The document summarizes key findings from the Energy Modeling Forum Study #37 on pathways to achieving net-zero carbon emissions in the United States by 2050. Models in the study reached net-zero primarily through electrification, reductions in fossil fuel use, and reliance on carbon dioxide removal technologies. Decarbonizing the industrial sector, including technologies like direct reduced iron with hydrogen for steel production, presents challenges to model and requires consideration of impacts beyond just carbon emissions.

1. Ozge Kaplan, Ph.D.
U.S. Environmental Protection Agency
Office of Research and Development
Durham, NC 27711
Industry Decarbonization Scenarios
Disclaimer: The views expressed in this presentation are those of the authors and do not necessarily represent the views or
policies of the U.S. Environmental Protection Agency.
International Energy Agency - ETSAP Summer Workshop
15 June 2023

2. Energy Modeling Forum Study #37:
Deep Electrification and NetZero Carbon
• EPA/ORD's energy team has been participating in Stanford University's Energy Modeling
Forum Study #37 since early 2020.
• The study brought together 16+ energy/economy/environment modeling groups.
• Modeling groups ran harmonized net-zero CO2 scenarios for U.S. to identify robust
insights and compare technology pathways across models.
• Models reach NetZero with some residual emissions
• reliance on carbon dioxide removal (CDR) technologies
• carbon capture and utilization vs. direct air capture
• Low-to-zero carbon electricity followed by building electrification.
• Transportation and industry reductions are at similar level when it comes to full
decarbonization.
• Includes sub-committees focusing on individual end-use sectors and cross-cut themes.
• focusing on in-depth analysis of the industrial decarbonization options.
• Will be published in Special issue of Energy and Climate Change Journal.

3. In the scenario design:
An exogenous 800 Mt CO2/yr
carbon land storage sink was
assumed for the models that do
not include Land Use Land Cover
detail.
Browning, M., et al. (2023) “Net-Zero CO2 by 2050 Scenarios for the United States in the Energy
Modeling Forum 37 Study” Energy and Climate Change 4, 100104
https://doi.org/10.1016/j.egycc.2023.100104
Energy Modeling Forum Study #37:
Deep Electrification and NetZero Carbon
CO2 Emissions in 2050: Reference and Net Zero

4. • Deep decarbonization and net-zero scenarios rely heavily on electrification (Panel a).
• CO2 reductions are directly correlated with reductions in fossil fuel use (Panel b).
• More fossil fuel reductions yield more CO2 reductions.
• Lower reductions in gross emissions mean achieving net-zero emissions requires additional CDR.
Browning, M., et al. (2023) “Net-Zero CO2 by 2050 Scenarios for the United States in the Energy Modeling Forum 37 Study” https://doi.org/10.1016/j.egycc.2023.100104
Bistline, J. E. T. (2021). "Roadmaps to net-zero emissions systems: Emerging insights and modeling challenges." https://doi.org/10.1016/j.joule.2021.09.012
Energy Modeling Forum Study #37:
Deep Electrification and NetZero Carbon
Presenting EMF 37 results in
the context of the literature
(Bistline 2021) (denoted with
an asterisk * in the legend) and
International Panel on Climate
Change (IPCC) reports for the
year 2050. IPCC AR6 results are
only shown for scenarios that
achieve net-zero CO2
emissions by 2050.

5. Browning, M., et al. (2023) “Net-Zero CO2 by 2050 Scenarios for the United States in the Energy Modeling Forum 37 Study” Energy and Climate Change 4, 100104
https://doi.org/10.1016/j.egycc.2023.100104
Energy Modeling Forum Study #37:
Deep Electrification and NetZero Carbon
Change in 2050 final energy use (EJ/yr) by fuel, Net Zero scenario minus Reference scenario
In 2050, some models rely more
heavily on electrification in
industry sector, some rely on
increased use of hydrogen and
biomass.
Which industries are electrifying,
and which are more viable for
fuel switching?

6. Preliminary figures for industrial sector results
REF NetZero NetZero.Ind
REF NetZero NetZero.Ind
REF NetZero NetZero.Ind
REF NetZero NetZero.Ind
Industry working group writing team (Boyd, Kaplan, Browing, Doolin, Dutrow, Perl, Supekar, Victor, Worrell). “Is the Industrial Sector Hard to Decarbonize or Hard
to Model? A Comparison of Industrial Model Structure and Net Zero Scenario Results in EMF37” (in preparation)

7. Is industry “hard to model or hard to decarbonize?”
49.5 billion metric tons of CO2eq 5.9 billion metric tons of CO2eq 1.5 billion metric tons of CO2eq
• Decarbonization of power sector leads reductions
• What are the next most cost-effective industrial sector opportunities?
• Should we tackle fuel combustion and process emissions?
• Growing energy demand for non-manufacturing industry –> which uses low temp heat
• Trade/border adjustment/standard development and market competitiveness
World US US-Industry

8. Transitions need to consider not just carbon emissions
• How will industrial energy consumption and
emissions in the U.S. grow over the next
decades?
• Which industries are the major contributors?
• What type of changes we can anticipate?
• structural and technology change
• fuel switching and energy efficiency
• material efficiency and circular economy
• What are the impacts on endpoints:
• Air quality,
• Human health,
• Environmental and energy justice,
• Climate mitigation and adaptation,
• Resilience and security

9. Economic Modelling
ECONOMY
ENERGY
ENVIRONMENT
Sectoral Characteristics:
e.g., I/O relationships
Economic Policy Characteristics:
e.g., tax/subsidy, Imp/exp
Production Factors:
e.g., capital, labor, goods
Environmental Policy:
e.g., regulations, trade
Energy Sector Modelling
Supply:
e.g., price, resource potential
Technology Characteristics:
e.g., availability, efficiency
Energy Policy:
e.g., subsidies, standards Environmental Limits, taxes, penalties
Emission Taxes
Emissions
Energy Costs
Useful Energy
Demand
Technology and Fuel
Choices, Energy
Consumption
GDP
Adapted from Karali, Xu, and Sathaye. Industrial Sector Energy Efficiency Modeling (ISEEM) Framework Documentation. United States: N. p., 2012. Web. doi:10.2172/1172249.

10. Modeling Technology Change with TIMES
Inputs
• Future-year energy service
demands for each region
• Primary energy resource
supplies
• Current & assumptions on
future technology
characteristics
• Emissions and energy policies
A dynamic, partial-equilibrium, bottom-up, large-scale, linear optimization modeling framework
• multi-period modeling horizon (until 2055)
• minimizes the total discounted system cost while meeting energy service demands
• regional coverage of nine US CENSUS Divisions
• simplified load duration curve
Outputs
• Technology penetrations for
meeting industrial, residential,
commercial, and transportation
end-use demands
• Fuel use by type and region
• Sectoral and system-wide GHG and
air emissions
• Marginal fuel and emissions
reduction prices

11. Modeling iron and steel industry energy service demands in TIMES
• Functioning more as energy accounting
model than an energy-technology
optimization model
• Used MECS data for capacity and fuel usage
in the current stock of establishments in
industrial sector.
• Fixed fractions in the model did not allow us
to explore and analyze a potential structural
shift in that industry.
• Lacked a formal structure to investigate more
advanced technologies, fuel switching and
changes in environmental regulations

12. NAICS: 331110 Iron and Steel Mills and Ferroalloy Manufacturing and 3312 Steel
Product Manufacturing from Purchased Steel
This sector includes all establishments:
(1) direct reduction of iron ore;
(2) manufacturing pig iron in molten or
solid form;
(3) converting pig iron into steel;
(4) making steel;
(5) making steel and manufacturing
shapes (e.g., bar, plate, rod, sheet,
strip, wire);
(6) making steel and forming pipe and
tube;
(7) manufacturing electrometallurgical
ferroalloys.

13. NAICS: 331110 Iron and Steel Mills and Ferroalloy Manufacturing and 3312 Steel
Product Manufacturing from Purchased Steel
Process Heating is major
component of iron and
steel sector – requiring
high temperature heat
Half of the carbon
emissions are process
related

14. Steel making process flow
Primary metals:
• Integrated mills typically use coke to heat and reduce
pelletized iron ore in a blast furnace (BF)
• This produces pig iron.
• Coke is made by heating coal in the absence of air in
coke ovens, and coking facilities are co-located with BFs.
• Pig iron is then fed into a basic oxygen furnace (BOF) ->
Heat + Oxygen further reduces the carbon to produce
crude steel.
Secondary metals:
• Metal recycled scrap can be converted back to steel via
two processes: 1. electric arc furnaces and 2. direct
reduced iron (DRI) process
• DRI process can use natural gas and/or hydrogen as fuel
reducing carbon footprint of steel.
Post processing:
• Steel product manufacturing from produced steel

15. Hybrid industrial sector modeling in TIMES
• 31-33 NAICS level representation
• Suitable for industries with heterogenous mix of
commodities
• Less diversity in end-use service demands
facility level modeling to allow for structural changes
and tracking of goods by physical terms for targeted
sectors
all industrial sectors can be represented with energy
service demands – one size fits all!??
• Appropriate for industries with homogenous outputs –
single commodity
• Candidate sectors: paper, iron and steel, aluminum,
cement, and agricultural chemicals
• Facility level representation with material demand
projections
Linkage between bottom-up models and an I/O economic models to simulate structural changes to industry

16. Heterogenous modeling accommodates…
• Changes to demands for goods and services in 20 sectors in 9 regions
• Where necessary represent demands in tons, rather than value of shipments/energy
• Structural shifts in the current makeup of each industry sector
• changes in product mix, relocation of an industry sector or changes in how industry makes
the goods → iron and steel
• Technological and emissions rate changes
• more efficient boilers, improvements in motors and engines
• Trading between regions
• producing region vs. demand centers → cement production, materials traded globally
• New and existing environmental and energy policies and regulations.

17. Overall Approach
• Change end use demand for selected industry sectors from PJ to physical units such as tons of product
• Gather future demand projections in quantities for major industrial outputs
• Inventory the existing major plant configurations for cement, iron and steel, aluminum, organic and
agricultural chemicals and pulp and paper sector
• Aggregate the individual plant data to come up with model plants per technology type per region
• E.g., one model plant is designed for wet kilns in R5 with current capacity, emission levels, O&M costs
and fuel use
• Review MECS and observed trends in fuel use and investigated fuel switching opportunities per industry
• Identify specific energy service categories that improved over time and have potential for additional
improvement through technology change, fuel switching and other energy efficiency measures
• Model vintage, fuel switching opportunities and emission controls for boilers and CHP
• Streamline fuel-use and process related emissions for industry sector (NEI, GHG reporting rule and industry
data)
• Identify applicable energy and environmental regulations and standards
What would be the future potential plant configurations?

18. Sectors Boilers1 Main Process
Machine
Drive2 Demand
Demand
methodology Advanced Technologies
Primary Steel NS Blast oxygen furnace NS MTonnes Forecasting DRI + H2/CH4
Secondary Steel NS Electric Arc Furnace (EAF) NS MTonnes Forecasting EE (scrap pre-heating)
Primary Aluminum NS Hall-Héroult electrolytic process NS MTonnes Forecasting Carbothermic Aluminum
Secondary Aluminum NS EAF NS MTonnes Forecasting Multi-chamber furnace
Other metals NS NS VOS3 AEO
Pulp Mills Detailed Kraft soda/ mechanical NS MTonnes Forecasting EE (recycled fiber)
Paper Mills Detailed Q MTonnes Forecasting EE
Paperboard Mills Detailed Q MTonnes Forecasting EE
Organic Chemicals Detailed
Ethylene from ethane
Varies
Q VOS AEO
Ethane to ethylene
(dehydrogeneration)
Inorganic Chemicals Detailed Varies by product Hi VOS AEO
Plastics, Fibers and
Resins
Detailed Varies by product Q VOS AEO
Agricultural
Chemicals
Detailed
Haber-Bausch for N
Q MTonnes Forecasting
Other Chemicals Detailed Varies by product Q VOS AEO
Glass NS Q
MTonnes/
VOS
Forecasting
Cement NS
Wet/Dry kilns; Precalciner and
pre-heater designs
NS MTonnes Forecasting Calera and CCS
Other non-metals NS Varies by product Hi VOS AEO
Food Detailed Varies by product Hi VOS AEO
All Others (rest of the
manufacturing)
NS Varies by product Hi VOS AEO
[1] NS: insignificant portion of the energy consumption is attributed to this category.
[2] NS: insignificant portion of the energy consumption is attributed to this category; Q: questionable, needs further evaluation; Hi: signification
portion of the energy consumption is attributed to this category.
[3] VOS: Value of shipments. These then converted into PJ of energy demand using AEO energy intensity values.

19. Key data sources
National Emission Inventory EPA Bottom-up; includes
criteria and hazardous
air pollutants
3-yr cycle; 2017 NEI released
in 2/2021; next one 2020 NEI
https://www.epa.gov/air-emissions-inventories
GHG Inventory EPA Bottom-up calculations;
all sources
Annual since 1990; 2022 https://www.epa.gov/ghgemissions/inventory-
us-greenhouse-gas-emissions-and-sinks
GHG Reporting Program – FLIGHT EPA Reported by facilities;
Large emitters (>25,000
MTCO2e)
Annual since 2010; 2022 https://www.epa.gov/ghgreporting
Manufacturing Energy Consumption
Survey
EIA Reported by facilities
(Form EIA-846A/B)
4-yr cycle; 2018 MECS
released in 2021
www.eia.gov/consumption/manufacturing
Annual Energy Outlook EIA Annual https://www.eia.gov/outlooks/aeo/data/browser
Consumption and Efficiency EIA https://www.eia.gov/consumption
Gross Output by Industry data series BEA Quarterly https://www.bea.gov/data/industries/gross-
output-by-industry
Manufacturing employee estimates Census https://www.census.gov/programs-
surveys/asm/data.html
Input-Output Accounts Data BEA Detailed benchmark
input-output statistics
Annual updates to 71
industry I/O table.
Roughly 4–5-year update
cycle for the 405-industry I/O
table; 2012 last one.
https://www.bea.gov/data/industries/input-
output-accounts-data
https://www.bea.gov/industry/input-output-
accounts-data#supplemental-estimate-tables

20. Iron and steel sector is the key contributor to 2021 GHG
emissions from metals sector
Hydrogen

21. Iron and steel sector
in the US
• Integrated Steel Mills are located near sources
of iron ore (e.g., the Mesabi Range and Upper
Peninsula of Michigan).
• Based on 2017 National Emission Inventory, top
PM2.5 polluters in Indiana are Steel Mills.
• All use BOFs to produce commodities
• Presents a lot of EJ issues.
• Decarbonization of the industry will have
implications on communities surrounding these
facilities

22. 0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
Region 2 Region 3 Region 4 Region 5 Region 6 Region 7 Region 8 Region 9
CO2 Intensity for BOF, EAF (tons/ton steel)
EAF BOF
0.00000
0.00005
0.00010
0.00015
0.00020
0.00025
0.00030
0.00035
0.00040
0.00045
Region 2 Region 3 Region 4 Region 5 Region 6 Region 7 Region 8 Region 9
PM2.5 and PM10 Intensity for BOF, EAF (tons/ton steel)
EAF PM2.5 EAF PM10
BOF PM10 BOF PM2.5
U.S. Census Region Map
Iron and steel sector in the US
More polluting than average coal powered
plant in the US 0.0643lb/MWh = 0.000032
tons/MWh
https://www.epa.gov/sites/default/files/2020-
07/documents/draft_egrid_pm_white_paper_7-20-20.pdf
Avg for BOF=1.6
Avg. for EAF=0.3

23. Industrial Facilities and Demographic Index Correlations

24. • EAF facilities: Located near areas where
labor costs are lower.
• Shifts to EAF will increase the demand for
electricity.
• In the U.S., the contribution of steel
manufactured by EAF is increasing. However,
total steel production is in decline.
Iron and steel sector
in the US

25. Outlook on Decarbonizing the Steel Production Process
60%
65%
70%
75%
80%
2020 2030 2040 2050
Proportion of U.S. steel produced via electric arc furnace
percentage of steel production
Low Oil and
Gas Supply
Reference
Low
Economic
Growth-Low
ZTC
Data source: U.S. Energy Information Administration, Annual Energy Outlook 2023 (AEO2023)
Note: All displayed data are projections. Shaded regions represent maximum and minimum
values for each projection year across the AEO2023 Reference case and side cases. ZTC=Zero-
Carbon Technology Cost.
5
6
7
8
9
10
11
2020 2030 2040 2050
Iron and steel sector energy intensity
thousand British thermal units per 2012 dollar
Low Economic
Growth-Low ZTC
Reference
Low Oil and Gas
Supply
In the Annual Energy Outlook 2023, EIA projects that the share of U.S. steel produced by EAFs is projected to
increase to anywhere from 71% to just above 75% of total U.S. steel production by 2050 while energy intensity
is expected to fall between 12 and 21%.

27. Crude Steel Production by Process
Today, world production of steel is at 1949.9 million tons of steel. BOFs produce around 70% of the world’s
crude steel. EAFs currently produce about 29%.
The steel industry produces approximately 7% of global carbon dioxide (CO2) emissions.
As countries implement roadmaps to reach net-zero emissions, industrial sectors like the iron and steel
industry will most likely be compelled to shift towards lower carbon-intensive processes for steel production.
Global Crude Steel Production (2021) is 1949.9 Million Tons of Steel
70.8
28.9
0.3
Basic Oxygen Furnace %
Electric Arc Furnace %
Open Hearth
Other
Source:
worldsteel.org

28. Steel Trade – United States
Source: U.S. Department of Commerce, Enforcement and Compliance. Trade data from S&P Global, Ltd.
According to the International Trade Administration,
Department of Commerce, the U.S. was the world’s
second largest importer of steel in 2022, and the world’s
18th largest steel exporter in 2021.
For years, the U.S. has maintained a trade deficit
between the amount of steel mill products that it
imports and exports.
In 2022, the U.S. imported 28 million metric tons of steel
products, down about 2% from the previous year (2021).
The U.S. has 81 countries that it trades steel with.
The top three markets for export are Canada, Mexico,
and China.
The top three markets for imports are Canada, Mexico,
and South Korea

29. Outlook on Decarbonizing the Steel Production Process
Europe
• Analysts from S&P Global Commodity
Insights believe that the transition to
EAF steel production in Europe has
challenges but may begin to shift
starting in 2025. A transition may
occur sooner if carbon tax prices are
higher, and S&P Global Platts analysts
believe that carbon prices will
continue to rise until CO2 reductions
begin to occur.
Global
• In 2022, the environmental research firm
Global Energy Monitor (GEM) reported that
a global switch away from BOF steelmaking
to scrap-fed EAF production is happening
slowly. The firm projected that by 2030,
approximately 69% of steel would be
produced in BOFs and 31% from EAFs. This
trend would “remain approximately the
same through to 2050.” Progress in replacing
BOF processes with EAF
technology was “stagnant.”

30. Outlook on Decarbonizing the Steel Production Process
According to DOE/EIA’s Internation Energy Outlook:
• Market incentives needed to shift production away from BF-BOF
processes towards EAF production
• Countries such as China or Japan or South Korea can benefit by shifting
steel production to EAF processes as older equipment nears the end of
its useful lifetime.
• Having sufficient renewable energy to power the EAF process will ensure
a greater reduction in the iron and steel industry’s carbon footprint
• Availability of steel scrap and recycling processes to ensure steady
feedstock for EAF process.
• Limitations on steel scrap and steel purity requirements for certain applications mean that EAFs powered by
renewable energy will not be only solutions. Steel produced by DRI sourced from blue hydrogen must scale
up dramatically.
• How quickly hydrogen produced from electrolysis can reach price parity from hydrogen produced from
fossil fuels like methane is a key question. There needs to be effective implementation of carbon capture
utilization and storage (CCUS) technologies for capturing emissions from fossil fuel derived hydrogen.
https://www.eia.gov/outlooks/ieo/IIF_industrial/

31. Key Data Challenges for Modeling
• Calibration issues
• Year
• Coverage – NAICS vs. other
categorization
• Misaligned data update cycles
• Open data
• Heterogeneity of sectors – one
solution fits all? No?
• Verification and Reconciliation
• Real-time analysis and measurement
• Adaptive decision-making from massive
heterogeneous data sources in
manufacturing space.
• Future Technology Assumptions
• Cost/efficiency
• Siting
• Adoption
• International competitiveness

32. Industrial Sector Modeling Enhancements
- Changes in material flows induced by energy efficiency- and emissions-
related end-use design changes or policy drivers
- Temporal/spatial/magnitude
Life-cycle
analysis
- Resulting industrial demand shifts for manufactured products, in both
related and unrelated industries?
- material efficiency/circular economy/import/export/trade
Economic
Input-Output
Model
- Changes in manufacturing processes and technologies
- Broader impacts on demands for energy system inputs
- fuel switching and energy efficiency
- System-wide energy and emissions implications
Bottom-up
Energy
Systems Model

33. Thank you!
Ozge Kaplan, Ph.D.: Kaplan.Ozge@epa.gov