NCEE Working Paper

Construction and Application of the
Micro-level Engineering,
Environmental, and Economic
Detail of Electricity (MEEDE)
Dataset, Version 2

Candise Henry, Jared Woollacott, Alison Bean de
Hernandez, Andrew Schreiber, and David A.
Evans

Working Paper 23-03
March, 2023

U.S. Environmental Protection Agency	|L|f*pp gf

National Center for Environmental Economics	livtt flr

https://www.epa.gov/environmental-economics	env'ronmental^conomics


-------
Construction and Application of the Micro-level Engineering,
Environmental, and Economic Detail of Electricity (MEEDE)

Dataset, Version 2

Candise Henry*, Jared Woollacott, Alison Bean de Hernandez,

Andrew Schreiber, David A. Evans

March 17, 2023

Abstract

ABSTRACT: The Micro-level Engineering, Economic, and Environmental Detail of Elec-
tricity (MEEDE, Version 2) dataset provides a unit-level representation of the United States
electricity sector based on public sources. The data draw on a disparate set of engineering,
environmental, and economic data, primarily from the Environmental Protection Agency and
the Department of Energy's Energy Information Administration, to characterize all utility-scale
electric generating units in the U.S. in terms of their physical inputs of energy, outputs of elec-
tricity and pollution, generating and pollution control equipment configurations, and economic
costs (capital, labor, energy, and materials) associated with their operation. The combination
of complete unit-level physical details of the US grid with economic characteristics is a key
distinction between the MEEDE data and other publicly-available sources. The MEEDE data
provide a highly-valuable tool for generating descriptive statistics and supporting advanced
partial or general equilibrium modeling efforts that require technology-rich representations of
the U.S. electricity grid. We demonstrate how these data can be integrated into social account-
ing matrices for use in economy-wide modeling applications.

* Corresponding Author: clhenry@rti.org

KEYWORDS: electricity pollution control, social accounting matrices, matrix balancing

JEL CODES: C69, Q43, Q52, Q53

Disclaimer

The views expressed in this paper are those of the author(s) and do not necessarily represent
those of the U.S. Environmental Protection Agency (EPA). In addition, although the research
described in this paper may have been funded entirely or in part by the U.S EPA, it has not been
subjected to the Agency's required peer and policy review. No official Agency endorsement
should be inferred. Further, the contractor's role did not include establishing Agency policy.

1


-------
Acknowledgements

The authors thank Ann Wolverton, Alex Marten, Wade Davis, Jim McFarland, Alex
MacPherson, and David Bielen for their support in developing these data and to Avery Tilley
for her dedicated research and data analysis.

2


-------
Contents

1	Introduction	5

2	Methods	7

2.1	Engineering		7

2.1.1	Activity Data: Energy Information Administration Forms 923 and 860 ....	7

2.1.2	Engineering Data Description		10

2.2	Economics		12

2.2.1	Generation Costs: Annual Technology Baseline 2020 		12

2.2.2	Pollution Control Equipment Costs: Integrated Planning Model Version 6 . .	16

2.2.3	Fuel Costs: Energy Information Administration Form 923 		17

2.2.4	Wholesale Prices: Federal Energy Regulatory Commission Form 714		18

2.2.5	Economic Data Description		20

2.3	Environment 		23

2.3.1	Particulate Matter Emissions: 2018 EPA Mercury and Air Toxics Standards
Residual Risk and Technology Review		23

2.3.2	Mercury Emissions: EPA Mercury and Air Toxics Standards 		25

2.3.3	SOx and NOx Emissions: EPA Air Markets Program 		25

2.3.4	Greenhouse Gas Emissions: EPA Greenhouse Gas Reporting Program ....	26

2.3.5	Environmental Data Description		26

2.4	Full MEEDE Dataset 		27

2.4.1	Annual Trends		30

2.4.2	Updates Since MEEDE Version 1		32

2.4.3	Comparison with Other Datasets		33

3	Application: SAM Integration	34

3.1	SAM Identities		34

3.2	Prior Formation 		36

3.2.1	Macro Priors		37

3.2.2	Micro Priors, Electricity		37

3.3	Balancing Routine		38

4	Conclusion	40
A Appendix I: MEEDE Data Dictionary	44
B MEEDE Version 2 Outputs for 2016	48

4


-------
1 Introduction

This methods report explains the construction and application of the second version of a detailed
dataset covering electricity generation in the United States based entirely on public sources: the
Micro-level Engineering, Economic, and Environmental Detail of Electricity (MEEDE). These data
provide a full economic and environmental profile of the physical equipment operating on the U.S.
electric grid in annual time series from 2013 to 2019.1 The detailed grid profile offered by the
MEEDE data is ideal for integration into partial and general equilibrium economic models. Toward
that end, this model documentation also provides a social accounting matrix sector disaggregation
and matrix balancing routine along with recommendations for general equilibrium model specifica-
tions for the electricity sector.

MEEDE is made up of power plant- and boiler- level information from various publicly-available
datasets provided by the Energy Information Administration (EIA), the Environmental Protection
Agency (EPA), Federal Energy Regulatory Commission (FERC), and the National Renewable En-
ergy Laboratory (NREL). This includes: (1) engineering data such as boiler-to-pollution control
equipment configurations, boiler-level fuel consumption quantities, and generator electricity output;
(2) economic data on power plant capital and operating costs, and emissions control equipment-
related costs; and (3) environmental data on power plant emissions rates of various pollutants and
greenhouse gases.

This version of MEEDE is the successor to the MEEDE dataset released in 2016 for the data year
2013 (MEEDE Version 1). Unlike the first version, which only covered the electricity sector land-
scape of 2013, Version 2 includes seven years of data, from 2013 to 2019. Several other improvements
(detailed below) have been made in this update as well, including the use of continuously-monitored
emissions data collected by the Environmental Protection Agency and the application of electric-
ity sales information from the Federal Energy Regulatory Commission. As such, the outputs in
MEEDE Version 2 do not correspond directly with those from MEEDE Version 1 including for the
year 2013. This updated version provides a more complete snapshot of the U.S. electricity sector.

The combined physical and economic detail compiled in MEEDE is ideal for use in partial and
general equilibrium economic models. Unit level information in MEEDE observations can be ag-
gregated readily on a variety of plant characteristics such as fuel type, primve mover, pollution
controls, or location to generate prototypical plants for modeling purposes. Such "bottom-up"
aggregations pose unique challenges when used with economy-wide models, whose base data do
not comport with bottom-up information. Combining multiple sources of economic data for use
in general equilibrium models therefore requires a method to re-balance the aggregated sources for
micro-consistency. We develop a numerical algorithm to re-balance our social accounting matrices
with integrated bottom-up data.

This report is organized as follows. First, the Methods section presents the construction of the
MEEDE dataset in three subsections by the type of data: engineering, economic, or environmental.
Within each of these sections, we introduce the datasets by source and describe how the data is
assembled as well as any assumptions we make regarding the data. At the end of each of the three
sections, we provide a summary to describe the assembled data. In a fourth subsection, we discuss

1At the time of MEEDE Version 2's release, the Energy Information Administration's Forms 923 and 860 only
published data up to 2019.

5


-------
the full MEEDE dataset by presenting trends through time (from 2013 to 2019). We also discuss
how this MEEDE update differs from the 2013 version.

Second, the Applications section describes the integration of the bottom-up information provided
by the MEEDE dataset into a balanced social accounting matrix. We discuss the social account-
ing matrix identities, priors, disaggregation, and balancing routine. We conclude this section by
presenting recommendations for general equilibrium model specifications for the electricity sector.

6


-------
2 Methods

Data provided in national accounts present an aggregated electricity generation, transmission, and
distribution sector. Capturing the heterogeneity of production and abatement alternatives requires
a finer-grain representation, disaggregated along several dimensions to the level of generation abate-
ment technology types. To achieve this, we integrate the EIA Forms 923 and 860 data (U.S. Energy
Information Administration, 2021a,b), the NREL Annual Technology Baseline (ATB) generation
cost estimates (National Renewable Energy Laboratory, 2021), the EPA Integrated Planning Model
(IPM) abatement cost estimates (U.S. Environmental Protection Agency, 2018), and EPA emissions
data to provide a comprehensive dataset covering 97 percent of electricity generation capacity in
the U.S. (U.S. Energy Information Administration, 2021c).2 This covers almost 100 percent of all
U.S. power plant emissions and abatement activity, as the 3 percent comprising smaller-scale (i.e.,
not utility-scale) systems are primarily solar.

2.1 Engineering

2.1.1 Activity Data: Energy Information Administration Forms 923 and 860
Data Assembly

The MEEDE data rely most heavily on EIA Forms 923 and 860 for the years 2013 to 2019 (U.S.
Energy Information Administration, 2021a,b). These forms collect information from electricity gen-
erating units (EGUs) primarily under the North American Industry Classification System (NAICS)
code 22 for Utilities (approximately 83 percent of all units), but also include data from other other
industries (U.S. Census Bureau, 2021; U.S. Energy Information Administration, 202Id).3 Form 923
data record activity levels of EGUs, including fuel use and net generation quantities. The data also
contain basic characteristics of each generating unit such as fuel type, prime mover, and location.
The Form 923 data provide 38 fuel types, which we collapse to 15 fuel codes. We do not collapse
the 18 unique prime movers.

Data from Form 923 page 1 summarize the total quantity of fuel input and electricity generation
output by fuel code, prime mover, and plant. From these data we were able to summarize generation
by fuel type, prime mover, and region. Table 1 presents the net electricity generation by each prime
mover across the North American Electric Reliability Corporation (NERC) regions.

Form 923 page 3 provides boiler-level detail on the type and total quantity of fuel inputs. Pollution
control equipment can be associated with the boiler(s) it serves using the Form 860 data. Identifying
the emissions profile of generation then requires a reliable mapping between boilers and prime
movers. This mapping is a many-to-many exercise: one boiler may serve multiple prime movers,
or multiple boilers may serve one prime mover. Fortunately, page 3 of the Form 923 data breaks

2The 97 percent of generation capacity includes only grid connected, utility-scale electricity generators, which are
defined as those located at power plants with at least 1 MW of total electricity generating capacity. This dataset
does not cover smaller-scale systems such as distributed generators and rooftop solar panels, which make up the
remaining approximately 3 percent of total capacity.

3Other industries include: Agriculture, Forestry, Fishing and Hunting (NAICS code 11; less than 1 percent of the
97 percent generator coverage); Mining (code 21; less than 1 percent); Manufacturing (codes 31-33; approximately
9 percent); Transportation and Warehousing (codes 48-49; less than 1 percent); and Administrative and Support
and Waste Management and Remediation Services (code 56; less than 1 percent). The remaining 5 percent of units
include those in the healthcare, real estate, and finance industries.

7


-------
Table 1: Annual net electricity generation from EI A Form 923 by prime mover and NERC region, 2019 (in
thousands of MWh)

(Thousands

Code

Description

SERC

RFC

WECC

MRO

TRE

NPCC

HICC

ASCC

Total USA

ST

Steam Turbine

709,119

570,254

238,591

216,269

131,440

93,375

5,476

700

1,965,224

CT

Combined Cycle (Combustion)

357,439

164,107

127,261

43,135

118,727

46,809

1,879

1,554

860,912

CA

Combined Cycle (Steam)

172,856

83,753

69,181

25,003

55,176

22,802

583

457

429,811

HY

Hydraulic Turbine

44,683

10,532

167,310

24,196

1,072

38,320

95

1,623

287,830

WT

Wind (Onshore)

5,676

30,267

54,377

119,566

76,063

8,072

529

143

294,693

GT

Gas Turbine

46,607

26,521

24,515

14,301

17,264

5,954

536

404

136,102

PV

Photovoltaic

16,302

2,655

41,133

1,861

4,330

2,062

268

0

68,611

CS

Combined Cycle (Single)

2,036

28,056

6,311

603

8,656

13,145

0

0

58,808

BT

Binary Cycle Turbines

0

0

4,340

0

0

0

0

0

4,340

IC

Internal Combustion (Diesel)

2,448

4,018

3,539

1,638

1,521

1,265

384

1,191

16,005

FC

Fuel Cell

59

294

911

0

0

381

0

0

1,645

CP

Concentrated Solar (Storage)

0

0

842

0

0

0

0

0

842

PS

Pumped Storage

-2,776

-1,680

-211

156

0

-750

0

0

-5,261

WS

Wind (Offshore)

0

0

0

0

0

118

0

0

118

BA

Battery

0

-38

-30

-2

-11

-8

-1

-4

-94

FW

Energy Storage (Flywheel)

0

-14

0

0

0

-11

0

0

-26

CE

Compressed Air (Storage)

-7

0

0

0

0

0

0

0

-7

OT

Other

110

235

18

325

0

59

0

0

747

Total



1,354,551

918,960

738,088

447,052

414,237

231,593

9,750

6,068

4,120,300

Regions: ASCC = Alaska Systems Coordinating Council; HICC = Hawaiian Islands Coordinating Council; MRO
= Midwest Reliability Organization; NPCC = Northeast Power Coordinating Council; RFC §= ReliabilityFirst Cor-
poration; SERC = Southeastern Electric Reliability Council; TRE = Texas Reliability Entity; WECC = Western
Electricity Coordinating Council.

Note: The values shown here may not sum to the totals shown due to independent rounding.

out boiler inputs (BID) by plant ID I'll), the prime movers (PM) they serve, and fuel type (FT).
Table 2 shows a meta-summary of the data available in Form 923. It includes the unit of observation
and variables provided by Form 923 pages 1 and 3.

We Can use the boiler-level fuel data from page 3 to allocate the generation data from page 1, giving
fuel use and estimated generation output data at the PI I) Pii FT-BID level. The second to last
column of Table 2 reveals that roughly 16 percent of power plants in Form 923 are served by boilers
(i.e., 1469 plants relative to 9809), but nearly half of all generation is provided through the use of
boilers (see steam turbines in Table: 1).

Pollution control processes and equipment are given control IDs that ate associated with plant
boilers, meaning environmental control equipment is associated only with steam generation in the
Form 860 data (Table 2). Through this association we can assign environmental control processes

8


-------
Table §t Heta-samixiary of EIA Form 923 data, 2019

Form

Schedule



Page

Observation

Variables

Number
of Plants

Number of
Observations

923

2_3_4_5_

M_

12

Page 1: Generation
and Fuel Data

{PID, PM, FT)

< Fuel, Generation >

9,809

14,345

923

2_3_4_5_

M.

12

Page 3: Boiler Fuel
Data

{PID, PM, FT, BID)

< Fuel, Sulfur Content, Ash Content >

1,469

9,551

923

2_3_4_5_

M_

12

Page 5: Fuel
Receipts and Costs

{PID}

< Monthly Fuel Prices >

1,006

38,107

860

3.1





Operable

{PID,GID, PM, FT}

< Nameplate Capacity, Operating Year,
Planned Retirement Year >

9,804

22,731

860

2





Plant

{PID}

< Plant Attributes >

11,833

11,833

860

6.1





Boiler-Generator

{PID,GID, BID}

< PID, GID, BID >

1,660

7,402

860

6.2





Boiler Info 8<
Design Parameters

{PID, BID}

< PID, BID, In-service Year, Firing Type,
Wet/Dry Bottom Tech >

1,661

4,680

860

6.1





Boiler Particulate
Matter

{ PID, BID, FGPJD }

< PID, ID, FGPJD >

835

2,423

860

6.1





Boiler S02

{PID, BID, FGDJD}

< PID, BID, FGDJD >

438

973

860

6.1





Boiler NOX

{ PID, BID, NOXJD }

< PID, BID, NOXJD >

790

1,742

860

6.1





Boiler Mercury

{PID, BID, HGJD }



491

1,455

860

6.1





Emissions Control
Equipment

{ PID, FGPJD, FGDJD, NOXJD, HGJD,
Equipment Type, Controls }

< Control ID (PM, SOX, NOX, HG),
Equipment Type >

1,184

5,787

860

6.2





Emission
Standards &
Strategies

{ PID, BID}

< PID, BID, Control Strategies: SOX,
NOX, HG >

1,661

4,680

860

6.1





Boiler Stack Flue

{PID, BID, FID}

< PID, BID, FID >

901

3,009

860

6.2





Stack Flue

{PID, FID}

< PID, FID, Service Year, Height,
Volume, Status >

904

2,459

Abbreviations: BID = boiler ID; PQD_ID = flue gas desulfurlzation equipment 11 >: K< H' ED = flue gas particulate
collector ID; FT s= fuel type; SID = generator ID; HG_ID = mercury equipment ID; NOXJD = nitrogen oxide
equipment ID; I'll") plant ID; PM = prime mover; SOX-ID = sulfur oxide equipment ID.

and equipment to the boilers represented in Form 923 page 3 boiler data. Then, with the boiler
technology characterized, we call identify what fuel input and electricity outputs are traveling
through which generation equipment and pollution control technology configurations.

We construct a plant boiler level dataset that identifies the type and age of environmental equip-
ment for all available boilers on the grid using the environmental association and equipment datasets
from Form 860 Schedule 6.1. The Emissions Control Equipment Page of Form 860 provides the
plant ID, associated control IDs, and the control equipment typo (e.g., dry sorbent injection for
S02 control)^ while the other pages in Schedule 6.1 provide the BID to control ID mapping, Envi-
ronmental equipment controls include those for sulfur, nitrogen, mercury, and particulates. Given
the importance of flue height for local health effects from pollution, we also generate the emissions-
weighted average flue height of each boiler (some boilers are served by multiple flues) . The Form
860 environmental association file provides, the type of environmental equipment installed at each

9


-------
plant and the boilers the equipment serves. The environmental equipment file provides additional
attributes for boilers (including boiler age) and some of the control equipment.

The Form 860 data also provide attributes at the generator and plant levels. Age and nameplate
capacity by generator are key variables in estimating the cost of generation. Because the final
dataset will be summarized at the PID-PM-FT-BID level, we sum nameplate capacity and took
a nameplate-weighted average age of the generators at the PID-PM level. The plant file provides
a variety of geographic and regulatory location characteristics that are merged with the genera-
tor and boiler data at the plant level. The final engineering dataset summarizes the generation
and abatement technologies operating on the grid with plant-, generator-, and boiler-level detail.
When applicable, we break generation out to the boiler level and identified associated environ-
mental equipment. The unit of observation is the PID-PM-FT-BID tuple wherever the boiler is
applicable and PID-PM-FT otherwise. Across all seven years (2013-2019), there are 91,296 distinct
observations in total and 133 variables. In 2019 alone, there are 14,347 distinct observations. The
variables provide detailed information on the geographic and regulatory location of the plants; the
ages of combustion, generation, and abatement equipment; the physical and thermal quantities of
fuel consumed; the flues and environmental equipment involved in the combustion of those quan-
tities; the boiler cooling equipment associated with each boiler; and the water use of the cooling
equipment.

Data Assumptions

The EIA Forms 923 and 860 datasets are our most comprehensive in terms of plant coverage
and plant-level technological details. This is the dataset on which we map the economics and
environment-related data. However, it reports only utility-scale electricity generators, which are
defined as those located at power plants with at least 1 MW of total electricity generating capacity
(U.S. Energy Information Administration, 2021c). Therefore, this dataset does not cover smaller-
scale systems such as rooftop solar panels.

2.1.2 Engineering Data Description

Table 1 shows that steam turbines generated the majority of electricity in 2019 (1,965 million MWh
of the U.S. total of 4,120 million MWh generation), followed by combined-cycle combustion prime
movers (CT; 861 million MWh) and then combined-cycle steam prime movers (CA; 430 million
MWh). In terms of regional variability in electricity generation, Table 1 shows that the Southeastern
Electric Reliability Council (SERC) region has the largest amount of generation nationally. It also
dominates in generation by steam turbine, combined cycle, gas turbine prime movers. The NERC
region with the second largest annual net generation is the ReliabilityFirst Corporation (RFC)
region, which covers the Mid-Atlantic and parts of the Midwest. The RFC region generates about 30
percent less electricity than SERC. The Western Electricity Coordinating Council (WECC) region
follows next in terms of total annual net generation, but leads in generation by hydropower (roughly
58 percent of all hydropower generation) and solar (almost 60 percent of all solar generation). The
Midwest Reliability Organization (MRO) region, which are the West North Central states, leads in
terms of onshore wind generation, covering 40 percent of all wind generation. Overall, utility-scale
solar across all regions contributes only a fraction of a percentage to total generation.

Meanwhile, in terms of the type of fuel, the Form 923 data show that most electricity was generated
by natural gas in 2019 (1,582 million MWh; Table 3) followed by coal (965 million MWh). This

10


-------
differs from the 20f 3 trend, when electricity was generated primarily by coal. Interestingly total
annual net generation has changed very little over the seven years 2013 to 2019.

Note that the fuel groupings shown in Table 3 are more aggregated than the fuel types presented
in the EIA Forms (and thereby in the full MEEDE dataset). For instance, coal in the EIA Forms
is categorized into bituminous, sub-bituminous, anthracite, and more. While we do not show these
groupings in Table 3, fuel type-prime mover pairs are an important part of the data because they
allow us to attribute the economic costs of generation in the next section.

Table 3: Annual net electricity generation from EIA Form 923 by fuel, 2013-2019 (in thousands of MWh)

Annual Net Generation
(Thousands of MWh)

Type

2013

2014

2015

2016

2017

2018

2019

Coal

1,586,100

1,585,806

1,351,017

1,243,511

1,207,435

1,151,497

965,196

Natural Gas

1,121,215

1,122,676

1,320,041

1,376,748

1,294,531

1,465,410

1,582,594

Nuclear

789,016

797,166

797,178

805,694

804,950

807,084

809,409

Hydropower

263,851

253,168

243,940

261,126

293,831

286,609

282,569

Wind

167,697

181,204

190,641

226,926

254,211

272,601

294,810

Petroleum

24,228

26,338

24,075

20,959

18,823

21,584

15,301

Geothermal

15,775

15,877

15,918

15,826

15,855

15,967

15,473

Solar

8,951

17,556

24,745

35,945

53,135

63,571

71,829

Other

86,959

88,456

89,465

88,798

87,875

87,864

83,118

Total

4,063,792

4,088,246

4,057,021

4,075,533

4,030,646

4,172,188

4,120,300

As previously discussed, Form 923 also provides boiler-level details on the type and quantity of
fuel inputs. Table 4 presents the heat (Btu) quantities of fuel consumption by various types of
fuel. Natural gas accounts for 53 percent of fuel combustion in electricity generation relative to
45 percent for coal in terms of Btu (based on totals in the Btu column in Table 4). Table 4 also
shows the pollutant content (sulfur and ash) for the fuel types. Coal is the primary sulfur- and
ash-bearing fuel, followed by oil and other petroleum derivatives. Of the two dominant types of
coal burned for electricity, bituminous has a higher sulfur and ash content than sub-bituminous.
Meanwhile, natural gas has negligible ash and sulfur content.

Lastly, the percentage of net fossil fuel generation that is controlled by each environmental technol-
ogy is shown in Table 5. Technologies are grouped by the primary pollutant they are designed to
target; however, environmental control technologies influence emissions of multiple pollutants. This
is particularly true with respect to mercury. The EIA Form 860 data identify 33 specific pollution
control technologies (and one "other" category). The majority of these are end-of-pipe controls,
but, for nitrogen in particular, there are a number of change-in-process controls. These controls
rely primarily on changes to how boilers are operated and may require less physical equipment rela-
tive to an end-of-pipe technology like a sulfur scrubber or particulate baghouse. Change-in-process
technologies are labeled with an asterisk in Table 5. Boilers use control technologies in over 1,000
different configurations ranging from no controls to nine control technologies. The prevalence of

11


-------
Table 4: U.S. electric grid fuel heat, sulfur, and ash content, 2019





Fuel Code

PollutantContent (%)



Volume/Mass



Fuel Code

Description

Sulfur

Ash

BTU (Quadrillion)

Quantity

Units

Coal

BIT

Bituminous coal

2.41

10.29

3.12

142.62

MM Sh. Ton

Coal

SUB

Subbituminous coal

0.35

6.04

3.77

222.58

MM Sh. Ton







1.13

10.43



183.08

MM Sh.Ton

Coal

COL

Other coal





3.24











0.00

0.00



102.87

Bn. Cu. Ft.

Gas

GAS

Natural gas & propane

0.00

0.00

11.98

11,205.58

Bn. Cu. Ft.

Gas

OGS

Other gases

0.00

0.00

0.24

1,069.70

Bn. Cu. Ft.







0.55

0.00



17.93

MM Barrels

Oil

OIL

Oil & petroleum deriv.

0.50

0.00

0.20

0.01

Bn. Cu. Ft.







5.08

0.80



3.32

MM Sh.Ton

Other

BML

Biomass liquids

0.00

0.00

0.00

6.24

MM Barrels

Other

BMS

Biomass solids

0.00

0.00

0.35

116.77

MM Sh.Ton

Other

MSW

Municipal solid wastes

0.00

0.00

0.26

28.89

MM Sh.Ton

Other OTH	Miscellaneous other	0.00	0.00	0.02	8.46 MMSh. Ton

Note: " Other coal" includes both solid and gaseous coal forms, including Coal-Derived Synthesis Gas.

control technologies varies significantly by region. For example, the Hawaiian Islands Coordinating
Council (HICC), Northeast Power Coordinating Council (NPCC), and Alaska Systems Coordi-
nating Council (ASCC) regions have relatively little fossil fuel generation equipped with sulfur,
particulate, or nitrogen control technologies, whereas RFC and SERC have relatively high fractions
of fossil fuel generation equipped with these controls.

2.2 Economics

Given a full technological and environmental characterization of the grid (see the engineering section
above), we specify the economic costs associated with electricity generation and pollution control.
We categorize cost components for generation and abatement technologies as either overnight capital
or fixed or variable operations and maintenance (O&M) costs and separate them by equipment
(i.e. generation and all associated abatement equipment). In all, the economic characterization
can provide an estimate of the capital, labor, energy, and materials requirements for operating
the electric grid. These costs are produced at the level of individual installations of generation
and pollution abatement equipment in true bottom-up fashion. They can be readily summarized
at higher-level regional or technology aggregations for modeling applications for comparison with
other cost or wholesale revenue estimates.

2.2.1 Generation Costs: Annual Technology Baseline 2020
Data Assembly

We use capital (CAPEX) and O&M expenditure estimates from the 2020 edition of the National Re-

12


-------
Table 5: Percent of net MWh fossil fuel generation with environmental controls per region, 2019 (%)

1 Control Equipment Type

Control ID

ASCC

HICC

MRO

NPCC

RFC

SERC

TRE

WECC

Total USA I

Sulfur Control Technologies



10.1

15.9

51.2

1.0

42.3

25.7

22.6

38.2

32.6

Circulating dry scrubber

CD

0.0

0.0

6.0

0.1

1.1

0.3

0.0

0.8

1.1

Dry sorbent (powder) injection type (DSI)

DSI

10.1

15.8

3.8

0.0

5.9

6.4

1.5

1.0

4.4

Jet bubbling reactor (wet) scrubber

JB

0.0

0.0

0.0

0.0

3.9

2.4

0.0

0.0

1.8

Mechanically aided type (wet) scrubber

MA

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.1

0.0

Packed type (wet) scrubber

PA

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

Spray dryer type / dry FGD / semi-dry FGD

SD

5.5

0.1

21.9

0.0

2.0

2.6

1.7

10.3

5.4

Spray type (wet) scrubber

SP

0.0

0.0

19.5

0.9

28.7

17.7

18.9

16.8

19.7

Tray type (wet) scrubber

TR

0.0

0.0

0.0

0.0

9.3

1.8

1.8

5.8

3.9

Venturi type (wet) scrubber

VE

0.0

0.0

2.5

0.1

0.0

0.1

0.0

3.4

0.8

Particulate Control Technologies



19.6

15.9

60.2

4.1

45.9

33.8

26.9

33.2

37.2

Baghouse (fabric filter), pulse

BP

15.4

0.0

31.7

0.1

7.3

9.1

7.1

11.7

10.8

Baghouse (fabric filter), reverse air

BR

4.3

15.8

9.4

0.0

0.6

1.4

6.8

11.4

4.1

Baghouse (fabric filter), shake and deflate

BS

0.0

0.0

1.9

0.0

0.0

0.0

0.8

0.0

0.3

Electrostatic precipitator, cold side, with flue gas conditioning

EC

0.0

0.0

6.1

0.1

6.1

7.6

4.2

2.3

5.6

Electrostatic precipitator, hot side, with flue gas conditioning

EH

0.0

0.0

2.4

0.2

0.0

0.2

0.0

0.0

0.3

Electrostatic precipitator, cold side, without flue gas conditioning

EK

0.0

0.0

16.0

3.4

32.6

19.0

7.9

6.0

17.9

Electrostatic precipitator, hot side, without flue gas conditioning

EW

0.0

0.0

5.1

0.2

1.7

3.0

0.0

3.4

2.5

Multiple cyclone

MC

5.9

0.0

0.7

0.3

0.0

1.6

0.0

0.7

0.8

Single cyclone

SC

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

Nitrogen Control Technologies



12.2

15.9

71.6

27.3

61.7

49.6

45.7

55.0

54.0

Advanced overfire air*

AA

0.0

0.0

10.9

0.4

0.1

0.2

2.5

9.7

3.0

Biased firing*

BF

0.0

0.0

0.0

0.0

0.2

0.0

0.2

0.0

0.1

Fluidized bed combustor*

CF

0.0

15.8

1.7

0.1

1.3

1.3

1.2

0.3

1.2

Flue recirculation*

FR

0.0

0.0

1.0

0.5

0.7

1.1

1.2

0.3

0.9

Fuel reburning*

FU

0.0

0.0

0.0

0.1

0.0

0.7

0.0

0.1

0.3

Water injection*

H20

2.1

0.0

0.1

2.9

0.2

0.4

0.2

0.0

0.4

Low excess air*

LA

5.5

0.0

0.6

1.6

0.1

1.6

1.1

0.7

0.9

Low NOx burner*

LN

4.3

0.0

58.4

12.4

41.0

29.4

33.9

43.6

36.9

Ammonia injection*

NH3

5.5

0.0

2.6

3.7

6.9

2.0

5.2

8.4

4.6

Overfire air*

OV

4.3

0.1

48.1

5.5

19.8

18.0

16.0

19.3

20.9

Selective noncatalytic reduction

SN

4.3

15.8

8.7

0.3

4.1

3.3

1.9

4.1

3.9

Selective catalytic reduction

SR

5.8

0.0

25.8

20.0

50.6

34.2

22.7

24.4

33.7

Steam injection*

STM

0.0

0.0

0.0

0.2

0.1

0.2

0.3

0.3

0.2

Mercury Control Technologies



0.0

0.0

52.9

0.2

11.1

19.8

26.8

26.8

22.2

Activated carbon injection system

ACI

0.0

0.0

52.7

0.2

10.9

19.2

26.8

26.8

21.9

Lime injection*

LIJ

0.0

0.0

2.4

0.1

0.3

1.3

0.0

0.0

0.8

Other

OT

0.0

0.0

6.9

0.0

1.2

2.0

0.0

10.1

3.2

Total fossil fuel generation (Millions of MWh)



3.5

8.3

262.2

99.8

595.9

937.6

291.0

393.2

2,591.5

Note: Change-in-process technologies are marked with an asterisk.


-------
newable Energy Laboratory Annual Technology Baseline (National Renewable Energy Laboratory,
2021) to estimate the generation costs of new installations. The NREL ATB provides cost informa-
tion for over 20 technologies with particular focus on renewable technologies including onshore and
offshore wind, utility-scale solar PV, hydropower, geothermal, biomass power, and battery storage.
As our focus is on estimating capital and compliance cost of the current operating fleet, which
is becoming more dominated by renewable and natural gas-fired technologies, we elect to use the
ATB cost information over other potential sources. Other sources for cost data are discussed in the
"Data Assumptions" section.

For some renewable technologies, such as onshore wind and hydropower, there is further classifica-
tion of CAPEX values based on resource potential. Fossil technologies (coal and natural gas) have
a range of CAPEX values depending on whether carbon capture controls are included. Finally,
for other technologies like nuclear, ATB reports only one CAPEX value. For the relevant generat-
ing sources, the costs of pollution controls such as nitrogen oxides (NOx) and sulfur oxides (SOx)
technologies are included in the capital expenditure in ATB. Regional variability in CAPEX is not
incorporated into ATB. Costs could be weighted using the regional indices from the Integrated
Planning Model Version 6 (U.S. Environmental Protection Agency, 2018), but that variation is
modest and we do not incorporate it here.

We map the ATB cost information onto Form 860 power plants based on their PM-FT configuration.
EIA prime movers without ATB cost information include compressed air energy storage, pumped
storage, flywheel energy storage, hydrokinetic axial flow turbines, hydrokinetic wave buoys, and
fuel cells. We do not fill these with substitute capital and O&M expenditure data, but these units
account for less than 0.05 percent of total net generation across the 7 years of EIA data.

The NREL ATB data has been released annually since 2015. Costs in each ATB release, including
the 2020 version, are projected for multiple years beginning in the base year. We apply the base
year cost data (2020) from ATB 2020 and do not use costs from projected years, as these data are
modeled rather than historical. To produce annual capital cost amounts, we amortize overnight
capital costs for 30 years at a weighted average cost of capital of 5.5 percent per ATB assumptions
(National Renewable Energy Laboratory, 2021). Installations older than 30 years are set to operate
with no capital cost of life extension (i.e., amortized value of overnight capital times zero percent).
This is a limitation of the dataset and the true sector-wide generation cost might be higher than
in MEEDE due to the presence of life extension costs in the real world.

Variable costs exclude fuels and include materials costs such as water and chemical solvents used in
operating the generation and control equipment. Fixed O&M costs pertain largely to labor expenses
incurred in the daily operations of the plant (U.S. Energy Information Administration, 2020).

Data Assumptions

The generation cost estimates from ATB are for new installations only. Applying these cost esti-
mates to facilities with a wide variety of vintages leads to an inaccurate assessment of their true
costs. True capital costs are likely lower and O&M costs higher than the ATB estimates. Capi-
tal costs from ATB are reduced through the amortization of overnight costs, which mitigates this
limitation, but the overall level is likely still higher than those prevailing at construction time for
older-vintage installations. This limitation could be mitigated with better available data; however,
data for new installations are often proprietary and difficult to obtain.

Relately, the cost data from ATB represents fewer generation technology specifications for fossil

14


-------
fuel generators than the EIA Forms 923 and 860 data provide. For example, coal CAPEX (in
$/kW) in ATB is estimated for an advanced supercritical power plant, but there are coal plants
in the EIA dataset with different boiler technologies (e.g., subcritical boilers), which would lead to
different capital costs. Here, we assume that the costs from ATB can be applied to all of the fossil
fuel plants within each EIA technology specification. This likely results in MEEDE having higher
than expected total capital expenditure for these units because new builds from ATB likely include
higher efficiency boilers than found at older vintage plants.

Note that while the ATB data gives us lower cost resolution for fossil fuel technologies compared
to other data sources (discussed in more detail below), it gives higher resolution cost information
for renewable technologies. For example, land-based wind is categorized into ten cost classes based
on wind speed range. Here, we apply the average cost across the ten wind classes, but note that
this data could be leveraged in future versions of MEEDE. As previously mentioned, due to the
increased penetration of renewables on the grid in recent years, we elect to use data with higher
resolution of renewable technology costs.

Alternative sources for capital and O&M expenditure data include the EPA's Integrated Planning
Model (IPM) and the EIA's National Energy Modeling System (NEMS). The EPA IPM dataset
presents cost data for a similar set of technologies as in ATB, but has added regional and age-
dependent variability (U.S. Environmental Protection Agency, 2018). However, IPM has fewer
classifications within renewable technologies than the ATB data, for example, in comparison with
ATB's land-based wind cost classes as described above. Moreover, the IPM cost data is not released
annually but rather when there is a model update. We do not use IPM for total capital and O&M
information but we do use it for air pollution control technology costs, discussed in more detail in
the "Pollution Control Equipment Costs" section.

It should be noted that IPM relies on FERC Form 1 as its underlying source for historical plant-level
capital and O&M cost data. FERC Form 1 is publicly available and provides plant-level financial
and operational information reported annually by major electricity utilities across the U.S. (Federal
Energy Regulatory Commission, 2023). The primary benefit of FERC Form 1 is that the cost data
are reported by those subject to the jurisdiction of the Federal Energy Regulatory Commission,
and thus have broad coverage and high fidelity. The major disadvantage of FERC Form 1 is that
the raw data is time consuming to parse and clean. As such, most users of FERC Form 1 rely on
paid data services like Energy Velocity or S&P for cleaned versions of the data, which decreases the
transparency of the data. The use of FERC Form 1 can be explored in future MEEDE updates,
and could potentially rely on the Catalyst Cooperative's open-source efforts to clean FERC Form
1 (Catalyst Cooperative, 2023).

The EIA National Energy Modeling System cost dataset uses a top-down capital cost estimation
methodology derived from parametric evaluations of costs from actual or planned projects (U.S.
Energy Information Administration, 2019a). The data cover a similar set of technologies to ATB
and IPM, and do not provide regional or age variability. Unlike the ATB data, the NEMS data do
not provide a wide range of classifications within specific renewable technologies.

Figure 1 shows a comparison of capital costs for select technologies between the ATB, IPM, and EIA
sources. These represent the original data and are not the costs after being mapped to EIA's EGUs.
The ranges mark the minimum and maximum costs of each technology from each data source, where
the ranges within each technology type come from differences in technology specifications (e.g., the
different wind classes for onshore and offshore wind). The dots indicate either the given value (for

15


-------
single points without a range) or the mean value (for points with a range).

Figure 1: Capital cost comparison between data sources for select technologies
8000-

.6000

• 4000

o

o

2000-

o-

5 ^



ATB EIA IPM

Coal Steam

ATB EIA IPM
Combined Cycle

ATB EIA IPM ATB EIA IPM

Combined
Cycle +¦ CCS

Combustion
Turbine

ATB EIA IPM

Nuclear

ATB EIA IPM

Onshore Wind

ATB EIA IPM

Solar PV

Abbreviations: CCS = carbon capture and storage.

Note: The error bars mark the minimum and maximum costs of each technology from each data source. The absence
of error bars on some points does not indicate greater precision.

2.2.2 Pollution Control Equipment Costs: Integrated Planning Model Version 6
Data Assembly

In MEEDE, the costs of air pollution control equipment are developed separately from other costs
of generation for two reasons. First, the control technologies of a power plant can change through
time, so separate accounting of control equipment costs allows calculations of total generation costs
to properly reflect the changing controls. Second, in the broader application of MEEDE for the
social accounting matrix, this information can be used to demonstrate how abatement activities
can be accounted for separately in a disaggregated framework. See Section 3 for a discussion of the
application of MEEDE for a social accounting matrix.

As previously mentioned, the ATB total generation cost estimates we use are for new installations of
technologies and are inclusive of air pollution control costs required of new units. Thus, to estimate
the costs of generation alone, we subtract the cost estimates for pollution controls. Pollution control
cost estimates come from EPA's IPM Version 6 (U.S. Environmental Protection Agency, 2018). New
coal builds include sulfur oxides and nitrogen oxides controls based on ATB documentation. We
specifically assume the use of limestone forced oxidation scrubbers and selective catalytic reduction
systems, even though this level of detail is not provided by ATB. We also assume that the capital
and O&M costs of new natural gas builds do not include SOx and NOx controls because this is not
explicitly specified in ATB documentation. We do assume that both coal and natural gas ATB costs

16


-------
include activated carbon injection with an existing baghouse for mercury and particulate matter
control. A generation-only cost estimate enables us to build up the total costs from the generation
and pollution control cost components for any configuration that may be operating on the grid.

In order to build up to total costs from the generation-only and IPM control technology costs, we
need to first map the variety of control technologies identified in the Form 860 data to the smaller
set of technologies that IPM covers. Depending on the technology, this mapping varies with the type
of fuel, the firing configuration of the boiler, and the specific control equipment installations. Table
6 provides the total costs of generation and pollution control. Controlling sulfur oxides requires the
greatest expense (sum across all SOx control technologies), and total nitrogen oxides and particulate
matter controls are each less than half the cost of sulfur oxide control. Approximately one third
of all grid generation (1,422 million MWh) employs some form of pollution control. Table 7 shows
total capital, fixed O&M, and variable O&M costs for select prime movers across the EIA power
plants using calculated generation-only costs with added IPM control costs.

Data Assumptions

The approach we use to integrate ATB and IPM cost data has some limitations. First, IPM's
pollution control cost estimates are based on retrofitted installations whose costs likely differ from
new installations. Two factors address this concern: (1) older installations are more likely to have
installed their control equipment as retrofits, making the estimates appropriate, and (2) for new
installations, the control equipment costs will still be included in the total costs of the installation;
the adjustment for control equipment only shifts costs between generation and control.

Second, our method for determining generation-only costs makes assumptions about the specific
control equipment that are included in ATB capital costs costs for each fossil fuel technology.
For instance, the ATB data includes NOx controls in the costing of new coal units, but it does
not specify whether it uses selective catalytic reduction equipment, low NOx burners, or another
technology. Our specific assumptions might differ from what is actually considered in the ATB cost
data. This limitation could be addressed in the future if more granular information on the specific
control technologies considered in the ATB cost data becomes available.

2.2.3 Fuel Costs: Energy Information Administration Form 923
Data Assembly

The Form 923 data provide fuel costs and quantities at the PID-FT level but do not attribute
these to the boiler level. As such, we use local, quantity-weighted average prices for the different
fuel types to estimate total fuel costs for each BID-PID pair. The locality of the average varies
depending on data availability (i.e., where a state-level average is not available, we take a regional
average if available, a national average if not). Given fuel price data and estimates for capital
and O&M costs, we fully cost the generation capital, labor, energy, and materials inputs where we
associate fixed O&M with labor and variable O&M with materials.

Data Assumptions

The fuel receipts found in EIA Form 923 are retail or delivered prices and not wholesale prices.
Wholesale fuel price data might be preferred in some instances such as when using MEEDE in a
social accounting matrix that applies wholesale prices. A method for determining wholesale fuel
prices is through mapping the contract start year for each fuel receipt to the EIA wholesale price

17


-------
Table .6: Costs of generation and pollution control, 2019

Installation Attribute



Total Costs Excluding Fuel (MM 2019$)

Cost Component

Net Generation (in thousands of MWh)

Capital

Fixed 08cM

Variable 08tM

Total

Total system, excluding fuel

4,120,300

85,094

56,171

15,428

156,693

Generation only

4,120,300

80,008

52,770

9,428

142,206

Controls

1,422,334

5,086

3,401

5,999

14,486

SOX Controls

873,182

2,278

2,593

3,356

8,226

Limestone forced oxidation scrubber

639,225

1,344

1,985

2,443

5,772

Lime spray dryer scrubber

249,216

1,131

859

1,032

3,022

Dry sorbent injection

117,034

246

342

1,054

1,642

PM Controls

1,009,184

1,341

372

79

1,792

Electrostatic precipitator (cold side)

621,454

349

224

49

621

Fabric filter baghouse

414,560

977

171

38

1,185

Electrostatic precipitator (cold side), with flue gas conditioning

145,072

88

44

11

144

Electrostatic precipitator (hot side)

91,291

235

53

9

298

Electrostatic precipitator (hot side), with flue gas

11,185

107

8

1

117

NOX Controls

1,406,151

1,432

416

1,120

2,969

Selective catalytic reduction

875,431

1,325

307

966

2,599

Low-NOx burner (wall fired)

687,370

1,074

269

764

2,107

Low-NOx burner with advanced overfire air (tangentially fired)

303,937

25

48

6

80

Low-NOx burner with overfire air (wall fired)

251,206

579

147

396

1,122

Selective noncatalytic reduction (fluidized bed)

18,208

21

3

53

77

Vertically fired

4,584

15

1

7

23

HG Control - Active Carbon Injection

586,773

35

20

1,444

1,498

Abbreviations: Hg = mercury; MWh = megawatt-hour; \O X = nitrogen oxide; O&M = operationsand maintenance;
PM = particulate matter; ;SOX = sulfur oxide.

Notes: Total system cost = cost of generation plus cost .of controls. Amounts do not sum across: controls because
many installations run multiple controls (e.g., selective catalytic reduction with a low-NOx wall-fired burner,).

data from the corresponding year. While Form 923 does not currently provide information on the
contract start date for each fuel receipt (it provides only the contract end date], it might do so in
the future.

2.2.4 Wholesale Prices: Federal Energy Regulatory Commission Form 714
Data Assembly

The final step in costing is to estimate the wholesale value of the electricity produced. Wholesale
electricity price data is useful for assessing the revenue from wholesale electricity sales in the; U.S.
Note that this can differ substantially from the retail price of electricity for a variety of reasons,
particularly in regulated markets (Cappers and Murphy, 2019).: We do not assess electricity sector

18


-------
Table 7: Capital and O&M costs across EIA power plants for select prime movers, 2019



Installation Attribute



Total Costs Including Pollution Control Technologies
(MM 2019$)

Prime Mover Type

Code

Net Generation (in thousands of
MWh)

Capital Cost

Fixed O&M Cost

Variable O&M Cost

Combined Cycle (Steam)

CA

429,811

1,789

296

1,006

Combined Cycle (Single)

CS

58,808

1,081

195

133

Combined Cycle
(Combustion)

CT

860,912

20,324

3,563

1,922

Gas Turbine

GT

136,102

9,792

2,113

651

Hydraulic Turbine

HY

287,830

1,940

7,340

0

Internal Combustion
(Diesel)

IC

16,005

589

145

77

Photovoltaic

PV

68,611

4,328

740

0

Steam Turbine

ST

1,965,224

29,205

36,602

11,635

Wind (Onshore)

WT

294,693

15,306

4,955

0

Note: Amounts do not sum to totals because these show a selection of all prime mover types).

revenue from retail sales. For estimating the wholesale value, we rely on data from FERC Form
714, which are available for years 2006 to 2019 (Federal Energy Regulatory Commission, 2021).

The FERC Form 714 provides hourly system marginal cost ($/MWh) and hourly electricity demand
(MWh) data for over 200 balancing authority areas (BAs). The system marginal cost (or the
"system lambda") is the marginal cost of the generating plant that meets the last MWh of electricity
demanded. We use the hourly electricity demand to weight the system lambda data, whereby hours
with greater electricity demand volume have higher weights for the system lambda.

In order to map FERC's BA-level data to the plant-level data produced in EIA Forms 860 and 923,
we match Form 860's Balancing Authority Code for each PID to Form 714's respondent ID descrip-
tion. Approximately 65 percent of BID-PID pairs in 2019 can be matched to a FERC respondent
ID. This percentage is lower for earlier EIA years (56 percent in 2013) due to discrepancies and
changes in FERC respondent IDs, which resulted in more gaps when mapped to the EIA data.
Plants without available Balancing Authority Codes across all years are assigned the NERC region
volume-weighted average price. These plants are primarily in the Alaska Systems Coordinating
Council (ASCC) and the Hawaiian Islands Coordinating Council (HICC). In 2019, the U.S. average
annual wholesale price of electricity was around $30/MWh, which aligns with other sources such
as the U.S. EIA.

Because FERC Form 714 data are available at the hourly level, we aggregate the values into lower
temporal resolution. Specifically, for each balancing area in each year from 2013 to 2019, we report
the annual weighted system lambda in dollars per MWh generated. The hourly-level data can be
aggregated in any way required by the user, for example by month or season.

Data Assumptions

19


-------
In collecting the wholesale cost data and assigning a value to each EIA plant based on the balancing
area, we make two main assumptions. First, we assume that the plants within each EIA Balancing
Authority Code overlap entirely with the plants covered by each corresponding Form 714 respondent
ID. For example, FERC Form 714 lists "Southern Company" as a respondent, which we then
matched to EIA BA Code "SOCO". All plants with Balancing Authority Code "SOCO" would
then be assumed to fall under FERC balancing area "Southern Company". However, it is possible
that there are discrepancies between the FERC and EIA datasets in terms of which plants fall
under each balancing authority.

Second, we assume that state-level averages are representative estimates for plants where FERC
data is not available. Utilities that are not required to report Form 714 may have different wholesale
cost and demand profiles than those required to report. However, without additional information
indicating where these differences may occur, the most reliable approximation is to use state-wide
averages based on available data.

An alternative source for wholesale price data is the Intercontinental Exchange (ICE), which reports
wholesale daily spot prices at hubs across the U.S. The prices listed on the ICE are only those for
deregulated markets so may differ from the data provided in FERC Form 714.

2.2.5 Economic Data Description

Table 8 shows the capital costs (in $/MWh) for the technology configurations used to make the
MEEDE dataset. The generation-only costs columns show the revised costs net of included con-
trol equipment as costed from IPM Version 6. This adjustment affects all coal and natural gas
technologies. Control equipment cost adjustments vary by technology and are greatest for coal
units (up to 25 percent). With the exception of carbon capture and sequestration controls, control
cost adjustments for combined-cycle plants are relatively minor (3 to 10 percent difference between
generation-only and EIA total costs).

Meanwhile, Table 9 summarizes the total generation costs or expenditures (generation plus controls)
for the different plant types by region. Advanced supercritical coal installations dominate by total
generation costs across all regions, followed by natural gas combined cycle and combustion turbine
plants. Between regions, the South Atlantic (SERC) has the highest overall costs because the
greatest number of plants exist in these states. The West North Central states (MRO) have high
coal and gas total generation costs but low solar generation costs, due to lower prevalence of solar
installations. The WECC region has similar coal generation costs but much higher solar generation
costs.

Lastly, Table 10 summarizes the wholesale revenue and total costs of electricity generation by re-
gion. The national wholesale revenue—again, represented by hourly system lambdas—is slightly
more than half of total cost. This does not necessarily indicate that sector-wide costs of generation
are greater than revenue earned. Rather, this can be at least in part attributed to the regional dif-
ferences in wholesale prices and other sources of generator revenue (e.g., ancillary service markets).
Wholesale prices may be far less than total system revenue. Depending on the market, retail prices
of electricity can be more than two times the wholesale price of generation (Cappers and Murphy,
2019) as retail prices include taxes and state surcharges, costs of delivery, and purchasing costs in
addition to generation and ancillary service costs. In recent years, changes in retail prices have

20


-------
Table 8: Total capital and O&M costs and generation-only capital and O&M costs by plant type

1



ATB Total Costs





Generation-Only Costs



Configuration

Capital
(S/MW)

Fixed O&M
(S/MW-yr)

Variable O&M
($/MWh)

Capital
(S/MW)

Fixed O&M
(S/MW-yr)

Variable O&M
(S/MWh)

Coal



Advanced Supercritical Coal Avg CF +
CCS

6.06

0.06

8.83

4.95

0.05

723

Advanced Supercritical Coal High CF +
CCS

6.06

0.06

8.83

5.03

0.05

723

Coal IGCC Avg CF

4.74

0.06

7.88

3.69

0.05

6.39

Coal IGCC High CF

4.74

0.06

7.88

3.77

0.05

6.39

Advanced Supercritical Coal Avg CF

4.15

0.04

4.40

3.10

0.03

2.91

Advanced Supercritical Coal High CF

4.15

0.04

4.40

3.18

0.03

2.91

Natural Gas



Natural Gas Combined Cycle Avg CF

1.09

0.01

2.16

0.99

0.01

2.01

Natural Gas Combined Cycle High CF +
CCS

2.78

0.03

5.72

2.68

0.03

5.56

Natural Gas Combined Cycle High CF

1.09

0.01

2.16

0.99

0.01

2.01

Natural Gas Combustion Turbine Avg CF

0.98

0.01

4.50

0.89

0.01

4.34

Nuclear



Advanced Nuclear

7.19

0.12

2.32

7.19

0.12

2.32

Renewables



Biomass

435

0.12

4.72

4.35

0.12

4.72

Concentrated Solar Power

7.83

0.07

4.20

7.83

0.07

4.20

Geothermal - Binary

8.70

0.18

0.00

8.70

0.18

0.00

Geothermal - Dual Flash

6.70

0.14

0.00

6.70

0.14

0.00

Onshore Wind

1.96

0.04

0.00

1.96

0.04

0.00

Hydropower New Stream-Reach
Development

6.95

0.08

0.00

6.95

0.08

0.00

Offshore Wind

534

0.11

0.00

534

0.11

0.00

Solar Photovoltaic

1.60

0.02

0.00

1.60

0.02

0.00

Abbreviations: Avg CF = average capacity factor; CCS = carbon capture and storage; High CF = high capacity
factor; IGCC = integrated gasification combined cycle.

Note: Generation-only costs equal ATB total costs less EPA Integrated Planning Model retrofit costs of the ATB
technology specification's control equipment.

become decoupled farther from wholesale prices and the costs of generation (Cappers and Murphy,
2019), This is particularly so in regulated markets (e.g., SERC) where retail revenue can support
cost recovery for integrated utilities' generation assets. Table 10 indicates considerable regional
variation, with some regions presenting greater discrepancy between total costs and wholesale rev-
enue (e.g., SERC). The MEEDE dataset's estimated annual cost of operating generation and control

21


-------
Table 9: Total generation cost by plant type and region, 2019

Generation Costs (MM 2019$)

ATB Technology

MRO

NPCC

RFC

SERC

TRE

WECC

ASCC

HICC

Total USA

Advanced Nuclear

697

2,062

5,182

7,744

2,100

1,160

0

0

18,946

Advanced Supercritical Coal Avg CF

5,935

990

4,941

8,382

1,352

3,939

53

784

26,377

Advanced Supercritical Coal High CF

5,596

704

15,657

20,344

6,275

5,379

0

0

53,955

Biomass

550

395

789

2,715

86

483

0

57

5,076

Coal IGCC Avg CF

0

0

493

0

0

0

0

0

493

Concentrated Solar Power

0

0

0

587

0

757

0

0

1,344

Geothermal - Binary

0

0

0

0

0

870

0

0

870

Geothermal - Dual Flash

0

0

0

0

0

566

0

0

566

Hydropower New Stream-Reach Development

507

726

407

1,748

55

5,661

162

13

9,280

Natural Gas Combined Cycle Avg CF

2,512

4,253

7,295

16,015

4,955

9,168

212

358

44,768

Natural Gas Combined Cycle High CF

479

722

4,675

7,988

2,072

1,206

0

0

17,142

Natural Gas Combustion Turbine Avg CF

2,144

880

4,218

6,350

1,088

2,842

260

128

17,911

Offshore Wind

0

16

0

0

0

0

0

0

16

Onshore Wind

7,530

671

2,251

421

5,023

4,312

12

40

20,261

Solar Photovoltaic

172

226

286

1,402

343

2,658

0

38

5,126

Abbreviations: Avg CF = average capacity factor; High CF = high capacity factor; IGCC = integrated gasification
combined cycle.

Regions: ASCC = Alaska Systems Coordinating Council; HICC = Hawaiian Islands Coordinating Council; MRO
= Midwest Reliability Organization; NPCC = Northeast Power Coordinating Council; RFC = ReliabilityFirst Cor-
poration; SERC = Southeastern Electric Reliability Council; TRE = Texas Reliability Entity; WECC = Western
Electricity Coordinating Council.

Note: Recall that each unit in MEEDE is assigned a cost from one of these ATB technologies.

equipment in 2019 is $222,452 million (2019 dollars). While wholesale prices in regulated markets
do not cover all generation costs, they do cover all variable costs (i.e., fuel and variable O&M),
which should be the basis for generators' bids in competitive markets.

Generators have other revenue streams in addition to wholesale revenues. Generators in many
regions have opportunities to participate in a variety of auctions and contract types, such as in
ancillary service or derivative markets (e.g., futures). Revenue from this type of market participation
would not be captured by hourly system lambdas and wholesale revenue. In addition, infrequent
and often brief market conditions such as extreme weather events may have outsized impacts on
generator revenue from participation in other markets, which are not captured by the wholesale
revenue.

On the cost side, the total cost of electricity generation may be smaller in reality than estimated
here. This is because MEEDE's amortization of capital costs may differ from the actual annual
costs of a plant. For instance, plants may payoff more of their debt obligations in early years when
O&M cost are lower than is assumed with a fixed amortization schedule at an assumed debt term
of 30 years.

22


-------
Table 10: Wholesale revenue and total costs by NERC region, 2019



MM 2019$

I NERC Region

Wholesale Revenue

Fuel Costs

Capital Costs

Fixed O&M Costs

Variable O&M Costs

Total Costs

ASCC

362

370

220



91

18

699

HICC

1,152

1,021

279



101

42

1,444

MRO

11,915

5,602

12,041



6,638

1,849

26,130

NPCC

6,133

3,113

4,316



3,802

483

11,714

RFC

24,894

14,786

14,806



12,691

3,978

46,262

SERC

32,831

24,826

26,279



17,136

5,550

73,790

TRE

11,884

5,668

11,732



4,564

1,385

23,350

WECC

31,006

10,347

15,421



11,148

2,121

39,036

Total USA

120,177

65,732

85,094



56,171

15,428

222,425

Regions: ASCC = Alaska Systems Coordinating Council; HICC = Hawaiian Islands Coordinating Council; MRO
= Midwest Reliability Organization; NPCC = Northeast Power Coordinating Council; RFC = ReliabilityFirst Cor-
poration; SERC = Southeastern Electric Reliability Council; TRE = Texas Reliability Entity; WECC = Western
Electricity Coordinating Council.

2.3 Environment

The final step in creating the full MEEDE dataset is to identify the emissions associated with each
observation generated in the previous steps. We include emissions data in MEEDE because the
electricity sector is a primary contributor of greenhouse gas and criteria pollutant emissions. In
the U.S., the electricity sector produces an estimated 25 percent of total greenhouse gas emissions
per year (U.S. Environmental Protection Agency, 2021a). Users of the MEEDE dataset can apply
the emissions information for example into economy-wide models for simulating changes in total
emissions under various scenarios. We estimate emissions for eight pollutants: three greenhouse
gases (C02, CH4, N2O), three criteria pollutants (particulate matter, SOx, NOx), and mercury.
For this, we rely on data from a variety of EPA sources, described in more detail in the following
sections.

Note that in this update of MEEDE, we do not include emissions from fluorinated gases (F-gases).
F-gas emissions from the electricity sector are primarily fugitive emissions from transmission infras-
tructure, especially from older equipment (U.S. Environmental Protection Agency, 2021b). Because
of this, F-gas emissions cannot be readily allocated to generators based on generation capacity or
other attributes, and are therefore excluded from MEEDE.

2.3.1 Particulate Matter Emissions: 2018 EPA Mercury and Air Toxics Standards
Residual Risk and Technology Review

Data Assembly

Emissions of particulate matter 2.5 in 2018 are compiled for the 2020 Mercury and Air Toxics Stan-
dards (MATS) residual risk and technology review (RTR; U.S. Environmental Protection Agency
(2017)). This dataset provides particulate matter emissions quantity (annual total in lbs, which

23


-------
we convert to tons) and emissions rate (lbs/MMBtu, converted to tons/MMBtu) for all generating
units that were required to report tests which were conducted for compliance with the final MATS
rule. These include any fossil fuel-fired combustion units greater than 25 megawatts-electric (MWe)
that serve a generator producing electricity for sale (U.S. Environmental Protection Agency, 2011).
Unlike many of the other emissions datasets, which are discussed in more detail in the following
sections, the particulate matter data are not released annually nor are they presented at sub-annual
resolution. Rather, this dataset was produced only in 2018 with one annual average emissions per
generating unit. The MATS RTR reports emissions of filterable PM 2.5 but not condensible PM
2.5 or any PM 10.

We map the particulate matter emissions onto EIA Forms 860 and 923 by matching the plant IDs
in the MATS RTR with PID in the EIA Forms. While the MATS RTR reports at the generation
unit level, we aggregate particulate matter emissions to the plant level for mapping because unit
ID codes in MATS RTR do not match BIDs in the EIA Forms. Particulate matter emissions
are not equally distributed among BIDs per each PID, but depend on the quantity of fuel input,
the fuel type, and the emissions controls associated with each BID, so we have to apportion the
particulate matter emissions to each BID-PID pair. To do this, we first use the EPA's AP-42
(U.S. Environmental Protection Agency, 1995) to assign emissions abatement efficiencies to each
BID-PID pair based on the associated control equipment. We also use AP-42 to allocate emission
factors to each BID-PID depending on the fuel type(s) used. Then, we use each BID-PID's heat
input to weight the emissions.

As previously mentioned, not all EGUs report emissions within the MATS RTR due to reporting
requirements. We calculate a weighted average emissions rate (tons/MMBtu) by prime mover and
fuel type and apply it to non-reporting units. Doing this increases PM emissions by 4 percent
relative to the original MATS RTR data.

Data Assumptions

Because annual data is not available for particulate matter emissions, we assume that the 2018 emis-
sions are representative of other years. In reality, it is possible that power plant control equipment
configurations change through time, thereby changing the emissions associated with each BID-PID
pair. An alternative approach for assigning particulate matter emissions to Forms 860 and 923 is to
do so by matching on the control equipment configuration (e.g., all of the possible combinations of
SOx, NOx, etc. controls) between the two datasets. For this, we would use data from the 2012 EPA
Maximum Achievable Control Technology (MACT) Information Collection Request (ICR) rather
than the 2018 MATS RTR because the former reports more detail in terms of the type of pollutant
control equipment associated with each generating unit (U.S. Environmental Protection Agency,
2012). However, for this version of MEEDE, we do not take this approach because we elect to use
the more recent particulate matter emissions data.

Another uncertainty in the PM emissions data stems from the lack of data for relevant (fossil
fuel and other PM emitting) units smaller than 25 MWe, which generated less than 1 percent
of electricity output in 2019. In applying the weighted average emissions rates to fill this missing
data, we assume that the emissions rates of larger units are representative of smaller units. Without
additional information on emissions rates of small EGUs, this method provides the best estimate
given available data and should not impact sector-wide results substantially as emissions only
increase by 4 percent.

24


-------
2.3.2	Mercury Emissions: EPA Mercury and Air Toxics Standards
Data Assembly

Mercury emissions are continuously monitored under MATS and available from the EPA Air Mar-
kets Program Data (AMPD; U.S. Environmental Protection Agency (2021c)). As with particulate
matter emissions under MATS, all fossil fuel-fired combustion units that serve a generator greater
than 25 MWe are required to report mercury emissions. Hourly emissions data (quantity given in
lbs and rate in lb/MMBtu) from 2015 to 2020 are available per generating unit per plant for all
plants that must report under MATS. Because the raw data is hourly, we aggregate the emissions
to annual as well as to predefined "time-slice" resolution. These time-slices correspond to NREL's
Regional Energy Deployment Systems (ReEDS) model's default 16 time-slices that are meant to
represent each of the four seasons as well as four times of day when electricity demand might vary
(overnight, morning, afternoon, and evening; Cohen et al. (2019)). MEEDE provides a total of
36 reported statistics for Hg emissions in each year: one annual emissions total (tons), one annual
median value to represent the hourly emissions (tons), one annual 25th percentile value (tons), one
annual 75th percentile value (tons), 16 time-slice emissions totals (tons), and 16 time-slice median
rates (tons/MMBtu).

An interesting point to note is that hourly emissions of mercury can be excessively high when
generators are ramping up or down. In the entire hourly time series from 2015 to 2019 for all
reporting plants, outlier values (greater than 1.5 times the interquartile range) occurred in less
than 5 percent of data. We do not remove these data points from our aggregated annual totals or
time-slice medians because they represent real power plant operations and emissions during these
times. However, we note that these values could skew the totals and medians in one year compared
to another if more frequent ramping occurs (for instance, if coal-fired power plants are treated as
load following plants due to the increased penetration of renewable generators).

Similar to the particulate matter data, the unit ID codes presented in the EPA AMPD do not
correspond perfectly with the BIDs in the EIA Forms. As such, we aggregate mercury emissions
to the plant level and then apportion emissions to each BID-PID pair using abatement efficien-
cies and emission factors from AP-42, and heat input data from Form 923. Moreover, since not
all units report mercury emissions under MATS, we calculate a weighted average emissions rate
(tons/MMBtu) by prime mover and fuel type and apply it to non-reporting units. This increases
emissions by less than 3 percent relative to the totals reported in the original source.

Data Assumptions

In addition to limitations introduced by the lack of emissions data for smaller EGUs, our use of AP-
42 data adds uncertainty to the final mercury emissions as well. The apportionment of plant-level
mercury emissions to boiler level relies on AP-42, which was developed based on 1995 data. This is
the most accurate data to date because more recent data is not available. As such, we have more
confidence in the plant-level emissions estimates than in the boiler-level estimates. In the future, if
EPA AMPD reported generation unit codes are able to align more closely with EIA boiler and/or
generator IDs (the way that plant IDs are standardized across reporting platforms), then the data
could be matched unit-to-unit rather than require allocation using AP-42.

2.3.3	SOx and NOx Emissions: EPA Air Markets Program
Data Assembly

25


-------
Emissions of sulfur and nitrogen oxides are provided by EPA's Office of Air and Radiation's Clean
Air Markets Division and also reported at the hourly level in the AMPD (U.S. Environmental
Protection Agency, 2021c). All fossil fuel-fired units serving a large generator (greater than 25
MWe) that provides electricity for sale are required to report SOx and NOx emissions. For units
less than 25 MWe that are not required to report, we apply the weighted average emissions rate
(lb/MMBtu) by plant and fuel type. Doing this increases SOx and NOx emissions by less than 1
percent each compared to the totals reported in the original sources.

The data provides unit-level emissions that we summarize at the plant level and allocate to the
boiler level based on the relative estimates from AP-42 emissions factors. As with mercury emissions,
hourly SOx and NOx emissions and emissions rates are aggregated to annual resolution from 2013
to 2019 and time-slice resolution per year.

Data Assumptions

We make the same data assumptions here as we do for PM and particularly mercury. Therefore,
the same improvements can be made in future iterations if EPA and EIA reporting units become
better aligned.

2.3.4	Greenhouse Gas Emissions: EPA Greenhouse Gas Reporting Program
Data Assembly

Annual emissions of carbon dioxide (CO2), nitrous oxide (N2O), and methane (CH4) from 2011-2019
are reported by the "Envirofacts" utility of EPA's Greenhouse Gas Reporting Program (GHGRP)
at the power plant level (U.S. Environmental Protection Agency, 2021d). Under the Greenhouse
Gas Reporting Program (codified at 40 CFR Part 98), facilities that generate more than 25,000
metric tons of C02e per year are required to submit annual greenhouse gas emissions reports (U.S.
Environmental Protection Agency, 2009). These include emissions quantity and emissions rates.
Power plants are assigned GHGRP "Facility IDs" and can be mapped directly to EIA PIDs. Using
AP-42 and EIA heat input, we allocate the annual plant-level emissions to the boiler level.

As with the other pollutants, emissions from units without GHGRP data are estimated as a weighted
average emissions rate by plant and fuel type. Emissions increase by less than 2 percent for carbon
dioxide, 6 percent for nitrous oxide, and 6 percent for methane relative to the totals reported in
the original sources.

Data Assumptions

In addition to the assumptions discussed regarding the other air pollutant emissions, further uncer-
tainty is introduced to this dataset through the manual mapping of GHGRP "Facility IDs" to EIA
PIDs. Approximately 40 percent of GHGRP power plant facilities are reported with an EIA plant
code. The remaining 60 percent are mapped using a supplementary crosswalk between GHGRP
facility IDs and EIA PIDs developed by the EPA (U.S. Environmental Protection Agency, 2020).
We assume its accuracy but note that there could be erroneous mappings.

2.3.5	Environmental Data Description

Table 11 summarizes the U.S. electricity sector emissions quantity by pollutant and region. The

26


-------
SERC region has the highest emissions across all criteria pollutants and greenhouse gases, but
also has the highest net electricity generation. The Mid-Atlantic and Midwest region (RFC) has
the second highest generation as well as the second highest level of emissions for all greenhouse
gases and pollutants except mercury. Meanwhile, the U.S. West (WEC-C) has the third highest net
generation but substantially lower SOx emissions than many of the other regions, including MRO
and TRE. Table 12 summarizes the U.S. electricity sector emissions rates by pollutant and region.
The West North Central states (MRO) have the highest emissions rates across most of the criteria
pollutants and greenhouse gases. Meanwhile, Hawaii (HICC) and the New England states (NPCC)
have the lowest emissions rates overall. Note that the HICC particulate matter emissions rate is an
outlier here, as it is the highest of the eight regions. This is because over 85 percent of the units in
Hawaii are oil-fired, which tends to have a higher particulate matter emissions rate.

Table 11: Criteria pollutant and greenhouse gas emissions by region, 2019

NERC

Output

Net Generation (in thousands of
MWh)

Criteria and Hazardous Air Pollutants (tons)

Sulfur Nitrogen Particulate
Oxides Oxides Matter Mercury

Greenhouse Gases (MMT C02e)

Carbon Nitrous
Dioxide Oxides Methane

ASCC

6,068

79

229

62

0.01

0

0.01

0.01

HICC

9,750

407

350

617

0.00

0

0.01

0.01

MRO

447,052

155,125

134,271

10,257

1.50

231

0.86

0.31

NPCC

231,593

2,558

10,666

181

0.01

48

0.04

0.02

RFC

918,960

247,095

199,421

21,279

1.18

425

1.47

0.71

SERC

1,354,551

270,898

253,891

36,514

1.52

582

1.74

0.97

TRE

414,237

106,528

65,224

5,365

1.02

183

0.49

0.27

WECC

738,088

78,558

142,187

17,840

0.81

274

0.86

0.46

Total

4,120,300

861,248

806,239

92,115

6.03

1,743

5.49

2.74

Regions: ASGG = Alaska Systems Coordinating Council; HICC = Hawaiian Islands Coordinating Council; MRO
= Midwest Reliability Organization; NPCC = Northeast. Power Coordinating Council; RFC = ReliabilityFirst. Cor-
poration; SERC = Southeastern Electric Reliability Council; TRE = Texas Reliability Entity; WECC = Western
Electricity Coordinating Council.

Table 13 illustrates the impact of control technologies on emissions of sulfur oxides. Among coal-
fired generators, 94 percent of generation is covered by some type of sulfur control equipment.
Limestone forced oxidation (LSFO) and lime spray dryer (LSD) technologies are most prevalent.
Emissions rates for controlled equipment are between 38 percent (dry sorbent injection for sub-
bituminous coal) and 87 percent (LSFO for sub-bituminous coal) lower than those for uncontrolled
generation. The use of multiple control devices can reduce emissions rates by upward of 98 percent.

2.4 Full MEEDE Dataset

The MEEDE dataset provides a highly detailed description of the engineering, economic, and en-
vironmental attributes and quantities of the U.S. electric grid at the level of individual generating
units. The aggregate quantities maintain good fidelity with estimates from secondary sources,
suggesting some success in methodology and construction. Table 14 shows a comparison of total

27


-------
Table 12: Criteria pollutant and greenhouse gas emissions rates by region, 2019



Output



Criteria and Hazardous Air Pollutants (tons/MWh)

Greenhouse Gases (tons C02e/MWh)

NERC

Net Generation (in thousands of
MWh)

Sulfur
Oxides

Nitrogen
Oxides

Particulate
Matter

Mercury

Carbon
Dioxide

Nitrous
Oxides

Methane

ASCC

6,068

0.013

0.038

0.010

0

72

1.97

1.06

HICC

9,750

0.042

0.036

0.063

0

42

1.04

0.54

MRO

447,052

0.347

0.300

0.023

0

516

1.92

0.69

NPCC

231,593

0.011

0.046

0.001

0

208

0.18

0.11

RFC

918,960

0.269

0.217

0.023

0

463

1.60

0.77

SERC

1,354,551

0.200

0.187

0.027

0

430

1.29

0.71

TRE

414,237

0.257

0.157

0.013

0

441

1.18

0.65

WECC



738,088

0.106

0.193

0.024

0

371

1.17

0.63

Total



4,120,300

0.209

0.196

0.022

0

423

1.33

0.67

Regions: ASCC = Alaska Systems Coordinating Council; HICC = Hawaiian Islands Coordinating Council; MRO
= Midwest Reliability Organization; NPCC = Northeast Power Coordinating Council; RFC = ReliabilityFirst Cor-
poration; SERC — Southeastern Electric Reliability Council; TRE = Texas Reliability Entity; WECC = Western
Electricity Coordinating Council.

Table 13: Coal-fired emissions of sulfur oxide by boiler-level control type, 2019



Emissions Rate (tons/MMBTU)

Percent Reduction from Uncontrolled (%)

Controls

Net Generation (in thousands of
MWh)

Other
Coal

Bituminous

Sub-
Bituminous

Other
Coal

Bituminous

Sub-
Bituminous

DSI

44,937

0.207

0.239

0.288

53.2

65.1

37.8

DSI+LSD

1,641

0.158

0.284

0.044

64.2

58.6

90.5

DSI+LSD+LSFO

5,807

0.029

0.365

0.000

93.4

46.7

100.0

DSI+LSFO

59,039

0.175

0.144

0.014

60.4

79.0

97.1

LSD

163,801

0.235

0.161

0.109

46.7

76.4

76.5

LSD+LSFO

54,388

0.008

0.161

0.125

98.3

76.4

73.0

LSFO

503,526

0.176

0.128

0.062

60.2

81.3

86.5

No Controls

126,492

0.442

0.685

0.464

0.0

0.0

0.0

All Coal-Fired Generation

959,630

0.231

0.143

0.180



Abbreviations: DSI = dry sorbent injection; LSD = lime spray dryer; LSFO = limestone forced oxidation.
Note: Combinations of controls shown here occur at the plant level, not boiler level.

heat input and total fuel expenditures by major fuel type (Goal, natural gas, and petroleum) in
2019 between MBEDB and the ElA?s State Energy Data System (SEDS; *D.,S. Energy Information
Administration (2019b)). SEDS also relies 011 EI A Forms 923: and 860, but applies different assump-
tions and accounting methodologies across its estimates to arrive: at the outputs shown in the table.
Coal and natural gas fuel consumption between the MEEDE and SEDS are within a few percent of
each other, demonstrating consistency between the datasets. Oil consumption, however, differs by
almost 15 percent. This is likely a difference in accounting (i.e., which El A fuel codes are classified

28


-------
as broader "Oil") rather than a discrepancy between the datasets themselves. On the other hand,
fuel expenditures diverge between the two datasets by 10-15 percent depending on the fuel type.
This indicates a difference in methodology for how state- and census-level fuel prices are calculated.
For coal, SEDS estimates fuel prices using unpublished cost data from EIA Form 923 (U.S. Energy
Information Administration, 2019b) whereas MEEDE relies only on the publicly-available version
of the data. It is possible that the publicly-available data is more heavily weighted towards higher
fuel prices. For natural gas, SEDS relies on the EIA "Natural Gas Annual" report, which uses EIA
Form 176 data and is distinct from Form 923 (U.S. Energy Information Administration, 2019b).
For oil or petroleum, SEDS estimates the price of distillate fuel oil using data from the EIA's Of-
fice of Energy Production, Conversion, Delivery (EPC-D; U.S. Energy Information Administration
(2019b)).

Table 14: Comparison of sector-wide outputs from MEEDE and SEDS, 2019



Fuel Consumption (Million MMBTU)
SEDS

MEEDE

Expenditure (MM 2019$)
SEDS

MEEDE

Coal

10,181

10,112

20,948

23,894

Natural Gas

11,674

11,979

33,975

39,950

Oil

188

165

1,689

1,887

The final MEEDE dataset for 2013 to 2019 has 91,296 observations and 133 variables. The data
provide a variety of plant-level attributes (e.g., plant name and ID, NERC and census regions,
wholesale electricity price) and boiler-level attributes (e.g., fuel type, MMBtu of fuel consumed,
emissions rates, and capital and O&M costs). Quantitative boiler-level attributes can be summed
over the entire dataset to generate grid totals, whereas plant-level attributes can apply to multiple
observations at the boiler level. Generator-level attributes (e.g., nameplate capacity, heat rate)
also apply to multiple observations. The data dictionary in Appendix I indicate the attribute level,
units (if quantitative), and data source of each variable.

To summarize the previously discussed uncertainties from the construction of the MEEDE dataset,
we have greatest confidence in the data provided by the EIA Forms 923 and 860, which includes
electricity generation, heat input, and power plant configuration information at the PID-PM-FT-
BID level. This data covers an estimated 97 percent of generation capacity in the U.S., and the
remainder is comprised primarily of small-scale solar photovoltaic systems (U.S. Energy Information
Administration, 2021c). In terms of capital and O&M costs, uncertainty arises from differences
between the air pollution control equipment included in the ATB costs and the equipment that
we assume ATB includes. This uncertainty is less relevant for renewable and nuclear technologies
without air pollution controls. The uncertainty in fuel costs comes mainly from the method we
apply to distribute EIA reported fuel cost data to the state, census, and then national level. States
and census regions with fewer EIA reported PID-PM-level fuel cost values will be more heavily
biased by the data that is available. For wholesale electricity revenue, the greatest uncertainty
arises from the method we use to fill missing data. Volume-weighted NERC average prices may
be biased if the balancing authorities that do report in FERC Form 714 are not representative
of those that do not. Finally, in terms of air pollution emissions, we have greater confidence in
the estimates for the greenhouse gases (C02, CH4, N2O) and NOx and SOx than we do for PM,
because of the availability of continuously monitored data. A limitation of the emissions data is

29


-------
the lack of information on smaller EGUs; however, our method for filling missing data increases
total emissions by no more than 6 percent across each pollutant relative to the original data, which
indicates that these non-reporting EGUs comprise only a small portion of total emissions. An
additional limitation of the mercury emissions data is the fact that it only begins in 2015.

2.4.1 Annual Trends

As can be seen in Figure 2, the number of steam turbine generators (e.g., coal) decreased between
2013 and 2019, while the number of solar and onshore wind generators both increased. In 2019,
the total number of solar generators was greater than the number of steam turbine generators, not
including the steam portion of combined cycle generators. However, in terms of actual electricity
generation, solar generates only a fraction of the total even in 2019. The decrease in coal steam
turbine net generation was met by primarily by natural gas generation (Table 3).

Due to the decrease in coal generators, along with a variety of other reasons such as increased
renewable penetration and emissions regulations, greenhouse gas and criteria pollutant emissions
decreased overall between 2013 and 2019 (Figure 3). Between 2013 and 2019, total methane and
nitrous oxide emissions from the electricity sector decreased almost 30 percent and 40 percent,
respectively. Though not shown here, total CO2 emissions experienced declines as well, decreasing
almost 25 percent between 2013 and 2019.

Figure 2: Count of select prime movers in EIA Form 923, 2013-2019

i

CA	CT

GT	HY

QJ

WT

Abbreviations: CA = combined cycle (steam); CT = combined cycles (combustion); GT = gas turbine; HY =
hydraulic turbine; IC = internal combustion (diesel); PV = photovoltaic; ST = steam turbine; WT = wind (onshore).

Total annual expenditure has decreased roughly 15 percent between 2013 and 2019 (Figure 4). This

30


-------
Figure 3: Annual emissions of criteria pollutants (nitrogen oxides and sulfur oxides) and greenhouse gases
(methane and nitrous oxide), 2013-2019

Year

NOX	SOX	CH4	N20

Note: The y-axes of the two panels differ in both range and units.

Figure 4: Annual capital, fixed O&M, variable O&M, and fuel expenditures, 2013-2019

300



g-100
LU

"ro

23

C
<

o-

type

Capital
Fixed
Fuel
Variable

2013

2014

2015

2016

2017

2018

2019

31


-------
comes primarily from large decreases in capital and fuel expenditures. Capital expenditures have
decreased since 2013 in part due to the fact that capital costs of older nuclear and coal plants
(greater than 30 years) are being paid off.4 Meanwhile, the capital expenditures of new capacity
additions, such as solar and wind, are relatively low. Fuel expenditures have decreased both due to
the decline in coal and natural gas prices and to the decrease in fuel purchases from coal retirements.
Operation and maintenance costs have remained fairly static through time.

Figure 5 shows the annual expenditure by NERC region through time. Trends vary by region, with
noticeable decreases in expenditure in some regions (e.g., RFC and WECC) and increases in others
(e.g., MRO). Meanwhile, expenditures in SERC decline between 2013 and 2016, and then increase
after 2017. These distinct trends can be attributed to the different generation portfolios of each
region.

Figure 5: Annual total expenditure by region, 2013-2019

ASCC FRCC HICC MRO NPCC RFC SERC SPP IRE WECC

2.4.2 Updates Since MEEDE Version 1

Several key components have been added to this updated version of the MEEDE dataset that are
not available in the original 2013 version (Woollacott and Depro, 2016). First is the annual EIA
time series data. While MEEDE Version 1 covers the U.S. electricity sector landscape in 2013 only,

4In 2019, the average unweighted age of nuclear plants was 38.4 years, coal-fired plants 40.8 years, natural gas-fired
plants 25 years, oil-fired plants 32 years, and solar power plants 3.5 years.

32


-------
this version includes data from 2013 to 2019. In part because of this, the overall dataset increased
by over 80,000 observations.

Second, hourly-level data of NOx, SOx, and Hg are incorporated into MEEDE Version 2, where
MEEDE Version 1 used data from AP-42. In this updated version, the data are presented at annual
as well as 16 representative time-slice resolution to demonstrate the temporal (daily, seasonal, and
yearly) variability in emissions. This addition expanded the MEEDE dataset by 36 variables per
BID-PID pair per year. In terms of NOx and SOx total emissions in 2013, MEEDE Version 1
produced 1,735,868 and 3,292,037 tons, respectively. Meanwhile, MEEDE Version 2 produced
1,607,832 and 2,952,037 for each in 2013.

Though not an addition, a change in the input data in this version is the use of time series wholesale
price data from FERC-714. MEEDE Version 1 relied on the Intercontinental Exchange for wholesale
price data at the hub-level. In this version, we use FERC wholesale price data at the balancing
authority-level and generate annual and time-slice values in dollars and dollars per MWh. This
change in wholesale price data decreased the 2013 U.S. wholesale revenue from $194,206 million
in MEEDE Version 1 to $120,177 million in Version 2. In MEEDE Version 1, the U.S. average
annual wholesale price of electricity was $48/MWh in 2013, while in Version 2, it is $37/MWh in
2013. This difference is likely due to the inclusion of wholesale price data from regulated markets,
which may be lower than in deregulated markets. MEEDE Version 1 assumes the ICE deregulated
wholesale prices are representative of the national average, which may have inflated the wholesale
revenue estimates.

Another change is the use of NREL's ATB generation costs in MEEDE Version 2. In Version 1,
generation costs from the EIA's Electricity Market Module are used. As previously mentioned, we
elected to use ATB cost information due to its higher resolution of renewable technology costs over
fossil fuel technology costs. The use of the ATB created a final MEEDE dataset with fewer cost
categorizations for some technologies: for instance, generation cost for biomass generators is no
longer differentiated by bubbling fluidized-bed (biomass BFB) and combined cycle (biomass CC).
The change in data source produced similar capital costs (in $/MW) for some technologies such as
natural gas and wind. However, for others such as biomass fuels, geothermal, and solar PV, capital
costs differed more significantly between the two version of MEEDE.

Finally, this updated version of MEEDE uses the MATS RTR dataset for particulate matter emis-
sions, whereas Version 1 relies on AP-42 for this information. As previously mentioned, MATS
RTR only includes filterable PM 2.5, so the 2013 total particulate matter emissions in MEEDE
Version 2 will be smaller than that in MEEDE Version 1.

2.4.3 Comparison with Other Datasets

There are several other sources that provide similar plant-level data for the U.S. electricity sector.
These include the EPA's National Electric Energy Data System (NEEDS), S&P Global's SNL
Energy dataset, and ABB Energy Velocity Suite (formerly Ventyx). All are based on public plant-
level data from multiple sources. However, the S&P Global and ABB Energy Velocity datasets are
both subscription based and proprietary. As such, they are not transparent and their data builds
are easily reproducible. On the other hand, while the NEEDS dataset is open-source, it is not as
comprehensive as MEEDE in its representation of pollution controls or its treatment of boilers-to-
generators mapping. However, the NEEDS, S&P Global, ABB Energy Velocity datasets do provide

33


-------
some additional information that MEEDE does not include, such as financial fillings and data
from more local entities including system operators and public utility commissions. Given these
differences, MEEDE is more useful for environmental analyses requiring detailed representation of
power plant-level emissions, as well as studies looking for transparent data builds.

3 Application: SAM Integration

The MEEDE dataset provides a comprehensive, bottom-up depiction of the U.S. electricity sector
that can provide a foundation for descriptive information, time-series analysis, and/or simulation
modeling efforts. This section considers how these data can be used to inform economy-wide simu-
lations with detailed electricity sector information. We illustrate this point by providing a flexible
routine for integrating the data into a social accounting matrix (SAM) for economy-wide model-
ing. We find that the reference accounts require substantial changes to align with an aggregated
representation of the MEEDE dataset. This finding highlights the importance of incorporating
information from the bottom-up depiction of the U.S. electricity sector into CGE models.

In this section, we begin by describing the core accounting identities implicit in the construction of
a social accounting matrix. In this description, we illustrate the formulation of these identities both
in terms of absolute and intensive terms. Casting accounting identities using intensities (or shares)
alleviates numerical issues with large numbers and provides flexibility in isolating recalibrated scale
and intensity changes. We then formulate a mathematical program to incorporate information
from MEEDE and other data sources into the SAM. We do this by minimizing the change in the
data subject to constraints on accounting identities and targetted values (or priors) from MEEDE.
The objective function of the mathematical program can impact the optimal change in the data.
Therefore, we present two alternate assumptions common in the literature (e.g., Rutherford and
Schreiber (2019)).

3.1 SAM Identities

The disaggregation routine starts with a balanced SAM based on the national accounts. We con-
struct a SAM in three component matrices denominated in value flows: one, the value-added matrix
of the factors of production for each productive sector; two, the intermediate demand matrix of
the commodity demands by sector; and three, the final demand matrix of commodity demands
by households, government, investment, and trade partners. For present purposes, we assume no
inter-regional transfer payments. SAM balance is defined by two identities:

Row-Column Balance

^i^rji ^g^rjg	(1)

Income Bala,nee

^fjVrfj = ^gdrig Vr	(2)

where

34


-------
vrfj\ components of the value-add matrix indexed by region (r), factor (/ = {capital,

labor, taxes}), and sector (j).
xrij\ components of the intermediate demand matrix indexed by region (r), supply-
ing sector (i), and demanding sector (j).
drig\ components of final demand indexed by region (r), supplying sector (i), and
demand category (g = {C, I,G, X, M}).

From the SAM components, we define their intensive counterparts using value totals. Stating the
mathematical program in intensive measures compresses the range of values entering the mathe-
matical program, which can aid the numerical algorithm, normalizes the penalty for data revisions
for both small and large value flows, and allows us to modify and constrain the level and intensity
of economic activity separately.

We restate the SAM balance conditions in intensive variables by combining equations (1) and (2)
with (3) and (4). Constraining the mathematical program with (3) and (4) imposes their definitions
in intensive terms.

Row-Column Balance

Value Totals

(3a)
(3b)
(3c)
(3d)

Intensives

(4a)
(4b)
(4c)
(4d)
(4e)
(4f)

(5a)
(5b)

35


-------
Income Balance

^ / j^r fj^rj ^igdrig Dr g	(6a)

^ vr-fj Yr-j 	 y drjg Drg

Jf3 Yri Yr Zj»S Dra Yr

y. .'Hiii Xii — y. d"g Drg

^f] Y„; Y„ ~ »S D„„ Y„	v /

Totals

1 — SfVr fj -\- SiXrij

1 = TijYrj	(7b)

1 = YirYr	(7c)

1 = T,idrig	(7d)

Equations (5), (6), and (7) are sufficient for ensuring a balanced SAM, however, in the current
formulations we have multiple interrelated SAMs, one for each region, so we must also ensure that
domestic trade markets clear to fully balance the set of SAMs. To ensure domestic trade market
clearance we specify

0 S)"	S)"

A 	 V I ^rixd £_rx±_~y _ ^rimd £_r^nA.V	— 1

r ' "rxd Yr lr Drmd Yr lr) 1

0 S)" rfjv j jv d D y. jv d	I ^ ! X ^

Vi

(8a)

Vi

(8b)

Vi

(8c)

where xd denotes domestic exports and md domestic imports. The accounting identities (5), (6),
and (7) will ensure a balanced set of regional SAMs. Balanced SAMs are necessary but often
insufficient for desired model calibrations and we may elect to impose additional constraints on the
mathematical program to capture information beyond the internal consistency of the SAM.

3.2 Prior Formation

MEEDE provides an alternative characterization of the costs of generating electricity. While the
social accounting matrix contains a cost structure for the electricity sector, it is both typically an
aggregate representation (total generation, transmission, and distribution) and may systematically
differ from MEEDE given different assumptions made in its construction. Targetting values from
MEEDE will produce imbalances in the fundamental accounting identities oulined above and the
magnitude of those imbalances depend on our priors for what we believe to be most accurate for
the sector.

We hold the accounting identities for a balanced SAM with equality as we are certain of their
necessity for our modeling purposes. We may also hold information with uncertainty (i.e. priors
on plausible ranges of values) that we would like to incorporate in our SAM balancing. We can

36


-------
allow values to vary only minimally from a specified prior using a penalty function that measure
the deviation from our priors. Within the penalty function, we can also weight the relative penalty
assigned to each component of the SAM depending on the strength of our priors. Last, we can fix
or bound variables to control the extent of deviation from priors.

3.2.1	Macro Priors

Specification of prices and quantities, taxes and subsidies, transfer payments among domestic and
foreign agents, and specific endowments all inform how a model is calibrated to balanced SAM
data. These values may depend in part or whole on values in the SAM and we may hold priors on
whether and to what extent these values should be revised. For example, we might elect to hold
the national current account balance fixed, which would require further constraining the relative
value of TirDrxd and Y,rDrmd. For another example, we can hold household and/or government
expenditures fixed relative to GDP to limit adjustments in implied transfer payments between them.
In this illustrative exercise, however, we refrain from enforcing any macro priors to highlight the
importance of priors formed specifically in the electricity sector.

3.2.2	Micro Priors, Electricity

The first step in forming our micro priors is aggregating the MEEDE data to a chosen technology
and geographic resolution with only cost and revenue variables. The choice of resolution for the
MEEDE aggregation is bounded by the feasibility of a solution to the disaggregation problem.
Larger numbers of technologies and/or regions, particularly if combined with numerous constraints
on macro priors, may inhibit the ability of the numerical solver to identify a solution and/or produce
less sensible results.

The MEEDE data provide information on the costs of generating electricity whereas the social
accounts characterize the broader electricity generation, transmission, and distribution sector. I.e.,
economic inputs from the MEEDE data represent a subset of what is included in national accounts.
For this exercise, we rely on social accounts based on IMPLAN data for 2016 used in EPAs SAGE
model (Marten et al., 2021). Comparing this with MEEDE, the input values for the electricity
sector are mostly greater than their corresponding values in MEEDE, though there are instances
where MEEDE inputs are greater, particularly for fuels whose use is predominately by generation.

We first compare IMPLAN and MEEDE input totals for consistency. Naively, the difference between
the two datasets represents transmission and distribution (T&D) costs, but we have not formed
priors on what T&D inputs other than labor should be, other than positive.5 We hold capital,
energy, and materials priors for T&D weakly as they are calculated as differences between generation
totals and IMPLAN sector totals. For generation inputs based on MEEDE, we can be relatively
more confident in the quality of the targetted values. To reflect these differences in the strength
of our priors, we include a weighting parameter in our objective function that allows for penalizing
deviations for certain values more than others.

The cost input categories from MEEDE are more aggregated than those from national accounts.
MEEDE summarizes capital, fixed and variable O&M, and energy costs for generators. We apply

5 In instances where the MEEDE data provide a larger value than IMPLAN data we assert a nominally positive
T&D value (i.e. $lbn.)

37


-------
Table 15: Micro Priors Data Sources

1 Input Cost

Generation Value

T&D Value 1

Capital

MEEDE

IMPLAN - MEEDE

Labor

QCEW

QCEW

Energy

MEEDE

IMPLAN - MEEDE (ex. coal = 0)

Materials

VOM + FOM + Fuel Margins - QCEW

IMPLAN - MEEDE

MEEDE capital costs directly to generation and the difference with IMPLAN to T&D. For labor
costs, we draw on data from the Bureau of Labor Statistics (BLS) Quarterly Census of Employment
and Wages (QCEW) (Bureau of Labor Statistics, 2021). The QCEW data provide labor expendi-
tures by sector including eight technology types of electricity generation.6 We rely on these data to
form a prior for T&D and generation labor inputs, which we deduct from O&M costs for generation.
To assign the labor data to generation technologies, we map our target technology resolution to
that provided in the QCEW data. Where the target technology resolution exceeds QCEW (i.e. our
QCEW to target mapping is one-to-many), we distribute total QCEW labor using MEEDE's fixed
O&M costs for the target technologies.

MEEDE fuel costs are reported as delivered from EI A Form 923. We assign MEEDE fuel costs less
delivery margins to generation technologies and the margin value to the materials total. IMPLAN
data provide wholesale fuel costs and report delivery margin costs separately, leaving IMPLAN fuel
costs less wholesale generation fuel costs to form our T&D prior except for coal, which we assign
to zero for T&D. O&M costs also add to materials total. With labor costs deducted from MEEDE
O&M, the remaining costs sum to materials. Last, we distribute total materials by the materials
input pattern in the IMPLAN electricity sector.

3.3 Balancing Routine

We impose our priors on a 23 sector, 4 region (North, South, Mid, and West) SAM from Marten
et al. (2021). We disaggregate the electricity sector to eight generation technologies and one T&D
sector. The set of priors we impose on the SAM data will be inconsistent with the SAM balance
identities in Section 3.1. Imbalances range from —39% to —15% of output across sectors nationally
and range wider regionally.

To resolve the imbalances we introduced by asserting our priors, we pose a mathematical program to
minimally revise the unbalanced SAM, conserving the information we provide, subject to meeting
the accounting identities for SAM balance. We minimize the extent of revision using a penalty
function that measures the distance between the candidate SAM, §", and a balanced SAM solution,
§6. Table 16 states the mathematical program we implement to balance the SAM.

We implement the mathematical program in GAMS using the PATHNLP solver. The least squares
and Kullback-Leibler divergence functions, both solved as NLPs, provide comparable results. We
present the results of the SAM balancing on the electricity sector in Table 17. Revisions from our
priors are very close to the candidate values and identical in share terms as generation input and
output shares were fixed in the program. Economy-wide sectoral output changes ranged from —26%
to 4% nationally for the least squares metric and —26% to 6% for the KL Divergence.

''The technologies include hydroelectric, fossil fuels, nuclear, solar, wind, geothermal, biomass, and other.

38


-------
Table 16: Mathematical program for balancing an unbalanced SAM

Given:
Find:

§" = {<
§6 = {¦ i

,u ~u ~yu ju i\u yru\
fj' i>3' r.i' is' rg i 1 r J

fP

rj i ig i rg
z.b yb Jb f)b
i>3 "> rj i %g i rg

Yb |

' r J

Candidate intensive values
Corresponding solution values

Minimizing: E„iXi!/id(§" - Sb)2
SmdSuln(Sb/S

u\2

Sum of squared deviations
Kullback-Leibler Divergence

Subject to:

Equation 5
Equation 6
Equation 7
Equation 8

Row-column balance
Income balance
Totals

Trade balance

The balancing routine is flexibly programmed to admit different technology and regional aggre-
gations or to be run on different years. Additional macro-economic constraints or bounding of
variables could further limit large or important differences between candidate and solution values.
Such refinements are helpful way to address outliers such as the increases in the T&D sector where
we may carry some prior on "reasonable" solution levels.

Table 17: Electricity costs (2019$ Billion) prior and post balancing routine using the least-squares metric

GT&D

Data

Capital

Labor

Energy

Materials

Outpi

T&D

Prior

19.2

26.4

0

447.7

752.4



Solution

21.1

25.3

19

656.3

945.2

Coal

Prior

32.1

5.3

23.7

34.6

95.7



Solution

28.6

5.1

21.4

31.3

86.4

Gas

Prior

32.6

3

30.1

18.5

84.2



Solution

29.5

2.9

26.9

16.6

76

Hydro

Prior

2.6

1.8

0

5.5

9.9



Solution

2.6

1.6

0

5.7

9.9

Nuclear

Prior

18.6

2

0

13.8

34.5



Solution

17.7

2.2

0

13.2

33.1

Oil

Prior

2.4

0.4

1.3

1.8

6



Solution

2.1

0.5

1.3

1.7

5.5

Other

Prior

5.8

0.3

0

3.8

9.8



Solution

5.4

0.3

0

3.4

9

Solar

Prior

3.8

0.3

0

0.3

4.4



Solution

3.6

0.3

0

0.3

4.2

Wind

Prior

12.1

0.4

0

3.5

16



Solution

11.6

0.4

0

3.4

15.3

39


-------
4 Conclusion

In this report, we introduced MEEDE Version 2 and the process by which the dataset was con-
structed. MEEDE relies on a combination of publicly-available data sources from the EPA, EIA,
FERC, and NREL. We illustrate MEEDE's engineering, economic, and environmental outputs for
the most recent year of data, 2019. We discuss some of the advantages and limitations of MEEDE
and its underlying data. The primary advantages of MEEDE include its transparency, public
availability, and comprehensive unit-level characterization of utility-scale generation in the United
States. We demonstrate how MEEDE data (or other bottom-up sources) can be successfully in-
tegrated into national accounts for use in economy-wide modeling by way of a matrix balancing
routine with differing objective functions. While not described here, the dataset has also been used
in partial equilibrium modeling of the electricity sector. For access to the MEEDE dataset, please
contact the corresponding author.

40


-------
References

U.S. Energy Information Administration. EIA Form 860. https://www.eia.gov/electricity/
data/eia860/, 2021a. Online, accessed 18 May 2021.

U.S. Energy Information Administration. EIA Form 923. https://www.eia.gov/electricity/
data/eia923/, 2021b. Online, accessed 18 May 2021.

National Renewable Energy Laboratory. Annual Technology Baseline 2020. https://atb.nrel.
gov/, 2021. Online, accessed 18 May 2021. As of 15 Feb 2023, data is archived and accessible at:
https://data.nrel.gov/submissions/145.

U.S. Environmental Protection Agency. Documentation for EPA's Power Sector Modeling Platform
v6: Using the Integrated Planning Model, https://www.epa.gov/sites/production/files/
2018-06/documents/epa_platform_v6_documentation_-_all_chapters_june_7_2018.pdf,

2018.

U.S. Energy Information Administration. Electricity generation, capacity, and sales
in the United States.	https://www.eia.gov/energyexplained/electricity/

electricity-in-the-us-generation-capacity-and-sales . php, 2021c. Online, accessed 18
May 2021.

U.S. Census Bureau. North American Industry Classification System (NAICS). https://www.
census.gov/naics/, 2021. Online, accessed 18 May 2021.

U.S. Energy Information Administration. Monthly Energy Review Notes: April 2021. Technical
report, U.S. Energy Information Administration, 2021d.

U.S. Energy Information Administration. Capital Costs and Performance Characteristics for Utility
Scale Power Generating Technologies. Technical Report SL-014940, U.S. Energy Information
Administration, 2020.

Federal Energy Regulatory Commission. FERC FINANCIAL REPORT FERC FORM No. 1: An-
nual Report of Major Electric Utilities, Licensees and Others and Supplemental Form 3-Q: Quar-
terly Financial Report, https://www.ferc.gov/sites/default/files/2020-04/form-l.pdf,
2023. Online, accessed 15 Feb 2023.

Catalyst Cooperative. PUDL Database: FERC Form 1. https: //catalystcoop-pudl.
readthedocs.io/en/latest/data_sources/fercl.html, 2023. Online, accessed 15 Feb 2023.

U.S. Energy Information Administration. The National Energy Modeling System: An Overview
2018, 2019a.

Peter Cappers and Sean Murphy. Unpacking the Disconnect Between Wholesale and Retail Electric
Rates, https://emp.lbl.gov/publications/unpacking-disconnect-between, 2019.

Federal Energy Regulatory Commission. FERC Form 714: Annual Electric Balancing Authority
Area and Planning Area Report. https://www.ferc.gov/industries-data/electric/
general-information/electric-industry-forms/form-no-714-annual-electric/data,
2021.

41


-------
U.S. Environmental Protection Agency. Sources of Greenhouse Gas Emissions: Overview, https:
//www.epa.gov/ghgemissions/sources-greenhouse-gas-emissions, 2021a.

U.S. Environmental Protection Agency. GHGRP Electrical Equipment Production and Use.
https://www.epa.gov/ghgreporting/ghgrp-electrical-equipment-production-and-use#
2019-subsector, 2021b.

U.S. Environmental Protection Agency. Screening Methodologies to Support Risk and Technology
Reviews (RTR): A Case Study Analysis, 2017.

U.S. Environmental Protection Agency. National Emission Standards for Hazardous Air Pollutants
From Coal and Oil-Fired Electric Utility Steam Generating Units and Standards of Performance
for Fossil-Fuel-Fired Electric Utility, Industrial-Commercial-Institutional, and Small Industrial-
Commercial-Institutional Steam Generating Units, https://www.govinfo.gov/content/pkg/
FR-2011-05-03/pdf/2011-7237.pdf, 2011.

U.S. Environmental Protection Agency. Compilation of Air Pollutant Emission Factors. Technical
report, U.S. Environmental Protection Agency, 1995.

U.S. Environmental Protection Agency. Reconsideration of the National Emission Standards for
Hazardous Air Pollutants (NESHAP) Maximum Achievable Control Technology (MACT) Floor
Analysis for Coal- and Oil-fired Electric Utility Steam Generating Units, Proposed Rule, 2012.

U.S. Environmental Protection Agency. Air markets program data, https: //ampd. epa. gov/ampd/,
2021c. Online, accessed 18 May 2021.

Stuart Cohen, Jon Becker, Dave Bielen, Maxwell Brown, Wesley Cole, Kelly Eurek, Will Ffazier,
Bethany Frew, Pieter Gagnon, Jonathan Ho, Paige Jadun, Trieu Mai, Matthew Mowers, Caitlin
Murphy, Andrew Reimers, James Richards, Nicole Ryan, Evangelia Spyrou, Daniel Steinberg, Yi-
nong Sun, Nina Vincent, and Matthew Zwerling. Regional Energy Deployment System (ReEDS)
Model Documentation: Version 2018. Technical Report NREL/TP-6A20- 72023, National Re-
newable Energy Laboratory, 2019.

U.S. Environmental Protection Agency. Greenhouse Gas Reporting Program Database, https:
//enviro.epa.gov/, 2021d. Online, accessed 18 May 2021.

U.S. Environmental Protection Agency. Title 40: Protection of Environment, Part 98:
Mandatory Greenhouse Gas Reporting. https://www.ecfr.gov/cgi-bin/text-idx?SID=
bdd8524bdce0787d01ea0cl307e5el26&mc=true&node=sp40.23.98.a&rgn=div6, 2009.

U.S. Environmental Protection Agency. Power Plant Crosswalk, https://www.epa.gov/sites/
production/files/2020-12/ghgrp_oris_power_plant_crosswalk_ll_24_20.xlsx, 2020.

U.S. Energy Information Administration. State Energy Data 2019: Technical Notes, https:
//www.eia.gov/state/seds/sep_fuel/notes/use_intro.pdf, 2019b.

Jared Woollacott and Brooks Depro. Construction and Application of the MEEDE Dataset. Tech-
nical report, RTI Press, 2016.

Thomas F Rutherford and Andrew Schreiber. Tools for Open Source, Subnational CGE Modeling
with an Illustrative Analysis of Carbon Leakage. Journal of Global Economic Analysis, 4(2),
2019.

42


-------
Alex Marten, Andrew Schreiber, and Ann Wolverton. SAGE Model Documentation (2.0.1).
U.S. Environmental Protection Agency: https://www.epa.gov/environmental-economics/
cge-modeling-regulatory-analysis, 2021.

Bureau of Labor Statistics. Quarterly Census of Employment and Wages, https://www.bls.gov/
cew/, 2021. Online, accessed 20 May 2021.

43


-------
A Appendix I: MEEDE Data Dictionary

Variable

Label

Aggregation
Level

ash_cont

[BLR]: Pet by wt. ash content [EIA.923J>3]

Boiler

baLauth

[PLNT]: Plant Balancing Authority [EIA.860-Plant]

Plant

BID

[BLR]: Boiler Identifier [EIA.923.P3]

Boiler

blr_btm

[BLR]: Type of boiler bottom (Wet or Dry) [EIA.860J5qp]

Boiler

blr_btu

[BLR]: Qty of fuel consumed in MMBTU [EIA.923.P3]

Boiler

blr_btu_ele

[BLR]: MMBTU cons, by boiler for ele gen [EIA.923.P1+P3]

Boiler

blr_btu_tot

[BLR]: MMBTU total for all boilers @PID-PM-FT [EIA.923.P3]

Boiler

blr_ct

[BLR]: No. boilers @PID-PM-FT [EIA.923J3]

Boiler

blr_fire_typ

[BLR]: Type of firing of boiler [EIA.860J5qp]

Boiler

blr_na

[BLR]: No data available @PID-PM-FT-BID [EIA.923.P3]

Boiler

blr_netgen

[BLR]: Net gen MWh as apportioned to boiler [EIA.923.P1-P3]

Boiler

blr_netgen_stup

[BLR]: Net generation (MWh) attributable to startup fuel by BTUs
[EIA.923]

Boiler

blr_netgen_tot

[BLR]: Net generation (MWh) total = startup + regular [EIA.923]

Boiler

blr_newsrc

[BLR]: Subject to new source review standards [EIA.860.Eqp]

Boiler

blr_noact

[BLR]: No activity data, available [EIA.923^3]

Boiler

blr_noattr

[BLR]: No attribute data available [EIA.860]

Boiler

blr_qty

[BLR]: Qty of fuel consumed in phys. units [EIA.923.P3]

Boiler

blr_qty_ele

[BLR]: Fuel phys unit cons, by boiler for ele gen [EIA.923.P1+P3]

Boiler

blr_svc_yr

[BLR]: Service year of boiler [EIA.860.Eqp]

Boiler

CAMD.GLOAD.

[PLNT]: Total GLOAD (MWh) of C02 [CAMD]

Plant

sum CO'2





CAMD.GLOAD.
sum _N OX

[PLNT]: Total GLOAD (MWh) of NOX [CAMD]

Plant

CAMD.GLOAD.

[PLNT]: Total GLOAD (MWh) of SOX [CAMD]

Plant

sum_SOX





CAMDJieatinput.
sum CO'2

[PLNT]: Heat input (mniBtu) of C02 [CAMD]

Plant

CAMDJieatinput.
sum _N OX

[PLNT]: Heat input (mniBtu) of NOX [CAMD]

Plant

CAMDJieatinput.
sum_SOX

[PLNT]: Heat input (mniBtu) of SOX [CAMD]

Plant

CAMD.tons.

[BLR]: Median CAMD emissions of C02 in tons [CAMD-AP42]

Boiler

med_C02





CAMD.tons.

[BLR]: Median CAMD emissions of NOX in tons [CAMD-AP42]

Boiler

medJ^OX





CAMD.tons.

[BLR]: Median CAMD emissions of SOX in tons [CAMD-AP42]

Boiler

mecLSOX





CAMD.tons.

[BLR]: 25th percentile CAMD emissions of C02 in tons [CAMD-AP42]

Boiler

p25_C02





CAMD.tons.

[BLR]: 25th percentile CAMD emissions of NOX in tons [CAMD-AP42]

Boiler

p25_NOX





CAMD.tons.

[BLR]: 25th percentile CAMD emissions of SOX in tons [CAMD-AP42]

Boiler

p25_SOX





44


-------
CAMD_tons_

[BLR]: 75th percentile CAMD emissions of C02 in tons [CAMD-AP42]

Boiler

p75_C02





CAMD_tons_

[BLR]: 75th percentile CAMD emissions of NOX in tons [CAMD-AP42]

Boiler

p75_NOX





CAMD_tons_

[BLR]: 75th percentile CAMD emissions of SOX in tons [CAMD-AP42]

Boiler

p75_SOX





CAMD_tons_

[BLR]: Total CAMD emissions of C02 in tons [CAMD-AP42]

Boiler

sum CQ2





CAMD_tons_

[BLR]: Total CAMD emissions of NOX in tons [CAMD-AP42]

Boiler

sum _N OX





CAMD_tons_

[BLR]: Total CAMD emissions of SOX in tons [CAMD-AP42]

Boiler

sum_SOX





census

[PLNT]: Census region code [923.PI]

Plant

CHP

[PLNT]: Combined heat & power plant [EIA.923_P1]

Plant

county

[PLNT]: Plant county location [EIA.860-Plant]

Plant

ctrl_typ_HG

[BLR]: Type of HG ctrl. from IPM cost data [IPM_Ch5]

Boiler

ctrl_typ_NOX

[BLR]: Type of NOx ctrl. from IPM cost data [IPM_Ch5]

Boiler

ctrl_typ_PM

[BLR]: Type of PM ctrl. from IPM cost data [IPM_Ch5]

Boiler

ctrl_typ_SOX

[BLR]: Type of SOx ctrl. from IPM cost data [IPM_Ch5]

Boiler

emissions_typ

Type of emissions corresponding to this row of data



FC

[BLR]: Fuel Code-Map based on EIA Fuel Types

Boiler

FC_st

[BLR]: Fuel Code-Map based on EIA Fuel Types of start-ups

Boiler

FERC_annual

[PLNT]: Total wholesale ELE price in $/MWh [FERC.714]

Plant

FLU_height

[BLR]: Wtd. avg. height of flue(s) serving boiler in ft [EIA.860_Equip-Flue]

Boiler

FOM_gen

[BLR]: Total boiler-level fixed O&M cost of generation ($)

Boiler

FOM_HG

[BLR]: Total boiler-level fixed O&M cost of HG ctrl ($)

Boiler

FOM_NOX

[BLR]: Total boiler-level fixed O&M cost of generation ($)

Boiler

FOM_PM

[BLR]: Total boiler-level fixed O&M cost of PM ctrl ($)

Boiler

FOM_SOX

[BLR]: Total boiler-level fixed O&M cost of SOX ctrl ($)

Boiler

FOM_tot

[BLR]: Total boiler-level fixed O&M cost of generation & control ($)

Boiler

FT

[BLR]: Fuel Type [EIA.923_P1-P3]

Boiler

FT_st

[BLR]: Fuel Type of start-up fuel

Boiler

FUE_cost

[BLR]: Boiler-level fuel cost of generation (ex. startup) ($)

Boiler

FUE_cost_st

[BLR]: Total boiler-level startup fuel cost of generation ($)

Boiler

FUE_tot

[BLR]: Total boiler-level fuel cost of generation & startup ($)

Boiler

fueLpxbtu

[BLR]: Fuel price ($/MMbtu) [EIA.923_P5]

Boiler

fuel_pxbtu_st

[BLR]: Fuel price of startup fuel ($/MMbtu) [EIA.923_P5]

Boiler

GEN_cst

[BLR]: Total boiler-level cost of generation & control ($)

Boiler

GEN_rev_annual

[PLNT]: Total revenue ($)

Plant

gen_svc_yr

[GEN]: NP-weighted average service year @PID-PM [EIA.860_Gen]

Generator

GHG_tons_

[BLR]: Median GHGRP emissions CH4 in tons [GHGRP-AP42]

Boiler

med_CH4





GHG_tons_

[BLR]: Median GHGRP emissions C02 in tons [GHGRP-AP42]

Boiler

med_C02





GHG_tons_

[BLR]: Median GHGRP emissions N20 in tons [GHGRP-AP42]

Boiler

medJ^20





GHG_tons_

[BLR]: 25th percentile GHGRP emissions CH4 in tons [GHGRP-AP42]

Boiler

p25_CH4





45


-------
GHG_tons_

[BLR]: 25th percentile GHGRP emissions C02 in tons [GHGRP-AP42]

Boiler

p25_C02





GHG_tons_

[BLR]: 25th percentile GHGRP emissions N20 in tons [GHGRP-AP42]

Boiler

p25_N20





GHG_tons_

[BLR]: 75th percentile GHGRP emissions CH4 in tons [GHGRP-AP42]

Boiler

p75_CH4





GHG_tons_

[BLR]: 75th percentile GHGRP emissions C02 in tons [GHGRP-AP42]

Boiler

p75_C02





GHG_tons_

[BLR]: 75th percentile GHGRP emissions N20 in tons [GHGRP-AP42]

Boiler

p75_N20





GHG_tons_

[BLR]: Total GHGRP emissions CH4 in tons [GHGRP-AP42]

Boiler

sum_CH4





GHG_tons_

[BLR]: Total GHGRP emissions C02 in tons [GHGRP-AP42]

Boiler

sum CQ2





GHG_tons_

[BLR]: Total GHGRP emissions N20 in tons [GHGRP-AP42]

Boiler

sum _N 20





heat_rt

[GEN]: Heat rate (MMBtu / MW) @PID-PT [EIA.923]

Generator

heat_rt_cat

[PLNT]: Heat rate category for cost data [EIA.923]

Plant

HG_erf

[BLR]: Emissions ctrl reduction factor for HG [AP42]

Boiler

HG_tons_med_HG

[BLR]: Median MATS-HG emissions in tons [MATS-HG-AP42]

Boiler

HG_tons_p25_HG

[BLR]: 25th percentile MATS-HG emissions in tons [MATS-HG-AP42]

Boiler

HG_tons_p75_HG

[BLR]: 75th percentile MATS-HG emissions in tons [MATS-HG-AP42]

Boiler

HG_tons_sum_HG

[BLR]: Total MATS-HG emissions in tons [MATS-HG-AP42]

Boiler

K_gen

[BLR]: Total boiler-level capital cost of generation ($)

Boiler

K_HG

[BLR]: Total boiler-level capital cost of HG ctrl ($)

Boiler

K_NOX

[BLR]: Total boiler-level capital cost of NOX ctrl ($)

Boiler

K_PM

[BLR]: Total boiler-level capital cost of PM ctrl ($)

Boiler

K_SOX

[BLR]: Total boiler-level capital cost of SOX ctrl ($)

Boiler

K_tot

[BLR]: Total boiler-level captial cost of generation & control ($)

Boiler

lat

[PLNT]: Plant latitude [EIA.860-Plant]

Plant

long

[PLNT]: Plant longitude [EIA.860-Plant]

Plant

NERC

[PLNT]: Plant NERC region code [EIA.860-Plant]

Plant

netgen

[PLNT]: Net gen (MWh) @PID-PM-FT [EIA.923_P1]

Plant

NOX_erf

[BLR]: Emissions ctrl reduction factor for NOx [AP42]

Boiler

NP

[GEN]: Nameplate capacity (MW) [EIA.860_Gen]

Generator

NP_cat

[PLNT]: Nameplate category for cost data [EIA.923]

Plant

NP_plant

[PLNT]: Nameplate capacity of entire plant (MW) [EIA.860_Gen]

Plant

phys_unit_FT

[BLR]: Physical Unit of fuel [EIA.923_P3]

Boiler

PID

[PLNT]: Plant Identifier [EIA.923_P3]

Plant

plant_nm

[PLNT]: Plant name[EIA.860_Plant]

Plant

plnt_noact

[PLNT]: No data available @PID-PM-FT [EIA.923_P1]

Plant

plnt_svc_yr

[GEN]: NP-weighted average service year @PID [EIA.860_Gen]

Generator

PM

[GEN]: Prime Mover [EIA923_P1-P3]

Generator

PM_erf

[BLR]: Emissions ctrl reduction factor for PM [AP42]

Boiler

PM_tons_sum_PM

[BLR]: Total MATS-PM emissions in tons [MATS-PM-AP42]

Boiler

PT

[PLNT]: Generation type categorized to ATB cost data

Plant

PT_btu_ele

[GEN]: Net fuel consumption for generation (MMBtus) @PID-PT [EIA.923]

Generator

PT_netgen

[GEN]: Net generation (MWh) @PID-PT [EIA.923]

Generator

46


-------
PT_NP

[PT]: Nameplate capacity (MW) @PID-PT

Plant Type

reg_stat

[PLNT]: Plant regulatory status [EIA.860-Plant]

Plant

RMV_eff_NOX

[GEN]: Control percent removal efficiency [IPM_Ch5]

Generator

RMV.efLSOX

[GEN]: Controf percent removaf efficiency [IPM_Ch5]

Generator

SOX_erf

[BLR]: Emissions ctrl reduction factor for SOx [AP42]

Boiler

state

[PLNT]: State where plant is located

Plant

sulf_cont

[BLR]: Pet by wt. S02 content [EIA.923_P3]

Boiler

tot_btu

[PLNT]: Total MMBTU cons. @PID-PM-FT [EIA.923_P1]

Plant

tot_btu_ele

[PLNT]: ELE gen MMBTU cons. @PID-PM-FT [EIA.923_P1]

Plant

tot_qty_ele

[PLNT]: ELE gen phys unit cons. @PID-PM-FT [EIA.923_P1]

Plant

VOM_gen

[BLR]: Total boiler-level var. O&M cost of generation ($)

Boiler

VOM_HG

[BLR]: Total boiler-level var. O&M cost of HG ctrl ($)

Boiler

VOM.NOX

[BLR]: Total boiler-level var. O&M cost of NOX ctrl ($)

Boiler

VOM_PM

[BLR]: Total boiler-level var. O&M cost of PM ctrl ($)

Boiler

VOM_SOX

[BLR]: Total boiler-level var. O&M cost of SOX ctrl ($)

Boiler

VOM_tot

[BLR]: Total boiler-level var. O&M cost of generation & control ($)

Boiler

year

Year [EIA.923_P5]

Dataset

zip

[PLNT]: Plant zipcode [EIA.860-Plant]

Plant

47


-------
B MEEDE Version 2 Outputs for 2016

Table Al: Annual net electricity generation from EIA Form 923 by prime mover and NERC region, 2016
(in millions of MWh)



Annual Net Generation
(Millions of MWh)

I Code

Description

SERC

RFC

WECC

TRE

MRO

SPP

NPCC

FRCC

HICC

ASCC

Total USA

ST

Steam Turbine

701

668

260

152

148

119

104

84

6

1

2,244

CT

Combined Cycle (Combustion)

204

123

119

104

9

36

51

89

2

2

738

HY

Hydraulic Turbine

32

9

174

1

12

6

33

0

0

2

268

CA

Combined Cycle (Steam)

108

64

63

47

5

21

26

48

1

0

383

WT

Wind (Onshore)

4

24

51

53

45

42

7

0

1

0

227

GT

Gas Turbine

42

27

22

15

3

6

6

7

0

1

130

PV

Photovoltaic

4

1

25

1

0

0

1

0

0

0

33

CS

Combined Cycle (Single)

1

6

9

10

0

1

14

1

0

0

41

BT

Binary Cycle Turbines

0

0

4

0

0

0

0

0

0

0

4

IC

Internal Combustion (Diesel)

2

3

3

1

1

1

1

0

0

1

14

CP

Concentrated Solar (Storage)

0

0

1

0

0

0

0

0

0

0

1

OT

Other

0

0

0

0

0

0

0

0

0

0

1

FC

Fuel Cell

0

0

0

0

0

0

0

0

0

0

1

PS

Pumped Storage

-4

-2

0

0

0

0

-1

0

0

0

-7

FW

Energy Storage (Flywheel)

0

0

0

0

0

0

0

0

0

0

0

WS

Wind (Offshore)

0

0

0

0

0

0

0

0

0

0

0

BA

Battery

0

0

0

0

0

0

0

0

0

0

0

CE

Compressed Air (Storage)

0

0

0

0

0

0

0

0

0

0

0

Total



1,095

924

731

384

222

231

242

229

10

6

4,076

Regions: ASCC = Alaska Systems Coordinating Council; HICC = Hawaiian Islands Coordinating Council; MRO
s= Midwest Reliability Organization; NPCC = Northeast Power Coordinating Council; RFC 3= ReliabilityFirst Cor-
poration; SERC •== Southeastern Electric Reliability Council; TRE = Texas Reliability Entity; WECC =t Western
Electricity Coordinating Council.

Note: The values shown here may not sum to the totals shown due to independent rounding.

48


-------
Table A2: Meta-summary of EI A Form 923 data, 2016

Form

Schedule



Page

Observation

Variables

Number
of Plants

Number of
Observations

923

2_3_4_5_

M_

.12

Page 1: Generation
and Fuel Data

{PID, PM, FT}

< Fuel, Generation >

8,120

12,787

923

2_3_4_5_

M.

12

Page 3: Boiler Fuel
Data

{PID, PM, FT, BID}

< Fuel, Sulfur Content, Ash Content >

1,543

9,925

923

2_3_4_5_

M.

12

Page 5: Fuel
Receipts and Costs

{PID}

< Monthly Fuel Prices >

1,046

30,751

860

3.1





Operable

{PID, GID, PM, FT}

< Nameplate Capacity, Operating Year,
Planned Retirement Year >

8,084

20,724

860

2





Plant

{PID}

< Plant Attributes >

9,711

9,711

860

6.1





Boiler-Generator

{PID, GID, BID}

< PID, GID, BID >

1,698

7,426

860

6.2





Boiler Info 8t
Design Parameters

{PID, BID}

< PID, BID, In-service Year, Firing Type,
Wet/Dry Bottom Tech >

1,700

4,732

860

6.1





Boiler Particulate
Matter

{ PID, BID, FGPJD }

< PID, ID, FGPJD >

868

2,476

860

6.1





Boiler S02

{ PID, BID, FGDJD}

< PID, BID, FGDJD >

445

991

860

6.1





Boiler NOX

{PID, BID, NOXJD }

< PID, BID, NOXJD >

773

1,671

860

6.1





Boiler Mercury

{PID, BID, HGJD}

< PID, BID, HGJD >

500

1,469

860

6.1





Emissions Control
Equipment

{ PID, FGPJD, FGDJD, NOXJD, HGJD,
Equipment Type, Controls }

< Control ID (PM, SOX, NOX, HG),
Equipment Type >

1,204

5,830

860

6.2





Emission
Standards 8*
Strategies

{ PID, BID}

< PID, BID, Control Strategies: SOX,
NOX, HG >

1,700

4,732

860

6.1





Boiler Stack Flue

{PID, BID, FID}

< PID, BID, FID >

873

2,936

860

6.2





Stack Flue

{PID, FID}

< PID, FID, Service Year, Height,
Volume, Status >

876

2,382

Abbreviations: BID = boiler ID; FGD_ID = flue gas desulfurization equipment ID; FGP_ID = flue gas particulate
collector ID; FT • fuel type; GID = generator ID; HG_ID = mercury equipment ID; NOX_ID = nitrogen oxide
equipment ID; PID = plant ID; PM §= prime mover; SOX_ID =t sulfur oxide equipment ID.

49


-------
Table A3: U.S. electric grid fuel heat, sulfur, and ash content, 2016





Fuel Code

Pollutant Content (%)



Volume/Mass



Fuel Code

Description

Sulfur

Ash

BTU (Quadrillion)

Quantity

Units

Coal

BIT

Bituminous coal

2.29

10.79

5.01

225.51

MM Sh. Ton

Coal

SUB

Subbituminous coal

0.35

5.95

4.85

285.46

MM Sh. Ton

Coal

COL

Other coal

1.11
0.00

11.64
0.00

3.12

183.49
109.80

MM Sh.Ton
Bn. Cu. Ft.

Gas

GAS

Natural gas & propane

0.00

0.00

10.51

9,825.88

Bn. Cu. Ft.

Gas

OGS

Other gases

0.00

0.00

0.26

1,162.26

Bn. Cu. Ft.

Oil

OIL

Oil & petroleum deriv.

0.58
5.24

0.00
0.74

0.25

20.06
17.21

MM Barrels
MM Sh.Ton

Other

BML

Biomass liquids

0.00

0.00

0.00

4.46

MM Barrels

Other

BMS

Biomass solids

0.00

0.00

0.38

124.83

MM Sh.Ton

Other

MSW

Municipal solid wastes

0.00

0.00

0.28

30.85

MM Sh.Ton

Other

OTH

Miscellaneous other

0.00

0.00

0.02

9.35

MM Sh.Ton

50


-------
Table A4: Percent of net MWh fossil fuel generation with environmental controls per region, 2016 (%)

Control Equipment Type

Control
ID

ASCC

FRCC

HICC

MRO

NPCC

RFC

SERC

SPP

TRE

WECC

Total
USA

Sulfur Control Technologies



5.4

19.3

18.1

71.8

3.2

54.6

37.7

36.3

26.8

41.8

39.7

Circulating dry scrubber

CD

0.0

0.3

0.0

7.8

0.5

1.7

0.1

3.8

1.5

0.9

1.4

Dry sorbent (powder) injection type (DSI)

DSI

5.4

2.2

17.9

7.1

0.0

7.4

8.9

3.6

0.9

0.9

5.2

Jet bubbling reactor (wet) scrubber

JB

0.0

0.0

0.0

0.0

0.0

6.0

4.1

0.0

0.0

0.0

2.5

Mechanically aided type (wet) scrubber

MA

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.1

0.0

Packed type (wet) scrubber

PA

0.0

0.0

0.0

0.0

0.0

0.3

0.0

0.0

0.0

0.0

0.1

Spray dryer type / dry FGD / semi-dry FGD

SD

0.0

2.0

0.0

37.6

1.1

2.2

3.8

8.3

1.5

11.4

6.0

Spray type (wet) scrubber

SP

1.0

9.9

0.0

19.3

1.5

35.3

25.4

22.6

21.9

19.4

23.6

Tray type (wet) scrubber

TR

0.0

7.0

0.0

0.0

0.0

10.4

2.0

0.0

1.9

5.9

4.5

Venturi type (wet) scrubber

VE

0.0

0.0

0.1

5.1

0.1

1.1

1.0

2.6

0.0

3.4

1.5

Particulate Control Technologies



16.4

21.4

18.0

80.4

7.4

59.2

48.3

53.9

35.4

39.4

46.2

Baghouse (fabric filter), pulse

BP

12.1

2.0

0.0

40.2

2.0

8.5

13.6

25.6

10.5

13.8

12.9

Baghouse (fabric filter), reverse air

BR

4.3

0.3

18.0

15.9

0.0

1.1

2.8

4.2

8.5

11.5

4.8

Baghouse (fabric filter), shake and deflate

BS

0.0

0.0

0.0

1.8

0.0

0.1

0.0

3.7

0.9

0.1

0.5

Electrostatic precipitator, cold side, with flue gas
conditioning

EC

0.0

3.7

0.0

14.3

0.1

12.2

10.3

3.4

8.7

2.7

8.2

Electrostatic precipitator, hot side, with flue gas
conditioning

EH

0.0

0.1

0.0

4.4

0.4

0.5

0.5

1.1

0.0

0.0

0.6

Electrostatic precipitator, cold side, without flue gas
conditioning

EK

0.0

14.5

0.0

21.9

6.1

38.0

27.7

18.3

10.4

5.2

21.9

Electrostatic precipitator, hot side, without flue gas
conditioning

EW

0.0

0.4

0.0

5.4

0.6

3.0

4.4

3.5

0.0

7.6

3.6

Multiple cyclone

MC

6.3

0.6

0.0

0.8

0.4

0.1

1.9

2.8

0.0

0.7

0.9

Single cyclone

SC

0.0

0.0

0.0

0.0

0.0

0.1

0.0

0.0

0.0

0.0

0.0

Nitrogen Control Technologies



9.5

27.7

18.0

86.5

31.5

69.0

63.3

68.7

51.3

55.9

59.6

Advanced overfire air*

AA

0.0

0.0

0.0

22.3

1.5

0.4

1.1

3.1

3.4

11.5

3.9

Biased firing*

BF

0.0

0.0

0.0

0.0

0.0

0.4

0.0

0.0

0.1

0.0

0.1

Fluidized bed combustor*

CF

0.0

4.3

17.9

1.8

0.1

1.6

1.3

4.1

2.6

0.4

1.8

Flue recirculation*

FR

0.0

0.1

0.0

0.1

1.0

0.5

1.3

1.5

0.7

0.5

0.8

Fuel reburning*

FU

0.0

0.6

0.0

0.0

0.3

0.0

0.5

0.0

0.0

0.1

0.2

Water injection*

H20

4.0

0.0

0.0

0.2

2.5

0.3

0.5

0.1

0.2

0.1

0.4

Low excess air*

LA

1.0

0.0

0.0

0.6

1.8

0.1

1.8

1.2

1.7

0.7

1.0

Low NOx burner*

LN

4.3

16.4

0.0

70.5

14.8

42.3

36.4

52.8

39.7

41.7

39.1

Ammonia injection*

NH3

0.0

0.0

0.0

1.1

3.7

6.0

2.1

5.3

5.7

8.0

4.3

Overfire air*

OV

4.3

9.5

0.0

52.8

9.9

24.6

23.1

48.3

21.5

19.1

24.1

Selective noncatalytic reduction

SN

0.0

2.5

18.0

13.7

0.5

5.9

4.4

9.9

3.4

1.5

4.7

Selective catalytic reduction

SR

0.2

21.0

0.0

21.2

19.0

53.8

44.8

26.2

22.4

20.4

35.4

Steam injection*

STM

0.0

0.0

0.0

0.0

0.3

0.2

0.2

0.0

0.3

0.3

0.2

Mercury Control Technologies



0.0

1.6

0.0

69.3

2.0

14.5

27.6

48.6

34.8

29.7

26.0

Activated carbon injection system

ACI

0.0

1.6

0.0

69.3

1.9

14.2

26.7

48.6

34.8

29.7

25.7

Lime injection*

UJ

0.0

0.0

0.0

0.0

0.1

0.6

1.7

6.8

0.0

0.0

1.1

Other

OT

0.0

0.0

0.0

11.0

0.0

1.6

2.5

0.1

0.6

8.0

2.9

Total fossil fuel generation (Millions of MWh)



3.6

194.3

8.4

134.9

117.1

608.8

750.1

172.4

286.4

395.0

2,671.0

Note: Change-in-process technologies are marked with an asterisk.


-------
Table A5: Costs of generation and pollution control, 2016

Installation Attribute

Total Costs Excluding Fuel (MM 2019$)

Cost Component

Net Generation (in thousands of MWh)

Capital

Fixed 08cM

Variable 08cM

Total

Total system, excluding fuel

4,075,533

90,651

57,733

17,343

165,727

Generation only

4,075,533

85,382

54,082

9,920

149,384

Controls

1,648,818

5,270

3,650

7,423

16,343

SOX Controls

1,086,353

2,378

2,765

4,145

9,288

Limestone forced oxidation scrubber

810,689

1,159

2,073

3,036

6,268

Lime spray dryer scrubber

274,000

1,490

899

1,087

3,477

Dry sorbent injection

141,601

403

357

1,296

2,056

PM Controls

1,282,767

1,548

426

102

2,076

Electrostatic precipitator (cold side)

817,015

463

273

66

802

Fabric filter baghouse

501,983

1,193

192

47

1,432

Electrostatic precipitator (cold side), with flue gas conditioning

221,310

139

72

18

228

Electrostatic precipitator (hot side)

129,977

273

59

13

345

Electrostatic precipitator (hot side), with flue gas

18,137

84

15

2

102

NOX Controls

1,593,788

1,301

436

1,385

3,121

Selective catalytic reduction

948,573

1,171

318

1,185

2,675

Low-NOx burner (wall fired)

732,258

945

263

841

2,048

Low-NOx burner with advanced overfire air (tangentially fired)

371,190

44

56

8

109

Low-NOx burner with overfire air (wall fired)

282,972

585

135

453

1,173

Selective noncatalytic reduction (fluidized bed)

28,083

26

4

81

111

Vertically fired

4,360

15

1

7

23

HG Control - Active Carbon Injection

707,979

43

23

1,792

1,858

Abbreviations: Hg = mercury; MWh = megawatt-hour; NOX = nitrogen oxide; O&M = operations and maintenance;
PM == particulate matter; SOX = sulfur oxide.

Notes: Total system cost = cost of generation plus cost of controls. Amounts do not sum across controls because
many installations run multiple controls (e.g., selective catalytic reduction with a low-NOx wall-fired burner). Capital
costs are amortized.

52


-------
Table A6; Capital and O&M costs, across I !l.\ power plants for select prime movers, 2016

Installation Attribute	Total Costs Including Pollution Control Technologies (MM 2019$)

Prime Mover Type	Code Net Generation (Millions of MWh) Capital Cost	Fixed O&M Cost	Variable O&M Cost

Combined Cycle (Steam) CA

383.10



1,583



260





896

Combined Cycle (Single) CS

40.63



813



149





90

Combined Cycle (Combustion) CT

738.36



18,944



3,287





1,651

Gas Turbine GT

130.03



10,194



2,122





622

Hydraulic Turbine HY

267.81



2,462



7,306





0

Internal Combustion (Diesel) IC

13.52



528



135





66

Photovoltaic PV

32.56



2,382



407





0

Steam Turbine ST

2,243.54



41,088



39,985





14,015

Wind (Onshore) WT

226.92



12,035



3,896





0

Table A7: Total generation cost by plant type and region, 2016















ATB Technology

MRO NPCC RFC

SERC

SPP

TRE

WECC ASCC

HICC

Total USA

Advanced Nuclear

511 2,149 9,321

9,512

187

3,574

3,458

0

0

28,525

Advanced Supercritical Coal Avg CF

3,017 1,162 6,187

7,255

2,828

1,963

3,513

113

773

23,983

Advanced Supercritical Coal High CF

5,890 1,048 16,411

19,419

5,344

5,562

5,932

0

0

54,261

Biomass

518 596 842

2,633

218

86

733

0

58

5,464

Coal IGCC Avg CF

0 0 459

0

0

0

0

0

0

459

Concentrated Solar Power

0 0 0

0

0

0

880

0

0

880

Geothermal - Binary

0 0 0

0

0

0

709

0

14

723

Geothermal - Dual Flash

0 0 0

0

0

0

931

0

23

954

Hydropower New Stream-Reach Development 365 825 474

1,698

222

62

5,924

162

9

9,518

Natural Gas Combined Cycle Avg CF

827 4,210 6,010

9,447

2,360

4,772

8,755

209

226

34,456

Natural Gas Combined Cycle High CF

7 368 1,896

3,907

213

2,205

1,322

0

0

9,704

Natural Gas Combustion Turbine Avg CF

1,012 788 4,154

5,466

989

1,055

2,570

244

62

15,350

Offshore Wind

0 16 0

0

0

0

0

0

0

16

Onshore Wind

2,983 617 1,860

322

2,608

3,705

3,785

12

40

13,324

Solar Photovoltaic

17 96 159

497

27

81

1,945

0

7

2,802

Abbreviations: Avg CF = average capacity factor; High CF = high capacity factor; IGCC = integrated gasification
combined cycle.

Regions: ASCC = Alaska Systems Coordinating Council; HICC = Hawaiian Islands Coordinating Council; MRO
= Midwest Reliability Organization; NPCC = Northeast Power Coordinating Council; RFC §= ReliabilityFirst Cor-
poration; SERC = Southeastern Electric Reliability Council; TRE = Texas Reliability Entity; WECC .= Western
Electricity Coordinating Council.

53


-------
Table A8: Wholesale revenue and total costs by NERC region, 2016



Wholesale Revenue

Fuel Costs

MM 2019$

Capital Costs Fixed O&M Costs

Variable O&M Costs

Total Costs

ASCC

373

335

300

88

17

740

FRCC

4,849

6,007

5,646

2,608

817

15,077

HICC

691

629

442

128

45

1,244

MRO

5,710

2,547

7,781

3,645

1,183

15,156

NPCC

6,444

3,169

4,326

3,914

557

11,967

RFC

26,638

14,324

15,699

13,260

4,560

47,843

SERC

31,462

18,415

20,936

15,462

5,401

60,213

SPP

4,836

3,986

6,705

3,285

1,030

15,005

TRE

9,275

6,336

10,987

4,244

1,498

23,064

WECC

26,974

9,304

17,831

11,099

2,235

40,468

Total USA

117,251

65,051

90,651

57,733

17,343

230,778

Regions: ASCC = Alaska Systems Coordinating Council; HICC = Hawaiian Islands Coordinating Council; MRO
= Midwest Reliability Organization; NPCC = Northeast Power Coordinating Council; RFC = ReliabilityFirst Cor-
poration; SERC — Southeastern Electric Reliability Council; TRE = Texas Reliability Entity; WECC = Western
Electricity Coordinating Council.

Table A9: Criteria pollutant and greenhouse gas emissions by region, 2016



Output



Criteria and Hazardous Air Pollutants (tons)

Greenhouse Gases (MMT C02e)



NERC

Net Generation (in thousands of
MWh)

Sulfur
Oxides

Nitrogen
Oxides

Particulate
Matter

Mercury

Carbon
Dioxide

Nitrous

Oxides Methane

ASCC

6,335

260

340

60

0.00

1

0.03

0.02

FRCC

229,408

34,692

44,973

4,332

0.08

112

0.23

0.14

HICC

9,949

773

526

615

0.00

1

0.02

0.01

MRO

222,471

141,011

98,218

6,033

0.56

136

0.59

0.23

NPCC

241,852

6,922

19,289

184

0.01

61

0.07

0.04

RFC

924,072

411,167

319,268

22,277

0.73

501

1.97

0.81

SERC

1,095,473

360,346

309,439

32,063

0.71

565

1.97

1.08

SPP

231,175

130,252

77,784

6,558

0.26

145

0.51

0.15

TRE



384,152

182,850

72,979

5,659

0.49

200

0.63

0.34

WECC



730,646

90,613

180,586

17,552

2.46

282

0.95

0.50

Total

4,075,533

1,358,887

1,123,401

95,332

5.31

2,003

6.97

3.32

Regions: ASCC ¦ Alaska Systems Coordinating Council; HICC = Hawaiian Islands Coordinating Council; MRO
= Midwest Reliability Organization; NPCC = Northeast Power Coordinating Council; RFC ¦ ReliabilityFirst Cor-
poration; SERC = Southeastern Electric Reliability Council; TRE = Texas Reliability Entity; WECC = Western
Electricity Coordinating Council.

54


-------
Table A10: Criteria pollutant and greenhouse gas emissions rates by region, 2016

NERC

Output

Net Generation (in thousands of
MWh)

Criteria and Hazardous Air Pollutants (tons/MWh)

Sulfur Nitrogen Particulate
Oxides Oxides Matter Mercury

Greenhouse Gases (tons C02e/MWh)

Carbon Nitrous
Dioxide Oxides Methane

ASCC

6,335

0.041

0.054

0.009

0

112

5.12

2.76

FRCC

229,408

0.151

0.196

0.019

0

490

1.00

0.61

HICC

9,949

0.078

0.053

0.062

0

68

2.12

1.11

MRO

222,471

0.634

0.441

0.027

0

613

2.67

1.04

NPCC

241,852

0.029

0.080

0.001

0

250

0.30

0.17

RFC

924,072

0.445

0.346

0.024

0

542

2.13

0.88

SERC

1,095,473

0.329

0.282

0.029

0

516

1.79

0.99

SPP

231,175

0.563

0.336

0.028

0

627

2.22

0.67

TRE

384,152

0.476

0.190

0.015

0

520

1.63

0.88

WECC

730,646

0.124

0.247

0.024

0

386

1.30

0.68

Total

4,075,533

0.333

0.276

0.023

0

491

1.71

0.82

Regions: ASCC = Alaska Systems Coordinating Council; HICC = Hawaiian Islands Coordinating Council; MRO
= Midwest Reliability Organization; NPCC = Northeast Power Coordinating Council; RFC = ReliabilityFirst Cor-
poration; SERC = Southeastern Electric Reliability Council; TRE = Texas Reliability Entity; WECC = Western
Electricity Coordinating Council.

55


-------