-74.28
-85.31
-86.2
-104.49
-101.41
-101.43
-90.15
-89.41
-84.26
-89.59
-75.37
Indirect Emissions Analysis'
Appendix B
52
-------
from
Monthly average interpolated wind speed (mph)
representative WBAN station in each NERC subregion
AZNM
CAMX
ERCT
FRCC
MROE
MROW
NEWE
NWPP
NYCW
NYLI
NYUP
RFCE
RFCM
RFCW
RMPA
SPNO
SPSO
SRMV
SRMW
SRSO
SRTV
SRVC
Jan
19.63
14.08
10.61
10.49
9.23
11.78
12.32
15.95
12.33
12.33
8.11
8.15
9.49
9.23
11.03
10.20
11.87
8.94
9.86
8.92
9.75
10.60
Feb
21.52
13.99
10.36
8.74
8.34
10.66
10.86
14.12
11.71
11.71
7.23
8.77
8.23
8.40
11.36
9.73
12.04
8.79
9.37
8.72
8.60
9.51
Mar
22.01
13.37
13.24
8.99
9.12
13.27
12.21
15.85
13.01
13.01
8.37
10.40
8.90
9.26
13.99
13.26
13.01
8.39
11.32
10.08
10.39
11.27
Apr
20.21
14.91
14.15
10.33
9.59
14.23
11.29
12.57
12.40
12.40
8.35
10.72
10.23
10.53
13.03
14.24
14.29
9.93
10.93
8.96
9.35
12.02
May
16.42
14.43
12.49
8.44
9.27
12.74
11.53
12.15
10.97
10.97
8.00
8.30
8.68
8.58
11.14
12.76
11.71
7.47
9.26
7.71
7.38
9.59
Jun
15.69
14.16
13.87
8.85
8.04
10.95
9.75
11.63
10.28
10.28
6.72
8.17
7.90
8.08
11.01
12.91
14.10
6.12
7.52
7.66
7.03
8.65
July
15.94
12.23
10.74
8.97
6.98
10.83
10.45
9.35
9.42
9.42
7.00
6.85
7.36
7.05
9.86
12.76
12.50
7.08
6.34
5.82
7.10
7.44
Aug
13.56
11.38
9.23
7.67
6.44
10.56
9.18
10.30
9.75
9.75
7.22
6.26
7.32
6.11
9.80
9.74
10.55
5.45
5.74
7.03
6.64
7.25
Sept
14.98
11.39
10.37
9.07
6.41
10.52
9.69
12.52
10.31
10.31
7.36
6.94
8.32
7.19
9.70
12.28
13.01
8.33
6.35
7.61
7.04
9.04
Oct
16.82
13.43
9.69
10.69
7.36
11.14
12.67
12.35
12.10
12.10
7.58
9.58
7.98
6.94
10.07
11.23
11.86
8.57
7.60
8.48
6.81
10.18
Nov
19.53
15.19
11.24
10.40
10.58
13.14
11.02
18.28
13.11
13.11
10.18
9.36
12.82
12.31
14.79
12.64
13.02
8.30
12.36
8.80
9.43
9.69
Dec
19.26
13.98
10.86
8.48
7.67
13.25
11.49
19.86
12.68
12.68
8.87
8.85
9.93
9.88
13.47
11.57
10.90
9.02
9.58
9.33
8.46
9.52
Indirect Emissions Analysis • 53
Appendix C
-------
AZNM
CAMX
ERCT
FRCC
MROE
MROW
NEWE
NWPP
NYCW
NYLI
NYUP
RFCE
RFCM
RFCW
RMPA
SPNO
SPSO
SRMV
SRMW
SRSO
SRTV
SRVC
Monthly average interpolated repr
scaled to Class 4 (16.25 mph average
Jan
38.64
28.04
21.49
13.75
8.36
16.50
14.79
18.86
18.14
23.51
14.24
13.85
18.69
16.96
20.79
13.88
21.61
12.18
20.22
12.60
13.77
15.10
Feb
42.36
27.86
20.98
11.45
7.55
14.93
13.03
16.70
17.23
22.32
12.68
14.91
16.23
15.44
21.40
13.23
21.92
11.97
19.22
12.31
12.14
13.55
Mar
43.33
26.63
26.82
11.79
8.26
18.60
14.66
18.74
19.14
24.81
14.69
17.68
17.54
17.02
26.36
18.03
23.68
11.43
23.23
14.24
14.68
16.06
Apr
39.78
29.68
28.65
13.55
8.68
19.94
13.55
14.87
18.24
23.64
14.65
18.22
20.16
19.35
24.56
19.36
26.02
13.53
22.42
12.65
13.21
17.13
May
32.32
28.74
25.29
11.06
8.40
17.85
13.83
14.37
16.13
20.91
14.04
14.10
17.11
15.76
20.99
17.36
21.31
10.18
19.00
10.89
10.42
13.67
Jun
30.89
28.19
28.09
11.61
7.28
15.34
11.70
13.75
15.12
19.60
11.80
13.88
15.58
14.85
20.76
17.55
25.66
8.34
15.42
10.83
9.93
12.33
esentative wind speed
for each NERC subregion
July
31.38
24.35
21.76
11.76
6.33
15.18
12.54
11.06
13.86
17.95
12.28
11.64
14.51
12.96
18.58
17.36
22.76
9.64
13.01
8.21
10.03
10.61
Aug
26.70
22.67
18.69
10.06
5.83
14.80
11.02
12.18
14.35
18.59
12.67
10.63
14.43
11.23
18.47
13.24
19.20
7.43
11.78
9.92
9.38
10.33
Sept
29.48
22.68
21.00
11.90
5.80
14.74
11.63
14.81
15.16
19.64
12.91
11.79
16.40
13.21
18.28
16.70
23.67
11.34
13.03
10.75
9.95
12.89
Oct
33.12
26.74
19.62
14.02
6.66
15.61
15.20
14.60
17.80
23.07
13.30
16.28
15.73
12.75
18.97
15.26
21.59
11.67
15.59
11.98
9.62
14.51
Nov
38.44
30.24
22.76
13.64
9.58
18.40
13.23
21.61
19.29
24.99
17.87
15.91
25.27
22.63
27.87
17.19
23.70
11.30
25.36
12.43
13.32
13.81
Dec
37.92
27.84
21.99
11.12
6.95
18.57
13.80
23.48
18.66
24.18
15.56
15.05
19.57
18.16
25.39
15.73
19.83
12.29
19.66
13.18
11.94
13.57
Indirect Emissions Analysis • 54
Appendix C
-------
Monthly average calculated wind power output from 1
in each NERC subregion (dropout at 56
.5 MW proxy turbine
mph)
AZNM
CAMX
ERCT
FRCC
MROE
MROW
NEWE
NWPP
NYCW
NYLI
NYUP
RFCE
RFCM
RFCW
RMPA
SPNO
SPSO
SRMV
SRMW
SRSO
SRTV
SRVC
Jan
883.49
971.55
680.73
286.16
58.62
439.92
338.89
572.73
546.35
822.24
335.36
356.54
638.43
506.60
644.99
290.58
752.59
202.22
677.87
253.66
258.66
353.12
Feb
919.73
941.26
670.23
166.17
23.10
343.82
213.95
479.22
504.41
767.69
324.15
394.50
477.22
413.12
659.34
217.16
766.10
177.89
624.12
227.19
200.36
226.16
Mar
910.01
865.85
929.82
190.93
51.41
573.12
272.13
574.52
589.69
848.34
368.73
540.16
605.68
549.15
895.00
515.08
815.82
160.16
797.53
310.51
335.47
407.72
Apr
823.61
1048.52
902.22
268.04
64.85
655.90
241.16
402.75
577.93
849.13
377.29
584.68
721.99
606.09
835.90
587.51
878.05
283.73
796.78
235.63
242.64
500.17
May
879.14
1101.11
899.96
137.00
47.93
518.31
286.37
358.02
406.64
677.99
294.03
341.90
549.50
457.05
664.56
515.45
735.24
114.13
637.27
127.92
97.73
228.59
Jun
1056.20
1005.35
1121.26
198.17
28.74
386.22
119.42
323.25
330.33
606.39
158.21
271.12
423.01
350.10
643.43
497.59
1004.26
67.34
390.96
139.60
87.17
172.78
July
1102.16
924.58
769.51
193.28
14.36
329.69
156.53
185.49
276.92
531.39
171.78
169.22
363.23
281.30
566.71
490.45
859.04
106.24
274.41
91.58
94.79
90.20
Aug
988.58
835.30
577.06
149.79
11.58
343.77
107.23
229.19
288.96
556.41
197.17
164.24
354.43
208.01
526.45
233.00
625.42
62.86
204.13
92.51
87.37
88.40
Sept
990.65
803.95
692.23
199.06
17.25
336.63
138.02
368.41
336.61
649.84
225.59
237.97
452.75
304.92
522.83
451.34
912.29
207.91
305.80
97.90
108.31
237.99
Oct
901.21
906.92
604.10
267.63
16.27
391.38
363.12
354.81
572.02
812.71
261.58
498.45
446.94
264.16
602.56
366.46
697.96
207.15
480.58
210.36
100.74
312.39
Nov
849.34
1037.77
675.85
256.74
113.47
559.79
251.55
786.22
633.20
851.24
491.64
501.62
890.43
780.63
813.64
414.62
848.84
209.46
898.76
218.47
289.18
281.43
Dec
987.33
971.48
672.41
164.98
21.29
532.01
254.57
880.60
574.40
858.86
368.78
415.26
673.93
548.80
838.43
362.94
601.09
216.39
648.20
209.75
168.19
262.02
Indirect Emissions Analysis • 55
Appendix C
-------
Appendix D:
Color intensity maps of hourly indirect C02, NOx and S02 emissions rates
for each eGRID subregion, using several alternative calculation
methodologies.
Each grid shows hour of the day on the vertical axis and days of the year
on the horizontal axis. Color scale is shown at the bottom of each page.
The indirect emissions metrics shown are:
Figure Title Page
D-1 Hourly Average CO2 Emissions Rate 56
D-2 Empirical Incremental CO2 Emissions Rate 59
n . Flexibility-Weighted Hourly Average CO2 Emissions R0
D'3 Rate b2
D-4 Load Following Incremental CO2 Emissions Rate 65
D-5 Hourly Average NOx Emissions Rate 68
D-6 Empirical Incremental NOx Emissions Rate 71
n _ Flexibility-Weighted Hourly Average NOx Emissions 7.
u~' Rate /4
D-8 Load Following Incremental NOx Emissions Rate 77
D-9 Hourly Average SO2 Emissions Rate 80
D-10 Empirical Incremental SO2 Emissions Rate 83
_. .. Flexibility-Weighted Hourly Average SO2 Emissions _„
D'11 Rate 86
D-12 Load Following Incremental SO2 Emissions Rate 89
The data are available in electronic format upon request.
Indirect Emissions Analysis • 56
Appendix D
-------
D-l: Hourly Average COi Emissions Rate(tons per MWh)
AZNM
5
10
15
20
•>,
CAMX
ERCT
5
10
15
20
'I HiMf «•>'*
-------
D-l (continued): Hourly Average COi Emissions Rate
NYCW
5
10
15
20
5
10
15
20
'sr-.-i -.in
NYLI
NYUP
RFCE
5
10
15
20
RFCM
RFCW
5
10
15
20
100
RMPA
5
10
15
20
I I I V I I
: :
r ' -
' ' I 1 1 1 1 1
SPNO
150
200
250
Emissions Rate (tons/MWh)
0.9
300
350
Indirect Emissions Analysis • 58
Appendix D
-------
D-l (continued): Hourly Average COi Emissions Rate (continued)
SPSO
5
10
15
20
SRMV
10
15
20
SRMW
5
10
15
20
SRSO
5
10
15
20
i i i i i i
-
-
i i i i i i
5
10
15
20
r
SRTV
SRVC
50
100
150
200
250
300
350
0.7
Emissions Rate (tons/MWh)
0.8
0.9
Indirect Emissions Analysis • 59
Appendix D
-------
D-2: Empirical Incremental COi Emissions Rate (tons per MWh)
AZNM
CAMX
5
10
15
20
ERCT
FRCC
MROE
MROW
NEWE
NWPP
100
150
200
250
300
350
Emissions Rate (tons/MWh)
Indirect Emissions Analysis • 60
Appendix D
-------
D-2 (continued): Empirical Incremental COi Emissions Rate
NYCW
NYLI
5
10
15
20
5
10
15
20
NYUP
RFCE
Pij£ll$^
4 jJrl»*s|iJjB&AJf>«'a"*\j!i{'>£ Ovf " ' !'?rirT-i =1 ?Cr • ••' -. 'r''~J- ""''<••!'* IJ^v^-j"iJ.l'S^}i 1,'-If•!• T^l
fe isfcSH^SS^s3rafi^'-^^ jf^^^^M^IMEi^
RFCM
50
RFCW
RMPA
SPNO
'
100
150
200
-. •• -
250
300
350
Emissions Rate (tons/MWh)
0.9
Indirect Emissions Analysis • 61
Appendix D
-------
D-2 (continued): Empirical Incremental COi Emissions Rate
AZNM
CAMX
ERCT
5
10
15
20
FRCC
MROE
MROW
aa&gi&Xi&ti^^^
- .*. Vt • •. I • . If1 .. J- J •. . *f
NEWE
NWPP
100
150
200
250
300
350
Emissions Rate (tons/MWh)
0.9
Indirect Emissions Analysis • 62
Appendix D
-------
D-2 (continued): Empirical Incremental COi Emissions Rate
SPSO
5
10
15
20
5
10
15
20
SRMV
SRMW
SRSO
5
10
15
20
5
10
15
20
SRTV
SRVC
50
100
150
200
250
Emissions Rate (tons/MWh)
0.9
300
350
1.1
1.2
Indirect Emissions Analysis • 63
Appendix D
-------
D-3: Flexibility-Weighted Hourly Average COi Emissions Rate
(tons per MWh)
5
10
15
20
III
AZNM
5
10
15
20
CAMX
5
10
15
20
ERCT
.
5
10
15
20
~- T
FRCC
MROE
5
10
15
20
MROW
5
10
15
20
NEWE
100
NWPP
150
200
250
300
350
Emissions Rate (tons/MWh)
0.9
Indirect Emissions Analysis • 64
Appendix D
-------
D-3 (continued): Flexibility-Weighted Hourly Average COi Emissions Rate
NYCW
5
10
15
20
f
ff *\ a $
5
10
15
20
NYLI
NYUP
RFCE
5
10
15
20
,
- - -
rr1 f "•*
I
5
10
15
20
RFCM
I hi < ^'iHMr
RFCW
5
10
15
20
i i i. _ ' i i
l -
-
ii [ [ i i [
5
10
15
20
RMPA
SPNO
100
150
200
250
300
350
Emissions Rate (tons/MWh)
0.9
Indirect Emissions Analysis • 65
Appendix D
-------
D-3 (continued): Flexibility Weighted Hourly Average COi Emissions Rate
SPSO
SRMV
SRMW
SRSO
10
15
20
5
10
15
20
SRTV
SRVC
5
10
15
20
',
50
100
150
200
250
300
350
Emissions Rate (tons/MWh)
0.9
Indirect Emissions Analysis • 66
Appendix D
-------
D-4: Load-Following Incremental COi Emissions Rate (tons per MWh)
AZNM
5
10
15
20
CAMX
ERCT
5
10
15
20
FRCC
MROE
MROW
5
10
15
20
NEWE
NWPP
50
100
150
200
250
300
350
Emissions Rate (tons/MWh)
0.9
Indirect Emissions Analysis • 67
Appendix D
-------
D-4 (continued): Load-Following Incremental COi Emissions Rate
NYCW
NYLI
NYUP
RFCE
5
10
15
20
RFCM
RFCW
5
10
15
20
RMPA
50
SPNO
100
150
200
250
300
350
Emissions Rate (tons/MWh)
0.9
Indirect Emissions Analysis • 68
Appendix D
-------
D-4 (continued): Load-Following Incremental COi Emissions Rate
SPSO
SRMV
SRMW
5
10
15
20
SRSO
>v-T.,:;iv. •', tfjl i-:.-.'•'• -t-?~ . -•'•"I;
M^mjMfit
5
10
15
20
4?£
SRTV
SRVC
5
10
15
20
iff«M5l
^> ?i-f&••••> --r^s. T.V-II*--
l •'"^•'^•'Iv.'--1.- - • -, '•?•
50
100
150
200
250
..
300
350
Emissions Rate (tons/MWh)
0.9
1.1
1.2
Indirect Emissions Analysis • 69
Appendix D
-------
D-5: Hourly Average NOx Emissions Rate (pounds per MWh)
AZNM
5
10
15
20
CAMX
I j
ERCT
5
10
15
20
FRCC
5
10
15
20
MROE
MROW
NEWE
10
15
20
\-
NWPP
5
10
15
20
50
100
150
200
250
300
350
Emissions Rate (Ibs/MWh)
0 0.5 1 1.5
2.5
3.5
Indirect Emissions Analysis • 70
Appendix D
-------
D-5 (continued): Hourly Average NOx Emissions Rate
NYCW
5
10
15
20
'II
NYLI
ioF
15h
20 \-
NYUP
5
10
15
20
_
5
10
15
20
5
10
15
20
5
10
15
20
5
10
15
20
RFCE
5
10
15
20
i i i i
:
I I I I I I
I
—
I
RFCM
RFCW
RMPA
50
100
SPNO
150
200
250
300
Emissions Rate (Ibs/MWh)
350
0 0.5 1 1.5
2.5
3.5
Indirect Emissions Analysis • 71
Appendix D
-------
D-5 (continued): Hourly Average NOx Emissions Rate
SPSO
5
10
15
20
SRMV
5
10
15
20
20
SRMW
fill
I 1 1 I
;
i i * i
SRSO
5
10
15
20
SRTV
SRVC
150
200
250
300
350
0.5
1.5
Emissions Rate (Ibs/MWh)
2.5
3.5
Indirect Emissions Analysis • 72
Appendix D
-------
D-6: Empirical Incremental NOx Emissions Rate (pounds per MWh)
AZNM
5
10
15
20
5
10
15
20
5
10
15
20
CAMX
ERCT
FRCC
MROE
MROW
NEWE
NWPP
100
150
200
250
300
350
Emissions Rate (Ibs/MWh)
0 0.5 1 1.5
2.5
3.5
Indirect Emissions Analysis • 73
Appendix D
-------
D-6 (continued): Empirical Incremental NOx Emissions Rate
NYCW
NYLI
NYUP
RFCE
^
5
10
15
20
5
10
15
20
RFCM
RFCW
RMPA
SPNO
50
100
150
200
250
300
350
Emissions Rate (Ibs/MWh)
0 0.5 1
1.5
2.5
3.5
Indirect Emissions Analysis • 74
Appendix D
-------
D-6 (continued): Empirical Incremental NOx Emissions Rate
SPSO
SRMV
SRMW
SRSO
SRTV
100
SRVC
150
200
250
300
350
Emissions Rate (Ibs/MWh)
0 0.5 1 1.5
2.5
3.5
4.5
Indirect Emissions Analysis • 75
Appendix D
-------
D-7: Flexibility-Weighted Hourly Average NOx Emissions Rate
(pounds per MWh)
5
10
15
20
5
10
15
20
AZNM
CAMX
ERCT
FRCC
MROE
5
10
15
20
MROW
5
10
15
20
NEWE
IT I
NWPP
5
10
15
20
''
*< <' «.iH
50
100
150
200
250
300
350
0.5 1 1.5
Emissions Rate (Ibs/MWh)
2.5
3.5
Indirect Emissions Analysis • 76
Appendix D
-------
D-7 (continued): Flexibility-Weighted Hourly Average NOx Emissions Rate
NYCW
5
10
15
20
5
10
15
20
5
10
15
20
NYLI
NYUP
n
RFCE
RFCM
5
10
15
20
RFCW
5
10
15
20
|.
RMPA
10
15
20
5
10
15
20
.ill
-.-j .. -^
SPNO
50
100
150
200
250
300
350
Emissions Rate (Ibs/MWh)
0 0.5 1 1.5
2.5
3.5
Indirect Emissions Analysis • 77
Appendix D
-------
D-7 (continued): Flexibility-Weighted Hourly Average NOx Emissions Rate
SPSO
5
10
15
20
E
SRMV
10
15
20
SRMW
SRSO
5
10
15
20
'Iff.
W;
SRTV
5
10
15
20
50
100
SRVC
150
200
250
300
Emissions Rate (Ibs/MWh)
0 0.5 1 1.5
2.5
3.5
4.5
Indirect Emissions Analysis • 78
Appendix D
-------
D-8: Load-Following Incremental NOx Emissions Rate (pounds per MWh)
AZNM
CAMX
ERCT
5
10
15
20
FRCC
5
10
15
20
MROE
MROW
NEWE
NWPP
50
100
150
200
250
300
350
Emissions Rate (Ibs/MWh)
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Indirect Emissions Analysis • 79
Appendix D
-------
D-8 (continued): Load-Following Incremental NOx Emissions Rate
NYCW
NYLI
NYUP
RFCE
RFCM
RFCW
RMPA
SPNO
100
150
200
250
300
350
Emissions Rate (Ibs/MWh)
0 0.5 1 1.5
2.5
3.5
Indirect Emissions Analysis • 80
Appendix D
-------
D-8 (continued): Load-Following Incremental NOx Emissions Rate
SPSO
SRMV
SRMW
SRSO
5
10
15
20
SRTV
SRVC
100
150
200
250
300
Emissions Rate (Ibs/MWh)
0 0.5 1
2.5
3.5
4.5
350
Indirect Emissions Analysis • 81
Appendix D
-------
D-9: Hourly Average SOi Emissions Rate (pounds per MWh)
AZNM
CAMX
5
10
15
20
ERCT
10
15
20
r
FRCC
MROE
5
10
15
20
I I I I I H <
-
-
I I I I I Mi .il
I
—
_
[
MROW
5
10
15
20
5
10
15
20
NEWE
100
NWPP
150
200
250
300
350
Emissions Rate (Ibs/MWh)
Indirect Emissions Analysis • 82
Appendix D
-------
JgW*"
I O
20
D-9 (continued): Hourly Average SOi Emissions Rate
NYCW
NYLI
NYUP
5
10
15
20
_,
_
5
10
15
20
5
10
15
20
RFCE
II • Mlljf Hi ii
RFCM
RFCW
!
RMPA
SPNO
5
10
15
20
50
100
150
200
250
300
Emissions Rate (Ibs/MWh)
350
Indirect Emissions Analysis • 83
Appendix D
-------
D-9 (continued): Hourly Average SOi Emissions Rate
SPSO
5
10
15
20
5
10
15
20
5
10
15
20
[ .I
SRMV
SRMW
5
10
15
20
i i i i i i
-
-
i i i i i i
SRSO
SRTV
SRVC
^^^^^
50
100
150
200
250
Emissions Rate (Ibs/MWh)
10
300
350
Indirect Emissions Analysis • 84
Appendix D
-------
D-10: Empirical Incremental SOi Emissions Rate (pounds per MWh)
AZNM
CAMX
15
20
ERCT
5
10
15
20
FRCC
MROE
MROW
NEWE
NWPP
100
150
200
250
300
350
Emissions Rate (Ibs/MWh)
Indirect Emissions Analysis • 85
Appendix D
-------
D-10 (continued): Empirical Incremental SOi Emissions Rate
NYCW
5
10
15
20
5
10
15
20
10
15
20
5
10
15
20
100
NYLI
NYUP
RFCE
RFCM
RFCW
RMPA
SPNO
150
200
250
Emissions Rate (Ibs/MWh)
300
350
Indirect Emissions Analysis • 86
Appendix D
-------
D-10 (continued): Empirical Incremental SOi Emissions Rate
SPSO
5
10
15
20
5
10
15
20
SRMV
100
SRMW
SRSO
SRTV
SRVC
150
200
250
300
350
Emissions Rate (Ibs/MWh)
10
Indirect Emissions Analysis • 87
Appendix D
-------
D-ll: Flexibility-Weighted Hourly Average SOi Emissions Rate
(pounds per MWh)
10
15
20
AZNM
CAMX
ERCT
FRCC
MROE
MROW
5
10
15
20
5
10
15
20
NEWE
NWPP
100
150
200
250
300
350
Emissions Rate (Ibs/MWh)
10
14
Indirect Emissions Analysis • 88
Appendix D
-------
D-ll (continued): Flexibility-Weighted Hourly Average SOi Emissions Rate
NYCW
5
10
15
20
NYLI
\
i
WWP
.
NYUP
RFCE
RFCM
5
10
15
20
5
10
15
20
RFCW
RMPA
SPNO
5
10
15
20
1 .
50
100
150
200
250
300
350
Emissions Rate (Ibs/MWh)
10
14
Indirect Emissions Analysis • 89
Appendix D
-------
D-ll (continued): Flexibility-Weighted Hourly Average SOi Emissions Rate
SPSO
SRMV
SRMW
If «<*i JIM
5
10
15
20
5
10
15
20
! 1
SRSO
50
100
SRTV
SRVC
150
200
250
300
350
Emissions Rate (Ibs/MWh)
10
12
14
Indirect Emissions Analysis • 90
Appendix D
-------
D-12: Load-Following Incremental SOi Emissions Rate (pounds per MWh)
AZNM
5
10
15
20
CAMX
5
10
15
20
ERCT
FRCC
5
10
15
20
MROW
NEWE
NWPP
100
150
200
250
300
350
Emissions Rate (Ibs/MWh)
Indirect Emissions Analysis • 91
Appendix D
-------
D-12 (continued): Load-Following Incremental SOi Emissions Rate
NYCW
NYLI
5
10
15
20
NYUP
RFCE
RFCM
5
10
15
20
RFCW
RMPA
SPNO
100
150
200
250
300
350
Emissions Rate (Ibs/MWh)
Indirect Emissions Analysis • 92
Appendix D
-------
D-12 (continued): Load-Following Incremental SOi Emissions Rate
SPSO
SRMV
SRMW
SRSO
5
10
15
20
SRTV
SRVC
100
150
200
250
300
350
Emissions Rate (Ibs/MWh)
10
Indirect Emissions Analysis • 93
Appendix D
------- |