* \
*1 PROt^
Documentation for Locomotive Component of
the National Emissions Inventory Methodology
-------
-------
EP A-454/B-20-016
May 2011
Documentation for Locomotive Component of the National Emissions Inventory Methodology
Prepared by:
Eastern Research Group
1600 Perimeter Park Drive
Morrisville, North Carolina 27560
Under Contract to:
E.H. Pechan & Associates, Inc.
3622 Lyckan Parkway
Suite 2002
Durham, North Carolina 27707
Prepared for:
Laurel Driver
Emissions, Monitoring and Analysis Division
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
Contract No. EP-D-07-097
ERG No.: 0245.03.402.001
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Air Quality Assessment Division
Research Triangle Park, NC
-------
TABLE OF CONTENTS
Section Page No.
1.0 INTRODUCTION 1-1
1.1 What are Locomotive Sources? 1-1
2.0 DEVELOPMENT OF THE LOCOMOTIVE COMPONENT FOR THE NEI 2-1
2.1 What Pollutants are Included in the National Emission Estimates for
Locomotives? 2-1
3.0 HOW WERE LOCOMOTIVE EMISSIONS ESTIMATED? 3-1
3.1 Line Haul Criteria Emissions Estimates 3-1
3.2 Rail Yard Criteria Emissions Estimates 3-1
3.3 Hazardous Air Pollutant Emissions Estimates 3-2
4.0 HOW WERE COUNTY LINE HAUL EMISSIONS REALLOCATED TO
INDIVIDUAL RAIL SEGMENTS? 4-1
4.1 Class I Line Haul Emissions Allocation 4-1
4.2 Class TT/TTT Line Haul Emissions Allocation 4-2
4.3 Rail Yard Emissions Allocation 4-3
4.4 State Provided Data 4-3
4.5 What are the Results? 4-3
5.0 REFERENCES 5-1
Appendix A - ERTAC Class I Line Haul Documentation
Appendix B - ERTAC Class II/III Line Haul Documentation
Appendix C - ERTAC Rail Yard Documentation
l
-------
LIST OF TABLES AND FIGURES
Tables Page No.
Table 2-1. Locomotive Pollutant List 2-1
Table 2-2. Methods Used to Develop Annual Emission Estimates for Nonroad
Mobile Sources 2-2
Table 3-1. Hazardous Air Pollutant Speciation Profile for 2008 Locomotive
Emission Estimation 3-2
Table 4-1. Line Haul Segment Activity (MGTM/Mi) Categories 4-1
li
-------
10 INTRODUCTION
1.1 What are Locomotive Sources?
The locomotive source category includes railroad locomotives powered by diesel-electric
engines. A diesel-electric locomotive uses 2-stroke or 4-stroke diesel engines and an alternator
or a generator to produce the electricity required to power its traction motors. The locomotive
source category does not include locomotives powered by electricity or steam. Emissions
associated with the operation of electric locomotives would be included in the point source utility
emission estimate. It is believed that the number of wood or coal driven steam locomotives is
currently very small; therefore, these types of locomotives are not included in this inventory.
The locomotive source category is further divided up into three categories: Class I line
haul, Class II/III line haul, and Class I yard. The national rail estimates were developed by the
Eastern Regional Technical Advisory Committee hereafter referenced as ERTAC Rail. This
group is comprised of eastern states' regulatory agencies in collaboration with the rail industry.
ERTAC Rail developed emissions estimates based on fuel data obtained from the American
Association of Railroads for each subcategory. California locomotive emission estimates were
handled separately from the rest of the United States because of their use of low sulfur
locomotive diesel fuels.
1-1
-------
2.0 DEVELOPMENT OF THE LOCOMOTIVE COMPONENT FOR THE
NEI
2.1 What Pollutants are Included in the National Emission Estimates for Locomotives?
All of the criteria pollutants, VOC, CO, NOx, SOx, PM, and PM2.5, are included in the
locomotive component of the NEI. OTAQ identified the HAPs for which data were available to
develop inventory estimates (Scarbro, 2001). The hazardous air pollutants (HAPs), listed below,
were identified based on available test data and accepted emission estimation procedures.
Emission estimation methods have changed over the history of the NEI, as outlined briefly in
Table 2-2 for nonroad sources.
Table 2-1. Locomotive Pollutant List
1,3-Butadiene
Beryllium
Napthalene
2,2,4-Trimethylpentane
Cadmium
n-Hexane
Acenaphthene
Chromium (Hexavalent)
Nickel
Acenaphthylene
Chromium (Trivalent)
Phenanthrene
Acetaldehyde
Chrysene
PAH Propionaldehyde
Acrolein
Dibenz(a,h) anthracene
Pyrene
Anthracene
Ethyl Benzene
Styrene
Arsenic
Fluoranthene
Toluene
Benzene
Fluorene
Xylene
Benzo(a)anthracene
Formaldehyde
Benzo[a]pyrene
Indeno(l,2,3-cd) pyrene
B enzo [b ] fluoranthene
Lead
Benzo[g,h,i,]perylene
Manganese
B enzo [k] fluoranthene
Mercury
2-1
-------
Table 2-2. Methods Used to Develop Annual Emission Estimates for
Nonroad Mobile Sources
(Categories included in this report are noted in bold print)
Category
Base Year
Pollutant(s)
Estimation Method*
NONROAD Categories
Nonroad Gasoline,
Diesel, LPG,
CNG
2008
VOC, NOx, CO, S02,
PM10, PM2 5, NH3, &
HAPs
Emission estimates for NONROAD model engines were developed using EPA's National Mobile Inventory
Model (NMIM), which incorporates NONROAD2008. Where states provided alternate NMIM nonroad
inputs, these data replaced EPA default inputs.
2005
VOC, NOx, CO, S02,,
PM10, PM2 5, NH3, &
HAPs
Emission estimates for NONROAD model engines were developed using EPA's NMIM, which incorporates
NONROAD2005. Where States provided alternate nonroad inputs, these data replaced EPA default inputs.
2002
VOC, NOx, CO, S02,
PM10, PM2 5, NH3, &
HAPs
Emission estimates for NONROAD model engines were developed using EPA's NMIM, which incorporates
NONROAD2004. Where states provided alternate nonroad inputs, these data replaced EPA default inputs.
State-supplied emissions data also replaced default EPA emission estimates.
1999
VOC, NOx, CO, S02,
PM10, PM2 5
Using emission estimates from two emission inventories including: 1) a 1996 county-level inventory,
developed using EPA's October 2001 draft NONROAD model; and 2) an updated 1999 national inventory,
based on EPA's draft Lockdown C NONROAD model (dated May 2002). Using the 1996 county-level
emission estimates, seasonal and daily county-to-national ratios were then developed for application to
updated national estimates per season estimated from the Lockdown C model. Replaced State-submitted data
for California for all NONROAD model categories; Pennsylvania for recreational marine and aircraft ground
support equipment, and Texas for select equipment categories.
1996, 1997,
1998, 2000 &
2001
VOC, NOx, CO, S02,
PM10, PM2 5
Using emission estimates from two emission inventories including: 1) a 1996 county-level inventory,
developed using EPA's October 2001 draft NONROAD model; and 2) updated year-specific national and
California inventories, based on EPA's draft Lockdown C NONROAD model (dated May 2002). Using the
1996 county-level emission estimates, seasonal and daily county-to-national ratios and California county-to-
state ratios were then developed for application to updated national estimates per season estimated from the
Lockdown C model. California results replace the diesel equipment emissions generated from prior
application of county-to-national ratios.
2-2
-------
Table 2-2. Methods Used to Develop Annual Emission Estimates for
Nonroad Mobile Sources (Continued)
(Categories included in this report are noted in bold print)
Category
Base Year
Pollutant(s)
Estimation Method*
Nonroad Gasoline,
Diesel, LPG, and
CNG
(Continued)
1991-1995
VOC, NOx, CO, S02,
PM10, PM2 5, NH3
Using 1990 and 1996 county-level emissions inventories, estimated emissions using linear interpolation of
national emissions between 1990 and 1996. From these emissions, calculated the average annual growth rate
for each pollutant/SCC combination for each year, and then applied the growth factors to 1990 county-level
emissions to estimate 1991-1995 emissions.
1990
VOC, NOx, CO, S02,
PM10, PM2 5
Using emission estimates from two emission inventories including: 1) a 1996 county-level inventory,
developed using EPA's October 2001 draft NONROAD model; and 2) updated 1990 national inventory,
based on EPA's draft Lockdown C NONROAD model (dated May 2002). Using the 1996 county-level
emission estimates, seasonal and daily county-to-national ratios were then developed for application to
updated national estimates per season estimated from the Lockdown C model.
1986, 1988, &
1989
VOC, NOx, CO, S02,
PM10, PM2 5, NH3
Using 1985 and 1990 county-level emissions inventories, estimated emissions using linear interpolation of
national emissions between 1985 and 1990. From these emissions, calculated the average annual growth rate
for each pollutant/SCC combination for each year, and then applied the growth factors to 1985 county-level
emissions to estimate 1986-1989 emissions.
1987
VOC, NOx, CO, S02,
PM10, PM2 5
Using EPA's draft Lockdown C NONROAD model (dated May 2002), developed updated national
emissions for 1987 by running 4 seasonal NONROAD model runs to estimate annual criteria pollutant
emissions. Also performed national NONROAD model runs to estimate typical summer weekday emissions.
1985
VOC, NOx, CO, S02,
PMio, PM2 5
Using emission estimates from two emission inventories including: 1) a 1996 county-level inventory,
developed using EPA's October 2001 draft NONROAD model; and 2) updated 1985 national inventory,
based on EPA's draft Lockdown C NONROAD model (dated May 2002). Using the 1996 county-level
emission estimates, seasonal and daily county-to-national ratios were then developed for application to
updated national estimates per season estimated from the Lockdown C model.
1970, 1975,
1978, & 1980
VOC, NOx, CO, S02,
PM10, PM2 5
Using EPA's draft Lockdown C NONROAD model (dated May 2002), developed updated national
emissions for all years by running 4 seasonal NONROAD model runs to estimate annual criteria pollutant
emissions. Also performed national NONROAD model runs to estimate typical summer weekday emissions.
2-3
-------
Table 2-2. Methods Used to Develop Annual Emission Estimates for
Nonroad Mobile Sources (Continued)
(Categories included in this report are noted in bold print)
Category
Base Year
Pollutant(s)
Estimation Method*
Nonroad Gasoline,
Diesel, LPG, and
CNG
(Continued)
1996, 1997,
1998, 1999,
2000,& 2001
nh3
Obtaining national fuel consumption estimates from the Lockdown C NONROAD model, multiplying by
NH3 emission factors, and distributing to counties using 1996 inventory, based on October 2001 draft
NONROAD. NH3 emissions for California were also recalculated using updated diesel fuel consumption
values generated for California-specific runs, and assuming the 1996 county-level distribution.
1985 & 1990
nh3
Obtaining national fuel consumption estimates from the Lockdown C NONROAD model, multiplying by
NH3 emission factors, and distributing to counties using 1996 inventory, based on October 2001 draft
NONROAD.
1987
nh3
Obtaining 1987 national fuel consumption estimates from Lockdown C NONROAD model and multiplying
by NH3 emission factors.
1970, 1975,
1978, & 1980
nh3
Obtaining national fuel consumption estimates from the Lockdown C NONROAD model and multiplying by
NH3 emission factors.
1990, 1996, &
1999
HAPs
Speciation profiles applied to county VOC and PM estimates. Metal HAPs were calculated using fuel and
activity-based emission factors. Some state data were provided and replaced national estimates. (2003)
Aircraft
Commercial Aircraft
2008
Criteria and HAPs
Federal Aviation Administration (FAA) Emissions and Dispersion Modeling System (EDMS) - Version
5.1.was run using BTS T-100 LTO data. (2009)
2002 and 2005
Criteria and HAPs
Federal Aviation Administration (FAA) Emissions and Dispersion and Modeling System (EDMS) was run
for criteria pollutants, VOC and PM emissions were speciated into HAP components. (2004)
1990, 1996,
1999, 2000,
2001
VOC, NOx, CO, SOx
Input landing and take-off (LTO) data into FAA EDMS. National emissions were assigned to airports based
on airport specific LTO data and BTS GIS data. State data replaced national estimates. (2003)
1970-1998
VOC, NOx, CO, SOx
Estimated emissions for interim years using linear interpolation between available base years. (2003)
1990, 1996,
1999
HAPs
Speciation profiles were applied to VOC estimates to get national HAP estimates. State data replaced
national estimates. (2003)
General Aviation,
Air Taxis
2008
Criteria and HAPs
Federal Aviation Administration (FAA) Emissions and Dispersion Modeling System (EDMS) - Version
5.1.was run using BTS T-100 LTO for aircraft identified as Air taxis. (2010)
Used FAA LTO data from TAF and OTAQ provided activity data for smaller airports derived from FAA
5010 master plans. EPA approved generic emission factors for criteria estimates. Speciation profiles were
applied to VOC and PM estimates to get national HAP estimates. (2010)
2-4
-------
Table 2-2. Methods Used to Develop Annual Emission Estimates for
Nonroad Mobile Sources (Continued)
(Categories included in this report are noted in bold print)
Category
Base Year
Pollutant(s)
Estimation Method*
General Aviation,
Air Taxis
(Continued)
2005
VOC, NOx, CO, S02,
PM10, PM2 5
2002 emissions for approximately 4,000 largest airports were calculated via EDMS and SIP guidance and
included in the 2005 NEI as point sources. Only airports in FAA's T100 and TAF databases were included.
State point source submittals were incorporated.
1978, 1987,
1990, 1996,
1999, 2000,
2001, & 2002
VOC, NOx, CO, S02,
PM10, PM2 5
Used FAA LTO data and EPA approved emission factors for criteria estimates. Speciation profiles were
applied to VOC estimates to get national HAP estimates. State data replaced national estimates. (2004)
1970-1998
VOC, NOx, CO, SOx,
PMio, PM2 5
Estimated emissions for interim years using linear interpolation between available base years. (2003)
1990, 1996,
1999, & 2002
HAPs
Used FAA LTO data and EPA approved emission factors for criteria estimates. Speciation profiles were
applied to VOC estimates to develop national HAP estimates. (2004)
1990, 1996,
1999, & 2002
Pb
Used Department of Energy (DOE) aviation gasoline usage data with lead concentration of aviation gasoline.
(2004)
1996
nh3
Applied NH3 emissions factors to 1996 national jet fuel and aviation gasoline consumption estimates.
Military Aircraft
2008
VOC, NOx, CO, S02,
PMio, PM2 5
Used FAA LTO data as reported in TAF and EPA approved emission factors for criteria estimates.
Representative HAP profiles were not readily available, therefore HAP estimates were not developed. (2010)
2005
VOC, NOx, CO, S02,
PMio, PM2 5
2002 emissions were included in the 2005 NEI as point sources similar to other TAF reported data.
1978, 1987,
1990, 1996,
1999, 2000,
2001, 2002,
2008
VOC, NOx, CO, S02,
PMio, PM2 5
Used FAA LTO data as reported in TAF and EPA approved emission factors for criteria estimates.
Representative HAP profiles were not readily available, therefore HAP estimates were not developed.
1970-1998
VOC, NOx, CO, SOx,
PMio, PM2 5
Estimated emissions for interim years using linear interpolation between available base years. (2003)
Auxiliary Power
Units and Ground
Support Equipment
2008
VOC, NOx, CO, S02,
PMio, PM2 HAPs
Federal Aviation Administration (FAA) Emissions and Dispersion and Modeling System (EDMS) - Version
5. l.was run using BTS T-100 LTO data. (2009)
2002 and 2005
VOC, NOx, CO, S02,
PMio, PM2 5, HAPs
Computed via NONROAD2005 model runs
1985-2001
VOC, NOx, CO, S02,
PMio, PM2 5
Grew 1996 emissions to each year using LTO operations data from the FAA. Estimation methods prior to
1996 reported in EPA, 1998.
2-5
-------
Table 2-2. Methods Used to Develop Annual Emission Estimates for
Nonroad Mobile Sources (Continued)
(Categories included in this report are noted in bold print)
Category
Base Year
Pollutant(s)
Estimation Method*
Unpaved Airstrips1
1985-2001
PM10, PM2 5
Grew 1996 emissions to each year using SIC 45-Air Transportation growth factors, consistent with the
current draft version of EGAS. Estimation methods prior to 1996 reported in EPA, 1998.
Aircraft Refueling1
1985-2001
voc
Grew 1996 emissions to each year using SIC 45-Air Transportation growth factors, consistent with the
current draft version of EGAS. Estimation methods prior to 1996 reported in EPA, 1998.
Commercial Marine Vessel (CMV)
All CMV Categories
2008
VOC, NOx, CO, S02,
PM10, PM2 5
OTAQ provided CAP emission estimates for all CMV categories. Note that the SCCs for this category have
changed such that the Diesel category refers to smaller vessels (Category 1 and 2) using distillate fuels and
the Residual category refers to larger (Category 3) vessels using a blend of residual fuels. Emissions were
allocated to segments using GIS shapefiles and adjusted based on limited state data (2010)
2008
HAPs
OTAQ's 2008 estimates were speciated into HAP components using SEPA profiles (2009)
CMV Diesel
2002 and
2005
VOC, NOx, CO, S02,
PMio, PM2 5
2001 Estimates carried over. Used state data when provided. (2004)
HAPs
1999 Estimates carried over. Used state data when provided. (2004)
1978, 1987,
1990, 1996,
1999, 2000, &
2001
VOC, NOx, CO, SOx,
PM10, & PM2 5
Used criteria emission estimates in the background document for marine diesel regulations for 2000.
Adjusted 2000 criteria emission estimates for other used based on fuel usage. Emissions were disaggregated
into port traffic and underway activities. Port emissions were assigned to specific ports based on amount of
cargo handled. Underway emissions were allocated based on Army Corp of Engineering waterway data.
State data replaced national estimates. (2003)
1970-1998
VOC, NOx, CO, SOx,
PMio, PM2 5
Estimated emissions for interim years using linear interpolation between available base years. (2003)
1990, 1996,
1999
HAPs
VOC and PM emission estimates were speciated into HAP components. State data replaced national
estimates. (2003)
1996
nh3
Applied NH3 emissions factors to 1996 distillate and residual fuel oil estimates (i.e., as reported in EIA,
1996).
1990-1995
nh3
Estimation methods reported in EPA, 1998.
2-6
-------
Table 2-2. Methods Used to Develop Annual Emission Estimates for
Nonroad Mobile Sources (Continued)
(Categories included in this report are noted in bold print)
Category
Base Year
Pollutant(s)
Estimation Method*
CMV Steam
Powered
2005
VOC, NOx, CO, SOx,
PM10, &PM25,HAPs
2002 estimates grown to 2005 (2008).
2002
VOC, NOx, CO, SOx,
PM10, &PM25,HAPs
2002 based estimates were developed for port and underway category 3 (C3) vessels as part of a rulemaking
effort. Emissions were developed separately for near port and underway emissions. For near port
emissions, inventories for 2002 were developed for 89 deep water and 28 Great Lake ports in the U.S. The
Waterway Network Ship Traffic, Energy, and Enviromnental Model (STEEM) was used to provide
emissions from ships traveling in shipping lanes between and near individual ports (2008)
1978, 1987,
1990, 1996,
1999, 2000, &
2001
VOC, NOx, CO, SOx,
PM10, & PM2 5
Calculated criteria emissions based on EPA SIP guidance. Emissions were disaggregated into port traffic
and under way activities. Port emissions were assigned to specific ports based on amount of cargo handled.
Underway emissions were allocated based on Army Corp of Engineering waterway data. State data replaced
national estimates. (2003)
1970-1998
VOC, NOx, CO, SOx,
PMio, PM2 5
Estimated emissions for interim years using linear interpolation between available base years. (2003)
1990, 1996, &
1999
HAPs
VOC and PM emission estimates were speciated into HAP components. State data replaced national
estimates. (2003)
Military Marine
1997-2001
VOC, NOx, CO, S02,
PMio, PM25
Applied EGAS growth factors to 1996 emissions estimates for this category.
CMV Coal,2 CMV,
Steam powered,
CMV Gasoline2
1997-1998
VOC, NOx, CO, S02,
PMio, PM25
Applied EGAS growth factors to 1996 emissions estimates for this category.
CM Coal, CMV,
Steam powered,
CMV Gasoline,
Military Marine
1991-1995
VOC, NOx, CO, S02,
PMio, PM25
Estimation methods reported in EPA, 1998.
Locomotives
Class I, II, III and
Yard operations
2008
VOC, NOx, CO, PM10,
PM2 5,SOx& HAPs
Criteria emission estimates were provided to EPA by ERTAC. These data were assigned to individual
railway segments using DOT shapefiles and guidance from ERTAC. HAP emissions were calculated
by applying speciation profiles to VOC and PM estimates. (2010)
2-7
-------
Table 2-2. Methods Used to Develop Annual Emission Estimates for
Nonroad Mobile Sources (Continued)
(Categories included in this report are noted in bold print)
Category
Base Year
Pollutant(s)
Estimation Method*
Class I, Class II,
Commuter,
Passenger, and Yard
Locomotives
1978, 1987,
1990, 1996,
1999, 2000,
2000, 2002, &
2005
VOC, NOx, CO, PM10,
pm25
Criteria pollutants were estimated by using locomotive fuel use data from DOE EIA and available emission
factors. County-level estimates were obtained by scaling the national estimates with the rail GIS data from
DOT. State data replaced national estimates. (2004)
1978, 1987,
1990, 1996,
1999, 2000,
2001, 2002, &
2005
S02
SOx emissions were calculated by using locomotive fuel use and fuel sulfur concentration data from EIA.
County-level estimates were obtained by scaling the national estimates with the county level rail activity
data from DOT. State data replaced national estimates. (2004)
1970-1998
VOC, NOx, CO, SOx,
PMio, PM2 5
Estimated emissions for interim years using linear interpolation between available base years. (2003)
1990, 1996,
1999, & 2002
HAPs
HAP emissions were calculated by applying speciation profiles to VOC and PM estimates. County-level
estimates were obtained by scaling the national estimates with the county level rail activity from DOT.
State data replaced national estimates. (2004)
1997-1998
nh3
Grew 1996 base year emissions using EGAS growth indicators.
1996
nh3
Applied NH3 emissions factors to diesel consumption estimates for 1996.
1990-1995
nh3
Estimation methods reported in EPA, 1998.
Notes:
* Dates included at the end of Estimation Method represent the year that the section was revised.
1 Emission estimates for unpaved airstrips and aircraft refueling are included in the area source NEI, since they represent non-engine emissions.
2 National Emission estimates for CMV Coal and CMV Gasoline were not developed though states and local agencies may have submitted estimates for these source
categories.
EPA, 1998. U.S. Enviromnental Protection Agency, Office of Air Quality Planning and Standards, Emission Factors and Inventory Group, National Air Pollutant
Emission Trends, Procedures Document, 1900-1996, EPA-454/R-98-008. May 1998.
2-8
-------
3 0 HOW WERE LOCOMOTIVE EMISSIONS ESTIMATED?
ERTAC Rail used confidential railroad-provided data to generate railroad-specific
criteria emission estimates for line haul and rail yards at the rail segment and rail yard level,
respectively. Appendices A-C provide more detail on how emissions were developed and
includes critical data used in calculating these estimates. This section of the report describes the
emission estimating methods used in general terms as well as the approach for reallocating the
emissions to protect confidential data. The data and documentation provided with respect to
ERTAC Rail's emission estimates pertain to the version that was incorporated into the NEI and
does not reflect recent revisions.
3.1 Line Haul Criteria Emissions Estimates
Criteria pollutant emissions were estimated by applying emission factors to the total
amount of distillate fuel oil used by line haul locomotives. Fuel usage was obtained from
publically available Class I Railroad Annual Reports (Form R-l). The R-l reports are submitted
to the Surface Transportation Board annually and include financial and operations data to be
used in monitoring rail industry health and identifying changes that may affect national
transportation policy. Additionally, each railroad provided fleet mix information that allowed
ERTAC Rail to calculate railroad-specific emission factors. Weighted Emission Factors (EF)
per pollutant for each gallon of fuel used (gm/gal or lbs/gal) were calculated for each Class I
railroad fleet based on its fraction of line haul locomotives at each regulated Tier level. EPA
emission factors were used for PM2.5, SO2, and NH3.
The weighted emission factors were then applied to the link-specific fuel consumption to
obtain emissions for each rail segment. Given the confidentiality of the activity data, emissions
for criteria pollutants were provided to EPA by ERTAC Rail by county for Class I line haul.
Class II/III rail was provided by railroad company and county. Appendices A and B provide
more detail on the Class I and Class II/III line haul emission development, respectively.
3.2 Rail Yard Criteria Emissions Estimates
Rail yard locations were identified using a database from the Federal Railroad
Administration. Criteria pollutant emissions were estimated by applying emission factors to the
total amount of distillate fuel used by locomotives. Each railroad provided fleet mix information
that allowed ERTAC to calculate railroad-specific emission factors. The company-specific,
system wide fleet mix was used to calculate weighted average emissions factors for switchers
operated by each Class I railroad. EPA emission factors were used for PM2.5, S02, and NH3.
R-l report-derived fuel use was allocated to rail yards using an approximation of line
haul activity data within the yard; see Appendix C for more details. These fuel consumption
values were further revised by direct input from the Class I railroads. The weighted emission
factors were then applied to the yard-specific fuel consumption to obtain emissions for each
yard. Since the rail yard inventory was based on publically-available data, the final criteria
emission estimates were provided per rail yard.
3-1
-------
3.3 Hazardous Air Pollutant Emissions Estimates
HAP emissions were estimated by applying speciation profiles to the VOC or PM
estimates. The speciation profiles were derived from Evaluation of Factors that Affect Diesel
Exhaust Toxicity (Truex and Norbeck, 1998), and data provided by OTAQ (Scarbro, 2001 and
2002). It should be noted that since California uses low sulfur diesel fuel and emission factors
specific for California railroad fuels were available, calculations of the state's emissions were
done separately from the other states. The HAP speciation profile used in this effort is shown in
Table 3-1. HAP estimates were calculated at the yard and link level, after the criteria emissions
had been allocated.
Table 3-1. Hazardous Air Pollutant Speciation Profile for 2008 Locomotive Emission
Estimation
Pollutant Name
California
All Other
States
Speciation
Base
1,3 Butadiene
0.0000615
0.0047735
PMio
2-2-4 Trimethylpentane
0.0022425
0.0022425
VOC
Acenaphthene
0.0000080
0.0000306
PMio
Acenaphthylene
0.0002182
0.0004275
PMio
Acetaldehyde
0.0004492
0.0276274
PMio
Acrolein
0.0000855
0.0045943
PMio
Anthracene
0.0000535
0.0001009
PMio
Arsenic
0.0000004
0.0000004
PMio
Benzene
0.0000517
0.0038020
PMio
Benzo(a)anthracene
0.0000121
0.0000160
PMio
Benzo(a)pyrene
0.0000044
0.0000027
PMio
B enzo(b )fluoranthene
0.0000044
0.0000064
PMio
B enzo(ghi )pery 1 ene
0.0000044
0.0000031
PMio
Benzo(k)fluoranthene
0.0000044
0.0000052
PMio
Beryllium
0.0000280
0.0000280
PMio
Cadium
0.0000280
0.0000280
PMio
Chromium (III)
0.0000001
0.0000040
PMio
Chromium (VI)
0.0000000
0.0000021
PMio
Chrysene
0.0000092
0.0000119
PMio
Dibenz(a,h)anthracene
0.0000000
0.0000000
PMio
Ethylbenzene
0.0020000
0.0020000
VOC
Fluoranthene
0.0000601
0.0000746
PMio
Fluorene
0.0000619
0.0001407
PMio
Formaldehyde
0.0009451
0.0636582
PMio
Indeno( 1,2,3 -cd)pyrene
0.0000033
0.0000027
PMio
Lead
0.0000840
0.0000840
PMio
Manganese
0.0000020
0.0000020
PMio
Mercury
0.0000280
0.0000280
PMio
Napthalene
0.0018505
0.0025756
PMio
n-Hexane
0.0055000
0.0055000
VOC
3-2
-------
Table 3-1. Hazardous Air Pollutant Speciation Profile for 2008 Locomotive Emission
Estimation (Cont.)
Pollutant Name
California
All Other
States
Speciation
Base
Nickel
0.0000066
0.0000066
PMio
Phenanthrene
0.0002822
0.0005671
PM10
Propionaldehyde
0.0061000
0.0061000
voc
Pyrene
0.0000771
0.0001054
PMio
Styrene
0.0021000
0.0021000
VOC
Toluene
0.0032000
0.0032000
voc
Xylene
0.0048000
0.0048000
voc
3-3
-------
4 0 HOW WERE COUNTY LINE HAUL EMISSIONS REALLOCATED
TO INDIVIDUAL RAIL SEGMENTS?
4.1 Class I Line Haul Emissions Allocation
Class I line haul emissions were allocated to rail segments based on segment-specific
railroad traffic data (ton miles) obtained from the Department of Transportation (BTS, 2009).
This dataset categorizes the segments' level of activity into ranges of MGTM and is populated
by FRA. Emissions were divided between all mainline segments using these activity ranges as a
proxy to allocate more emissions to segments with higher activity.
Since the activity data were provided as ranges, a single "allocation value", typically the
midpoint of the range, was selected for use in the emissions allocation. The exception to this
was the "0" activity category, which by definition had "unknown" activity. As a result, most
mainline segments with the "0" activity category were not included in the emissions
calculation/allocation. However, there was a small subset of segments that did have known
activity values in the confidential data set but were labeled as "unknown" in the publically
available data set. Those segment IDs were provided by ERTAC Rail for inclusion in the
emission allocation; however, the activity of these segments was averaged to protect confidential
data. Table 4-1 lists the activity categories along with their ranges in MGTM/mi and the
allocation value used in the emissions spatial allocation.
Table 4-1. Line Haul Segment Activity (MGTM/Mi) Categories
Category
Range
Minimum
Range
Maximum
Allocation
Value Used
0*
0.0003
0.09
0.01233
1
0.1
4.9
2.5
2
5
9.9
7.45
3
10
19.9
14.95
4
20
39.9
29.95
5
40
59.9
49.95
6
60
99.9
79.95
7
100
1000000
100
* The "0" category has "unknown" activity in the publically
available segment data. As a result, this table lists the minimum,
maximum, and average of the confidential activity data greater
than zero that were categorized as "unknown" in the public data.
The county emission sums were reallocated to the segments by multiplying the county
emissions by the segment's allocation value divided by the sum of the allocation values for all
links within the county.
4-1
-------
A
Eil =Elc*lr-^-
lAc
C=1
Where:
EiL = Emissions of pollutant i per link L (tons/year).
Etc = Emissions of pollutant i per county C (tons/year).
Al = Allocation value for link L per activity category from public BTS dataset
ALc= Sum of allocation values for all links in county C from public BTS dataset
Note that rail line data for Puerto Rico, U.S. Virgin Islands, and Hawaii data were not
included in ERTAC Rail's shapefile and were developed separately; however, since these areas
have exclusively Class II/III railroads present, these efforts are discussed in the following
section.
4.2 Class II/III Line Haul Emissions Allocation
ERTAC Rail created a shapefile of Class II/III mainline rail segments from their FRA-
provided proprietary shapefile as described in Appendix B for the contiguous 48 states and
Alaska. Raw rail line data for Puerto Rico were obtained from USGS (Scanlon and Briere,
2000), and rail line data for Hawaii was obtained from ESRI's Digital Chart of the World (ESRI
2010). The U.S. Virgin Islands have no rail lines. Because Class II/III railroads are less likely to
use rail segments that are heavily traveled by Class I railroads, the activity-based approach used
for Class I lines was not appropriate. Instead, Class II/III line haul emissions were allocated to
rail segments using segment length as a proxy.
The county emission sums were reallocated to the segments by multiplying the county
emissions by the segment's length divided by the sum of the length for all links within the
county.
E,l =E,c*-^~
I',c
C=1
Where:
EiL = Emissions of pollutant i per link L (tons/year).
Etc = Emissions of pollutant i per county C (tons/year).
Il = Allocation value for link L per activity category from public BTS dataset
Ilc = Sum of allocation values for all links in county C from public BTS dataset
Since ERTAC Rail used proprietary data to develop the shapefile, some segment IDs
were not found in the EIS data set. These segments were manually identified, and their
emissions were allocated to the nearest segment within the EIS data set.
4-2
-------
4.3
Rail Yard Emissions Allocation
Rail yard emissions were developed based on yard name and ownership properties. As a
result, unique yards needed to be identified and emissions summed. Unfortunately, the yard data
lacked detail necessary for confident duplicate checks and yard matching such as address,
detailed yard name, etc. As a result, a GIS was used to find the centroid of the yards based on
the latest public BTS rail network, using the yard name and FIPS. The list of unique yards was
further examined against ERTAC's data and within Google Earth to identify any yards that
required further revision. A crosswalk of original ERTAC data to new, consolidated yard IDs
facilitated the summing of activity and emissions. 753 unique yards were identified nationwide.
This underestimate of the total number of yards is most likely due to using line-haul-focused data
to identify locations and develop rail yard emissions.
Once the unique yards were identified and criteria emissions were summed at the yard,
the PM and VOC-based HAP speciation profile was applied to estimate HAP emissions at each
yard.
4.4 State Provided Data
In this version of NEI, state and local agencies were invited to provide locomotive data
that replaced the estimates based on national fuel consumption. However, only a small rail yard
dataset was received from Kentucky. Their rail yard list was compared with the ERTAC/ERG
yard list, and 2 yards were found in both sets. These yards were merged so as to avoid
duplication in activity or emissions.
4.5 What are the Results?
Table 3 summarizes the 2008 locomotive mobile source emission estimates.
Table 3. 2008 Locomotive Emissions Data
2008 Locomotive Criteria Emissions
Pollutant Name
Class I
Line Haul
Class II/III
Line Haul
Rail Yard
TOTAL
CO
110,969
5,055
9,152
125,176
NH,
347
16
27
390
NOx
754,433
51,342
73,741
879,516
PMio-PRI
25,477
1,264
2,086
28,827
pm25-pri
23,439
1,163
2,024
26,626
S02
7,836
357
619
8,811
VOC
37,941
1,896
4,824
44,661
2008 Locomotive Hazardous Air Pollutant Emissions
Pollutant Name
Class I Line
Haul
Class II/III
Line Haul
Rail Yard
TOTAL
1,3 Butadiene
116.7941
5.7969
9.3296
131.9206
2-2-4 Trimethylpentane
85.0832
4.2511
10.8178
100.1521
Acenaphthene
0.7569
0.0376
0.0609
0.8554
4-3
-------
Table 3. 2008 Locomotive Emissions Data (Cont.)
2008 Locomotive Hazardous Air Pollutant Emissions
Pollutant Name
Class I Line
Haul
Class II/III
Line Haul
Rail Yard
TOTAL
Acenaphthylene
10.6772
0.5298
0.8639
12.0709
Acetaldehyde
676.0572
33.5552
54.0089
763.6213
Acrolein
112.4351
5.5806
8.9828
126.9985
Anthracene
2.5231
0.1252
0.2042
2.8525
Arsenic
0.0091
0.0005
0.0007
0.0103
Benzene
93.0272
4.6173
7.4312
105.0757
Benzo(a)anthracene
0.4047
0.0201
0.0329
0.4577
Benzo(a)pyrene
0.0717
0.0036
0.0059
0.0812
Benzo(b)fluoranthene
0.1607
0.0079
0.0131
0.1817
Benzo(ghi)perylene
0.0798
0.0040
0.0066
0.0904
Benzo(k)fluoranthene
0.1312
0.0065
0.0107
0.1484
Beryllium
0.7138
0.0354
0.0584
0.8076
Cadium
0.7138
0.0354
0.0584
0.8076
Chromium (III)
0.0985
0.0049
0.0079
0.1113
Chromium (VI)
0.0508
0.0025
0.0041
0.0574
Chrysene
0.2998
0.0149
0.0244
0.3391
Ethylbenzene
75.8814
3.7914
9.6479
89.3207
Fluoranthene
1.8868
0.0936
0.1538
2.1342
Fluorene
3.5039
0.1739
0.2830
3.9608
Formaldehyde
1,557.66
77.3124
124.4335
1759.4059
Indeno( 1,2,3 -cd)pyrene
0.0684
0.0034
0.0056
0.0774
Lead
2.1413
0.1062
0.1753
2.4228
Manganese
0.0520
0.0026
0.0043
0.0589
Mercury
0.7138
0.0354
0.0584
0.8076
Napthalene
64.8766
3.2187
5.2765
73.3718
n-Hexane
208.6739
10.4263
26.5317
245.6319
Nickel
0.1669
0.00983
0.0137
0.19043
Phenanthrene
14.1555
0.7024
1.1450
16.0029
Propionaldehyde
231.4383
11.5637
29.4261
272.4281
Pyrene
2.6566
0.1318
0.2161
3.0045
Styrene
79.6755
3.9809
10.1303
93.7867
Toluene
121.4103
6.0662
15.4366
142.9131
Xylene
182.1154
9.0993
23.1549
214.3696
4-4
-------
5 0 REFERENCES
Bureau of Transportation Statistics, 2009. National Transportation Atlas Databases - National
Rail Network 1:2,000,000. Washington, DC, Publisher: Bureau of Transportation Statistics.
Energy Information Administration Form EIA-821, "Annual Fuel Oil and Kerosene Sales
Report" for 1999. Table 23: Adjusted Sales for Transportation Use: Distillate Fuel Oil Residual
Fuel Oil, 1999, U.S.
ESRI, Digital Chart of the World Hawaii Rail line dataset
http://data.geocomm.com/catalog/US/61094/group 103 .html July 27, 2010
Fritz, Steve, Diesel Fuel Effects on Locomotive Exhaust Emissions, California Air Resource
Board. SwRI 08.02062, October 2000.
Porter, Fred L., U.S. Environmental Protection Agency, Emission Standards Division. Note to
Anne Pope, U.S. EPA/Emissions, Monitoring and Analysis Division. Comments on combustion
source information in the Baseline Emission Inventory of HAP Emissions from MACT Sources -
Interim final Report {September, 18, 1998. November 13, 1998)
Scanlon, Kathryn and Peter R. Briere, U.S. Geological Survey Open-File Report 00-006. Puerto
Rico Marine Sediments, Terrestrial and Seafloor Imagery, and Tectonic Interpretations, 2000.
http://pubs.usgs.gov/of/2000/of00-006/htm/index.htm
Scarbro, Carl, E-mail entitled A Few Questions on the Rail Emissions - Reply, to Richard
Billings, and Roger Chang, Eastern Research Group, Inc., United States Environmental
Protection Agency Office of Transport and Air Quality. July 19, 2001
Scarbro, Carl, E-mail entitled Chromium in Loco's - Reply, to Richard Billings, Eastern Research
Group, Inc., United States Environmental Protection Agency Office of Transport and Air
Quality. June 1, 2001
Scarbro, Carl, E-mail entitled Better Railroad Numbers This Will Disaggregate Class I Work, to
Roger Chang, Eastern Research Group, Inc., United States Environmental Protection Agency
Office of Transport and Air Quality. May 8, 2001
Scarbro, Carl, E-mail entitled CMVSOx corrections - Reply, to Richard Billings, Eastern
Research Group, Inc., United States Environmental Protection Agency Office of Transport and
Air Quality. May 28, 2002
Scarbro, Carl, E-mail entitled 2, 2, 4-trimethylpentane, to Richard Billings, Eastern Research
Group, Inc., United States Environmental Protection Agency Office of Transport and Air
Quality. June 1, 2001
Scarbro, Carl, E-mail entitled 2, 2, 4-trimethylpentane, to Roger Chang, Eastern Research Group,
Inc., United States Environmental Protection Agency Office of Transport and Air Quality.
March 26, 2002
5-1
-------
Truex, Timothy J. and Joseph M. Norbeck. Evaluation of Factors that Affect Diesel Exhaust
Toxicity. University of California-Riverside, Center for Environmental Research and
Technology. Riverside, CA. March 16, 1998.
U.S. Environmental Protection Agency Form APR420-F-97-051, Emission Factors for
Locomotives, for 1996 Table 9: Fleet Average Emission Factors for All Locomotives (Projected
1999), December 1997
U.S. Environmental Protection Agency, Locomotive Emission Standards Regulatory Support
Document, page 109 April 1998.
U.S. Environmental Protection Agency, Procedures for Emission Inventory Preparation, Volume
IV: Mobile Sources. 1992.
U.S. Environmental Protection Agency. Procedures for Emission Inventory Preparation,
Volume IV: Mobile Sources. Office of Air Quality Planning and Standards. Research Triangle
Park, NC. 1989.
5-2
-------
Appendix A
ERTAC Class I Line Haul Documentation
-------
DRAFT
ERTAC Rail Emissions Inventory
Part 1: Class I Line-Haul Locomotives
Michelle Bergin, GA Environmental Protection Division
Matthew Harrell, IL Environmental Protection Agency
Mark Janssen, Lake Michigan Air Directors Consortium
Acknowl edgments:
Robert Fronczak, Association of American Railroads
Raquel Wright, Federal Railroad Administration
Julie McDill, Mid-Atlantic Regional Air Management Association
Patrick Davis, Mid-Atlantic Regional Air Management Association
Laurel Driver, US EPA, Office of Air Quality Planning and Support
Byeong Kim, GA Environmental Protection Division
Introduction
Air protection agencies from twenty-seven states, coordinated through the Eastern Regional
Technical Advisory Committee (ERTAC) and headed by the Lake Michigan Air Directors
Consortium (LADCO), identified a need to better quantify and characterize rail-related emissions
inventories. Traditional locomotives largely utilize diesel engines, resulting in emissions of
NOx, diesel PM, hydrocarbons, greenhouse gases, and other pollutants. These emissions are
sometimes concentrated in areas exceeding National Ambient Air Quality Standards. No
cohesive nationwide railroad emission estimates based on local operations are known to have
been made previously. Inventory development methods for locomotive emissions estimates vary
from state to state and, in general, lack the spatial or temporal resolution needed to support air
quality modeling and planning 1-5.
The ERTAC Rail Subcommittee (ERTAC Rail) was established with active representatives from
twelve member states, three regional planning offices, and the US EPA. The subcommittee's
goals are to (1) standardize agencies' inventory development methods through a collaborative
effort, (2) improve the quality of data received and the resulting emission inventories, and (3)
reduce the administrative burden on railroad companies of providing data.
With support from the Rail industry and assistance from the ERTAC Rail Data Workgroup
(Appendix A), ERTAC Rail has developed 3 inventories of locomotive emissions (Table 1);
from Class I line-haul, Shortline and Regional Railroads (Class II and III operations), and Class I
railyard switchers. Because of the difficulty in obtaining data and differences in states' needs for
inventory years, sources from both 2007 and 2008 were utilized (Appendix B.) Due to the
variability and uncertainty in much of the data, the results are considered applicable for either
2007 or 2008.
A-l
-------
The Surface Transportation Board (STB) defines Class I Railroads as having had minimum
carrier operating revenues of $401.4 million (USD) in 2008. There are 8 Class I Railroads
operating in the United States (Table 2), about 12 Regional Railroads (Class II), and
approximately 530 Class III Railroads (Shortlines). While categorized as a Class I Railroad,
Amtrak was excluded from these inventories because of significant differences in equipment and
operation characteristics. Line-haul locomotives travel long distances (e.g. between cities) while
switcher locomotives largely operate in railyards, splitting and joining rail cars with varying
destinations. Passenger and Commuter Rail (including Amtrak), industrial locomotives, and
associated non-locomotive equipment are not included in these inventories.
This paper documents the data sources and methodologies used for calculating the Class I line-
haul emissions inventory. Class I line-haul activities are the largest source of rail-related
emissions, with estimates of Class I line-haul fuel consumption totals to be from 74 to 84% of all
rail sources combined4,5. For this reason, characterizing Class I line-haul emissions were a focal
point of ERTAC Rail's inventory development efforts. Information on ERTAC Rail, Railroad
participation, the Rail industry, and effects of rail on air quality are available elsewhere6.
Table 1. Summary of ERTAC Rail Inventories: U.S. Locomotive Emissions and Fuel Use
for either 2007 or 2008*.
Fuel Use**
Emissions
(tons/yr)
(gal/yr)
NOx
pm2,
HC
S02
CO
nh3
Class I*** line-
3,770,914,002
754,443
23,439
37,941
7,836
110,969
347
haul
Class I switcher
300,492,223
73,741
2,024
4,824
619
9,152
27
Class II and III
157,800,000
51,367
1,163
1,897
357
5,058
16
*See Appendix B
for a description o
? the year and source of data utilized for each inventory.
"Locomotive grade diesel
* "Excluding Amtrak and including work train fuel use
Table 2. Class I Railroads, Reported Locomotive Fuel Use,
and Railroad Fuel Consumption Index (RFCI) 7.
R-l Reported Locomotive Fuel
Class I Railroads*
Use (gal/yr)
RFCI
Line-Haul
(2007)**
Switcher
(2008)
(ton-miles/gal)
BNSF
1,393,874,954
52,497,057
883.14
Canadian National
93,830,751
12,290,022
1190.79
Canadian Pacific***
50,320,233
4,594,067
1096.28
CSX
514,687,186
53,717,674
963.81
Kansas City Southern
69,787,071
1,816,759
785.89
Norfolk Southern
463,267,278
32,317,375
865.75
Union Pacific
1,185,146,529
143,470,336
974.64
Total
3,770,914,002
300,492,223
929.47
A-2
-------
* Excluding Amtrak
** Includes work trains
*** CP's line-haul fuel use values include 2008 data (rather than 2007) for their Delaware and
Hudson subsidiary.
Method
Earlier efforts to characterize line-haul railroad emissions relied on highly aggregated activity
data (Figure 1), and generally apportioned annual system-wide fuel use equally across all route
miles of track operated by a Class I railroad. However, the majority of freight tonnage carried by
Class I railroads is concentrated on a disproportionately small number of route miles. In
addition, emissions calculations were previously based on an estimate of annual nationwide-
average locomotive fleet mix to create one set of emissions factors.
For this inventory, the Class I Railroads allowed ERTAC Rail access under a confidentiality
agreement to a link-level (single lengths of track) line-haul GIS layer activity dataset managed
by the Federal Railroad Administration9 Each railroad also provided fleet mix information that
allowed ERTAC Rail to calculate weighted emission factors based on the fraction of their line-
haul fleet meeting each Tier level category. The use of this data, largely following a line-haul
inventory methodology recommended by Sierra Research'3, resulted in a link-level line-haul
locomotive emission inventory using railroad-specific emission factors. This segment-level
inventory is nationwide, aggregated to state and county level files, and will be released as
gridded emissions files for use in photochemical and dispersion modeling. Link-level emissions
may be provided for special study requests pending approval of any Class I railroads operating in
the study domain. The calculations are described below as a two-part process, calculating
railroad-specific factors and emissions per rail link.
Figure 1. US Railroad Traffic Density in 2006.8 MGT is million gross tons.
Traffic Density
0.1 -4.9 MGT
5.0 - 9.9 MGT
10.0- 19.9 MGT
20.0 - 39.9 MGT
40.0 - 59.9 MGT
60.0 - 99.9 MGT
100.0+ MGT
A-3
-------
1. Calculate Railroad-Specific Factors.
The EPA provides annual default Emission Factors for locomotives based on characteristic
operating cycles ('duty cycles') and the estimated nationwide fleet mixes for both switcher and
line-haul locomotives. However, fleet mixes vary from railroad to railroad and, as can be seen in
Figure 2, Class I railroad activity is highly regionalized in nature and subject to issues of local
terrain such as operation on plains vs. mountainous areas, which can have a significant impact on
fuel consumption and emissions.
Railroad Network of the
UNITED STATES 2008
Mapilnw ill tea ewiinlip Iriwi CN N5
oi .DCS NUumri Tntrapoil^Bri Mhk ^ ijp
LuuImh patfWMd by 9ib US. DOT* .
InadTrwOaUn. Clt ,hor L; e»,,en.l
4>1993-2D13, Assaaaton ol .Vercan RaBasta. For mare infon-atlon about i&Boasis ytsls «*w.aar.aTi orvw».Trela(ilfallwort.s.cr3. Fetrua-^'2310
Figure 2. Class I Railroad Territories in the United States10.
As an alternative approach to using a single nationwide set of emission factors, ERTAC Rail
requested each Class I company to provide a description of their line-haul fleet mix based on
Tier rating, which each company provided under a confidentiality agreement. An engine's Tier
level is based on the year the engine was built and determines allowable emission limits (Table
3).
A-4
-------
Table 3. EPA line-haul locomotive Emission Factors by Tier, 1997 standards (grams/gal).
Note that the new standards released in 2008 did not apply to fleets in the year 2008.11
PM10
HC
NOx
CO
Uncontrolled (pre-1973)
6.656
9.984
270.4
26.624
Tier 0(1973-2001)
6.656
9.984
178.88
26.624
Tier 1 (2002-2004)
6.656
9.776
139.36
26.624
Tier 2 (2005 + )
3.744
5.408
102.96
26.624
Based on values in EPA Technical Highlights: Emission Factors for Locomotives, EPA Office of
Transportation and Air Quality, EPA-420-F-09-025, April 2009.
Weighted Emission Factors (EF) per pollutant for each gallon of fuel used (gm/gal or lbs/gal)
were calculated for each Class I railroad fleet based on its fraction of line-haul locomotives at
each regulated Tier level (Eqn 1; Table 3).
EFm = ^(EFlT* fm)
Equation 1
EFiRR
EFiT
/TRR
= Weighted Emission Factor for pollutant i for Class I railroad RR (gm/gal).
= Emission Factor for pollutant i for locomotives in Tier T (gm/gal) (Table 3).
There were 4 Tiers of locomotives in the 2008 fleets.
= Fraction of railroad RR fleet in Tier T.
While engine emissions are variable within Tier categories, this approach likely provides better
regional estimates than uniformly applying the nationwide average emission factors. This
approach likely provides conservative emission estimates as locomotive engines are certified to
meet or exceed the emissions standard for each Tier, although emission levels may increase after
certification.
Other emission factors are not engine specific. For locomotives, PM2.5 is assumed to be 97% of
PMio u, and emission factors applied for SO2 and NH3 are 1.88 g/gal 11 and 83.3 mg/gal12
respectively. Greenhouse gases are estimated using emission factors shown in Table 4.
Table 4. EPA greenhouse gas emission factors for locomotive diesel fuel (grams/gal).13
C02
N20
ch4
Locomotive diesel
1.015E4
0.26
0.80
A Railroad Fuel Consumption Index (RFCI) was also calculated for each Class I railroad using
their system-wide line-haul fuel consumption (FC) and gross ton-mile (GTM) data reported in
their annual R-l reports submitted to the Surface Transportation Board7 (Eqn 2). This value
represents the average number of GTM produced per gallon of diesel fuel used over their system
in a year, and varies between railroad carriers depending on factors such as fleet mix, system
A-5
-------
terrain, speeds, loading/weight of cargo, train type (e.g., intermodal, unit, and manifest), and
operating practices. (Table 2).
GTMm ^ n
RFCIm = Equation 2
FCrr
RFCIrr = Railroad Fuel Consumption Index (gross ton-miles/gal) per Class I railroad
(RR). GTMrr = Gross Ton-Miles (GTM), annual system-wide gross ton miles of freight
transported per RR. (R-l Report Schedule 755, Line 104)
FCrr = Annual system-wide fuel consumption by line-haul and work trains per RR
(gal) (R-l Report Schedule 750, Lines 1 and 6).
2. Calculate Emissions per Link.
Emissions of pollutant i per link L (En) are then calculated by multiplying the gallons of diesel
fuel consumed by each Class I railroad on the link by that railroad's weighted Emission Factor
for the pollutant, and summed over all railroads operating on the link (Eqn 3). This approach
splits the activity on each link (represented by MGT) evenly between all railroads operating on
the link. Note that the weighted Emission Factors are converted to tons/gal for these
calculations, and that variables with units in tons may represent tons of freight hauled (MGT,
RFCI) or tons of pollutants (EF, E).
ElL = Z"
'MGTL *10M
. ^
RFCh,
'EF,
iRR
Equation 3
EiL = Emissions of pollutant i per link L (tons/year).
N = Number of Class I railroads operating on link L.
MGTl = Millions of Gross Tons hauled per link per year from the FRA database
(106 tons/yr)9.
lL = Link length from the FRA database (miles).
EFiRR = Weighted Emission Factor for pollutant i per railroad RR (Eqn 1; tons/gal).
RFCIrr = Railroad Fuel Consumption Index per railroad RR (Eqn 2; gross
ton-miles/gal).
Note that approximately 36% of Class I route miles in the United States are shared by more than
one Class I carrier, a fraction that drops to 26% when neglecting track only shared between one
Class I freight railroad and Amtrak. Accurately apportioning the specific fractions of tonnage
(MGT) per carrier per link was considered, but after comparing likely worst-case areas, the
difficultly of merging carrier-specific MGT with the aggregated FRA MGT dataset was
considered too great considering the potential gain in accuracy. Where warranted, MGT data
may be apportioned more accurately in the future.
A-6
-------
Limitations, Conclusions, and Future Work
Rail-related emissions can be important components of emissions inventories used to support
effective air quality management practices, at local, state, regional, and national levels. This line-
haul inventory, as well as the companion Class I railyard inventory and Class II/III inventory,
greatly improve our estimates of rail-related emissions. However, a systematic study of
variability and uncertainty in line-haul locomotive emissions and activity, by fleets, locations,
and through time, would give valuable information for identifying how to best improve this
inventory as well provide an indication of how representative the inventory may be. An
uncertainty study on the data used for this inventory, including the R-l reported fuel use and the
confidential link-level tonnage data, would also help in evaluating the quality of this inventory.
Localized studies should also examine how shared tracks are apportioned between multiple
carriers.
Early ERTAC Rail discussions concluded that link-level tonnage was the most important data to
obtain, while other variables such as track grade and track speed could not be addressed at this
time. ERTAC Rail calculated railroad-specific fleet-averaged emission factors rather than
applying the estimated national average; however, it is recognized that emissions from individual
engines are highly variable even within Tier categories depending on variables such as the
specific locomotive model, operation cycle, and conditions of operation. Future evaluation of
emission variability within Tiers and between certain types of operation and locations would also
be valuable.
Emissions inventory preparation guidance from the U.S. EPA describes locomotive activity as
relatively constant throughout the year (e.g. no daily, weekly, or seasonal variability); however,
actual activity levels do vary seasonally and annual averaging may dilute or exaggerate
concentrations during pollution episodes. ERTAC Rail and the Class I railroad community had
some discussions addressing if incorporating more specific fleet mix or monthly or seasonal
variation may be worthwhile, and these topics should be looked into further.
Finally, it is important to reiterate that the link-level MGT data maintained by the FRA is
proprietary and can only be released to agencies/groups outside the FRA with the express
permission of each Class I railroad. It is possible that one or more Class I railroads could
withhold permission for access, but data for specialized studies may be provided if requested.
This database can also be improved by better distinguishing between haulage and trackage rights,
and by apportioning tonnage hauled on links to specific carriers.
We would like to thank the Class I Railroads and their representatives for their assistance and
support in the development of this inventory.
References
1. Eastern Research Group (ERG) for E.H. Pechan & Associates, Inc., "Documentation for
Aircraft, Commercial Marine Vessel, Locomotive, and Other Nonroad Components of
the National Emissions Inventory, Volume I - Methodology"; EPA Contract No.: 68-D-
02-063. Prepared for the US EPA Emissions, Monitoring and Analysis Division, Sept.
A-7
-------
30, 2005.
ftp://ftp.epa.gov/EmisInventorv/2002finalnei/documentation/mobile/2002nei mobile no
nroad methods.pdf. Related documents at
ftp://ftp.epa.gov/EmisInventory/2002finalnei/documentation/mobile/2002nei_mobile_no
nroad_train.pdf
2. Sierra Research, Inc., "Revised Inventory Guidance For Locomotive Emissions"; Report
No. SR2004-06-01. Prepared for Southeastern States Air Resource Managers
(SESARM), June 2004. http://www.metro4-sesarm.org/pubs/railroad/FinalGuidance.pdf
3. Sierra Research, Inc., "Research Project: Development of Railroad Emission Inventory
Methodologies"; Report No. SR2004-06-02. Prepared for Southeastern States Air
Resource Managers (SESARM), June 2004. http://www.metro4-
sesarm.org/pubs/railroad/FinalMethodologies.pdf
4. Environ, "Draft LADCO 2005 Locomotive Emissions". Prepared for Lake Michigan Air
Director Consortium, Feb 2007.
http://www.ladco.org/reports/technical_support_document/references/ladco_2005_locom
otive_emissions.021406.pdf
5. Southern Research Institute, "NYSERDA Clean Diesel Technology: Non-Road Field
Demonstration Program, Development of the 2002 Locomotive Survey for New York
State"; Agreement Number 8958. Prepared for the New York State Energy Research And
Development Authority (NYSERDA), Feb. 09, 2007.
http://www.nvserda.org/publications/LocomotiveSurvevReportwithAppendices.pdf
6. M. Bergin; M. Harrell; J. McDill; M. Janssen; L. Driver; R. Fronczak; R. Nath,; and D.
Seep. "ERTAC Rail: A Collaborative Effort in Building a Railroad-Related Emissions
Inventory Between Eastern States Air Protection Agencies and Participation with the
Railroad Industry," 18th Annual International Emission Inventory Conference.
Baltimore, MD. April 14 - 17, 2009. Paper and presentation available at:
http://www.epa.gov/ttn/chief/conference/eil8/session6/bergin.pdf
7. Surface Transportation Board R-l Reports, available at:
http://www.stb.dot.gov/stb/industry/econ reports.html.
8. US DOT Bureau of Transportation Statistics' 2008 National Transportation Atlas
Database.
9. Confidential database was provided with assistance from Raquel Wright of the Federal
Railroad Administration. Similar public data providing ranges of tonnage hauled rather
than link-level tonnage is available from the Bureau of Transportation Statistics in the
NT AD 2009 shapefile data (data is representative for the year 2007):
http://www.bts.gov/publications/national transportation atlas database/2009 .
10. Freight Railroads in the United States 2008. Association of American Railroads.
Available at:
http://www.aar.0rg/~/media/AAR/InC0ngress RailroadsStates/2008unitedstates.ashx .
11. EPA Technical Highlights: Emission Factors for Locomotives, EPA Office of
Transportation and Air Quality, EPA-420-F-09-025, April 2009.
http://www.epa.gov/otaq/regs/nonroad/locomotv/420fD9025.pdf
A-8
-------
12. Estimating Ammonia Emissions From Anthropogenic Nonagricultural Sources - Draft
Final Report by E.H. Pechan & Assoc. April 2004. Prepared for EPA/STAPPA-
ALAPCO Emission Inventory Improvement Program. Supported by personal
communication (5/6/2010) with Craig Harvey, US EPA, OTAQ, and Robert Wooten, NC
DENR. http://www.epa.gov/ttnchiel/eiip/techreport/volume03/eiip areasourcesnh3.pdf
13. U.S. Environmental Protection Agency. Inventory of U.S. Greenhouse Gas Emissions and Sinks:
1990-2005, EPA 430-R-07-002, Annex 3.2, (April 2007), web site:
htto://www.epa.gov/climatechange/emissions/usinventorvreport.html.
Appendix A: ERTAC Rail Data Workgroup
REPRESENTATIVE
ORGANIZATION
Matt Harrell
IL EPA
Michelle Bergin (Co-Chair) and Byeong Kim
GAEPD
Mark Janssen (Co-Chair)
LADCO
Julie McDill and Patrick Davis
MARAMA
Laurel Driver
US EPA OAQPS
Robert Fronczak
AAR
Steven Sullivan
ASLRRA
Rick Nath
CSX
David Seep and Lyle Staley
BNSF
Ken Roberge
CPR
Carl Akins and Peter Conlon
KCS
Erika Akkerman
CN
M. John Germer
UP
Brent Mason and Richard Russell
NS
Joanne Maxwell
Amtrak
A-9
-------
Appendix B: Source and Year of Data Utilized for Each Inventory
Data
Year
Source
Class I Line-Haul
Annual Line-Haul Fuel Use
and Gross Ton-Miles
2007
STB R-l Reports (CP data for
D&H is for 2008.)
Line-haul fleet mix for
emission factors
2008
Each Class I railroad
Link-level tonnage
2007
FRA confidential database
Class I Railyards (Switcher Locomotives)
Annual Switcher Fuel Use
2008
R-l Reports
Switcher fleet mix for
emission factors
2008
Each Class I railroad
Link-level tonnage or
Density Code (for activity
estimate)
2007
FRA confidential database
Class I
and III Locomotives
Annual Total Fuel Use
2008
ASLRRA Annual Report (2008)
Track length and railroad
2008
ASLRRA Annual Report (2008)
Estimated fleet mix for
emission factors
Discussions with ASLRRA and
Class II and III representatives.
A-10
-------
Appendix B
ERTAC Class II/III Line Haul Documentation
-------
DRAFT
ERTAC - Class 2/3 Shapefile Documentation
13 Jul 2009
Introduction
This document outlines the methods and procedures used to compile a shapefile representing the
links in the FRA 1:100,000 railroad dataset that are owned or operated by Class II and III
railroad companies. It is important to note that there is a considerable amount of overlap
between the Class II's and Ill's and the Class I and passenger railroads. Class II's and Ill's can
operate on Class I or passenger rail links and vice versa. Although the final shapefile
specifically represents Class II and III links, there are many Class I and passenger railroads
represented as well.
Procedure
1. Started with all proprietary FRA links where "NET = 'M' and "STCNTYFIPS" <> ' '
(this definition query selects all active mainline links located within the United
States).
2. Ran 12 queries, one for each ownership and trackage rights field, to select all links
not associated with a Class I freight railroad or Amtrak and not containing a null
value (e.g., "RROWNER1" <> 'AMTK' AND "RROWNER1" <> 'BNSF' AND
"RROWNER1" <> 'CN' AND "RROWNER1" <> 'CPRS' AND "RROWNER1" <>
'CSXT' AND "RROWNER1" <> 'KCS' AND "RROWNER1" <> 'NS' AND
"RROWNER1" <> 'UP' AND "RROWNER1" <> ''). The first query was setup as a
new selection. Each of the 11 subsequent queries were setup to add records to initial
set of records. 26,261 links were selected and exported to a new shapefile.
3. Due to the multitude of railroad codes used to represent commuter rail operations
across the country, additional processing was required to remove any links that were
not operated by a Class II or III freight railroad. Each commuter railroad was queried
out of the new shapefile and the links analyzed to eliminate all links where no Class II
or III operations were occurring. The following commuter rail operations were
evaluated: NJT (New Jersey Transit), MNCW (Metro-North Commuter Railroad), LI
(Long Island Railroad), CDOT (Connecticut DOT), MBTA (Massachusetts Bay
Transportation Authority), SEPA (Southeastern Pennsylvania Transportation
Authority), MARC (Maryland Area Rail Commuter), VRE (Virginia Railway
Express), MTRA (Northeastern Illinois Regional Commuter Railroad), CSS
(Northern Indiana Commuter Transportation District), DART (Dallas Area Rapid
Transit), SCRA (Southern California Regional Rail Authority - including also SCAX,
LACM, LAPT, and LATC), TCRA (South Florida Regional Transportation
Authority), PJPB (Caltrain), and ACE (Altamont Commuter Express).
B-l
-------
Approximately 1581 links were identified with no Class II or III operations and were
deleted from the Class 2/3 shapefile.
4. The remaining Class II and III links were then compared to the regional maps
contained in the July-August issue of The Official Railway Guide to assess the
completeness of the Class 2/3 shapefile. Six specific edits were made to the shapefile
to correct the most glaring errors: 1) BMLP links deleted (Black Mesa & Lake
Powell, an electric coal hauling railway in Arizona); 2) DSNG links deleted (Durango
& Silverton steam tourist railroad in Colorado; 3) CIC haulage rights links on CN
from Chicago to Omaha deleted; 4) DMIR links deleted (Duluth, Missabe & Iron
Range, now owned and operated by CN in Minnesota; 5) EVWR's ex-CSXT links
coded from Evansville, IN to Okawville, IL (Evansville Western Railroad); 6) INRD
ex-CP links coded from Chicago, IL to Louisville, IN (Indiana Rail Road).
5. During the course of reviewing the FRA dataset, 555 "active" links were found to
have no ownership or trackage rights codes. 1005 links have no codes listed in the 3
ownership fields. In most cases these links are very short and scattered across the
country. Only the links representing the EVWR and INRD spanned large distances
and were fixed. The other problem links were deemed to be insignificant. A listing of
these links will be provided back to the FRA to assist with their coding in 1:100K
railway shapefile.
B-2
-------
Appendix C
ERTAC Rail Yard Documentation
-------
DRAFT
ERTAC Rail Emissions Inventory
Part 2: Class I Railyard Switcher Locomotives
Michelle Bergin, GA Environmental Protection Division
Matthew Harrell, IL Environmental Protection Agency
Mark Janssen, Lake Michigan Air Directors Consortium
Acknowledgments: Robert Fronczak, Association of American Railroads
Laurel Driver, US EPA, Office of Air Quality Planning and Support
Byeong Kim, GA Environmental Protection Division
Introduction
Air protection agencies from twenty-seven states, coordinated through the Eastern Regional
Technical Advisory Committee (ERTAC) and headed by the Lake Michigan Air Directors
Consortium (LADCO), identified a need to better quantify and characterize rail-related emissions
inventories. Traditional locomotives largely utilize diesel engines, resulting in emissions of
NOx, diesel PM, hydrocarbons, greenhouse gases, and other pollutants. These emissions are
sometimes concentrated in areas exceeding National Ambient Air Quality Standards. No
cohesive nationwide railroad emission estimates are known to have been made previously.
Inventory development methods for locomotive emissions estimates vary from state to state and,
in general, lack the spatial or temporal resolution needed to support air quality modeling and
planning 1_5.
The ERTAC Rail Subcommittee (ERTAC Rail) was established with active representatives from
twelve member states, three regional planning offices, and the US EPA. The subcommittee's
goals are to (1) standardize agencies' inventory development methods through a collaborative
effort, (2) improve the quality of data received and the resulting emission inventories, and (3)
reduce the administrative burden on railroad companies of providing data. With support from
the Rail industry and assistance from the ERTAC Rail Data Workgroup (Appendix), ERTAC
Rail has developed 3 inventories of locomotive emissions; from Class I line-haul, Shortline and
Regional Railroads, and Class I railyard switchers, for the year 2008 (Table 1).
The Surface Transportation Board (STB) defines Class I Railroads as having had minimum
carrier operating revenues of $401.4 million (USD) in 2008. There are 8 Class I Railroads
operating in the United States (Table 2), about 12 Regional Railroads (Class II), and
approximately 530 Class III Railroads (Shortlines). While categorized as a Class I Railroad,
Amtrak was excluded from these inventories because of significant differences in equipment and
operation characteristics. Line-haul locomotives travel long distances (e.g. between cities) while
switcher locomotives largely operate in railyards, splitting and joining rail cars with varying
destinations. Passenger and Commuter Rail (including Amtrak), industrial locomotives, and
associated non-locomotive equipment are not included in these inventories.
C-l
-------
Table 1. Summary of ERTAC Rail Inventories: U.S. Locomotive Emissions and Fuel Use for
either 2007 or 2008*.
Fuel Use**
Emissions
(tons/yr)
(gal/yr)
NOx
pm2,
HC
S02
CO
nh3
Class I*** line-
3,770,914,002
754,443
23,439
37,941
7,836
110,969
347
haul
Class I switcher
300,492,223
73,741
2,024
4,824
619
9,152
27
Class II and III
157,800,000
51,367
1,163
1,897
357
5,058
16
*See Appendix B
"or a description o
? the year and source of data utilized for each inventory.
"Locomotive grade diesel
* "Excluding Amtrak and including work train fuel use
Table 2. Class I Railroads and Reported Locomotive Fuel Use7.
Class I Railroads*
R-l Reported Locomotive Fuel
Use (gal/yr)
Line-Haul
(2007)**
Switcher
(2008)
BNSF
1,393,874,954
52,497,057
Canadian National
93,830,751
12,290,022
Canadian Pacific***
50,320,233
4,594,067
CSX
514,687,186
53,717,674
Kansas City Southern
69,787,071
1,816,759
Norfolk Southern
463,267,278
32,317,375
Union Pacific
1,185,146,529
143,470,336
Total
3,770,914,002
300,492,223
* Excluding Amtrak
** Includes work trains
*** CP's line-haul fuel use values include 2008 data (rather than 2007) for
their Delaware and Hudson subsidiary.
This paper documents the data sources and methodologies used for calculating the Class I
switcher ("Railyard") inventory. Information on ERTAC Rail, Railroad participation, the Rail
industry, and effects of rail on air quality are available elsewhere6.
Method
Switcher locomotives are expected to be the single largest source of air emissions in railyards.
Therefore, as a starting point for a comprehensive railyard inventory, a Class I switcher emission
inventory was developed. It is assumed that estimates for yards of interest, associated equipment
and activity, and smaller railroads could be refined later.
While ERTAC Rail represents states east of the Mississippi River, the railroad companies
specified they wanted this effort to result in a consistent nationwide inventory. ERTAC Rail
agreed to calculate emissions for all states when the data was available and when additional
C-2
-------
significant effort was not required. Because both the dataset of railyards and switcher fuel use
was nationwide in scope, the resulting initial railyard inventory is a nationwide, 'top-down'
derivation. However, railroad companies may have different levels and quality of data available,
and may have interpreted some data requests differently. Also, states are requested to update
yards they have detailed information on when possible, and a few states (i.e. California) have
unique railroad operations and equipment. Therefore, data for some areas will be more accurate
than for others, and locally-derived inventories may be more accurate.
This documentation describes development of the initial top-down inventory, which consisted of
three main activities:
1. Locate Class I Railyards
2. Select/Calculate Emission Factors
3. Estimate Locomotive Activity
4. Improve Estimates
1. Locate Class I Railyards.
Identification and correct placement of railyards was an important first step, requiring a
comprehensive electronic dataset. A confidential database was obtained from the Federal
Railroad Administration (FRA) with permission from the Class I Railroads (FRA database). A
similar public database compiled by the Bureau of Transportation Statistics is also available7 .
Data from this source will not match the confidential data exactly, but will be very similar. The
FRA database has rail links (track lengths) individually identified as parts of specific railyards.
While there may be discrepancies in how each railroad defined railyard links, this dataset
appears to identify most Class I railyards in the U.S., and shows a high density of yards in the
eastern states (Figure 1). The database gives length, up to 3 owners and 3 operators, and a
Federal Density Code (explained below) for each railyard link.
Railyards
NOX [TPY]
0.0 -40.9
41.0-112.0
4 112.7-241.5
241.6 - 535,0
O 530.0 - 1301.5
NOx emissions in ER~AC Railyard Emisson Inventory (craft ver. 1)
Created by Byeong-Uk Kim at GA EPD
Dala provided by Michelle Bergin at GA EPD
Figure 1. Class I Railyards in the United States and estimates of Annual NOx
emissions from switcher locomotives (tons/yr in 2008).
C-3
-------
2, Select/Calculate Emission Factors.
The EPA provides annual default emission factors based on characteristic operating cycles ('duty
cycles') and the estimated nationwide fleet mix for both switcher and line-haul locomotives.
However, switcher fleet mix is not uniform from company to company and, as can be seen in
Figure 2, Class I railroad activity is highly regional.
As an alternative approach, ERTAC Rail requested each Class I rail company to provide a
description of their switcher fleet mix based on Tier rating, which each company provided under
a confidentiality agreement. An engine's Tier determines allowable emission limits based on the
year the engine was built (Table 3). While engine emissions are variable within Tier categories,
this estimate likely provides a better regional estimate than the nationwide average. The
company-specific systemwide fleet mix was used to calculate weighted average emissions
factors for switchers operated by each Class I railroad.
Figure 2. Class I Railroad Territories in the United States.
PM10
HC
NOx
CO
Uncontrolled (pre-1973)
6.688
15.352
264.48
27.816
Tier 0(1973-2001)
6.688
15.352
191.52
27.816
Tier 1 (2002-2004)
6.536
15.352
150.48
27.816
Tier 2 (2005 + )
2.888
7.752
110.96
27.816
Listed years apply to the year the engine was built. Table based on values from8. Note that the new standards
released in 2008 did not apply to existing fleets in the year 2008.
C-4
Other Owners
National Network MS
AH Other Rail KCS
BNSF CN
-------
For locomotives, PM2.5 is assumed to be 97% of PM10 , and emission factors for SO2 and NH3
are 1.88 g/gal and 83.3 mg/gal respectively (add cites). Greenhouse gases are also estimated
using emission factors shown in Table 4.
Table 4. EPA greenhouse
gas emission factors for locomotive diesel fuel (grams/gal).
C02
N20
ch4
Locomotive diesel
1.015E4
0.26
0.80
Source: U.S. Environmental Protection Agency. Inventory of U.S. Greenhouse Gas Emissions and
Sinks: 1990-2005, EPA 430-R-07-002, Annex 3.2, (April 2007), web site:
http://www.epa.gov/climatechange/emissions/usinventoryreport.html
These emission factors are based on a characteristic duty cycle for switchers which assumes
operation over 24-hour per day 365 days per year. An evaluation of the effect of variability in
railyards and switching duties on emissions would be useful for future inventories.
3. Estimate Locomotive Activity.
Class I railroads report total annual switcher locomotive fuel use to the STB, which is reported in
publicly available 'R-l' reports (Table 2). There may be inconsistencies between railroads in
how fuel use is estimated to be apportioned between line-haul and switcher locomotive use, and
possibly in the total locomotive fuel use, so these values may be adjusted in the future.
However, the use of these values provides a starting point for estimating total U.S. Class I
locomotive-related emissions segregated by Class I carrier. The R-l report was used by ERTAC
for both the line-haul and switcher locomotive emissions inventories.
The next step for inventory development is to allocate switcher fuel use to each railyard. Two
methods were applied, one that relies on publicly available line-haul activity (the 'Dencode'
method), and the other using confidential line-haul activity (the 'MGT' method.) At this time,
Norfolk Southern and Kansas City Southern have provided input for use of the MGT method and
the Dencode method is applied for the other five railroads.
The Dencode Method - Publicly available data
Each link in both the publicly available BTS database and the confidential FRA database has a
'Federal Density Code' (Dencode) ranging from 1 to 7 assigned based on the cumulative annual
freight tonnage hauled on the link (track). Total Switcher Fuel Use in each railyard Y (SFUy) is
estimated as follows:
First the Switcher Activity Indicator per yard (SAIY) is estimated by multiplying the average
dencode of the links identified as part of the same railyard by the sum of the length of the links
for that railyard (Eqn 1).
SAIy= £(/,.* l'l)CnY) Equation 1
SAIY = Switcher Activity Indicator in Railyard Y
C-5
-------
nY = number of links identified as part of railyard Y
1y = length of link n in miles
FDCn = Federal Density Code (1 to 7) of link n
Next, this value is then weighted (SAIY') based on an ownership factor (OF) set between 0 and 1.
The OF depends on the number of owners listed for each railyard: if there is one owner the OF
is set to 1, if there are two owners the primary owner is set to 0.8 and the secondary is 0.2, and if
there are 3 owners the primary is 0.6, the secondary is 0.2, and the tertiary is 0.1.
Next, the SAIY' of all railyards belonging to a Class I railroad (RR) were summed, and the
fraction of the railroads total SAI associated with each railyard was multiplied by the railroads
total annual switcher fuel use reported in the R-l (TFUrr), resulting in the total Switcher Fuel
Use for each railyard Y (Eqn 2).
SFUy = Switcher Fuel Use at railyard Y
Finally, the SFUy is multiplied by the emission factors described in the previous section to
obtain annual switcher emissions at each railyard.
The MGT Method - Confidential data
Two railroads, Norfolk Southern and Kansas City Southern, provided confidential link-level
tonnage information and weighting factors to correct skewed estimates to improve estimated
switcher activity at important yards. Other railroads may also allow the use of this technique for
their inventories in the future.
The MGT Method also uses the FRA database for railyard identification and link lengths.
However, rather than using the average dencode per link, confidential annual gross tonnage
(MGT) hauled per link in the railyard was used to calculate the railyard switcher activity (SAIY).
This is calculated by replacing FDCn in Equation 2 with link-specific tonnage MGTn (Equation
SAIy' = OFy* SAIy
Equation 2
Equation 3
RR
4)
SAIy= IX
Equation 4
SAIy = Switcher Activity Indicator in Railyard Y
nY = number of links identified as part of railyard Y
C-6
-------
1y = length of link n in miles
MGTnY = million gross tons on link n
This method provides a more refined comparison between railyards than the use of the 7-
category dencodes; however, is more susceptible to errors for yards where tonnage is not
correlated to switching activity. For example, a yard with large coal trains pulling through used
for crews to change over would be assigned an overly high level of emissions for switching
activity. To account for this, a discretionary Switching Activity Factor (SAF) was introduced to
allow railroads to roughly weight yards with clearly higher or lower levels of switching activity
than what results from the mathematical allocation. Therefore, SAIY is weighted based on both
the ownership factor (OF) as well as the SAF (Equation 5). For example, a yard used for crew
changes and not switching may have an SAF of 0, while a yard at a major interchange between
cities may have an SAF of 3.
SAIy' = OFy*SAFy* SAIy Equation 5
Again, the SAIY' of all railyards belonging to a Class I railroad (RR) are summed, and the
fraction of the railroads total SAI associated with each railyard was multiplied by the railroads
total annual switcher fuel use reported in the R-l (TFUrr), resulting in the total Switcher Fuel
Use for each railyard Y (Eqn 6).
SAT '
SFUy = ^~ * TFIJ^ Equation 6
RR
While the SAF allows estimates of yard-specfic emissions to be adjusted, the total level of
emissions for each railroad, which is based on systemwide fuel use and systemwide emission
factors, remains unchanged. The MGT method SFUy is also later multiplied by the emission
factors described in the previous section to obtain annual switcher emissions at each railyard.
4. Improve estimates.
In addition to the Switching Activity Factor described above, direct input was also used to
improve emission estimates for important railyards. Each Class I railroad provided an estimate
of annual average switcher fuel use (generally much lower than the EPA default of 82,490
gal/yr) as well as the name, location, and number of operating switchers for railyards with 8 or
more switchers operating in ozone or PM2.5 nonattainment areas. This data was used to
overwrite the dencode or MGT derived emissions estimates for those railyards.
The difference in estimated fuel use for those railyards was re-allocated (added or removed)
between the remaining railyards belonging to that Class I railroad. It is important to note that
there are some discrepancies in how this data was reported for the large railyards by each
railroad. For example, some railroads reported all switchers located at a railyard while others
reported 'full time equivalent' switchers, meaning the number of switchers normalized to a full
working cycle (24-hours per day year-round.) This process should be standardized for future
inventory versions.
C-7
-------
States also have the option of updating specific railyard emissions estimates. Because this
inventory is derived 'top-down', local studies and familiarity with specific railyards is expected
to provide better estimates, which can be used to adjust this inventory. Care must be taken to
ensure the other railyard estimates are adjusted to account for increases or decreases in estimated
fuel use per yard.
Limitations, Conclusions, and Future Work
What this ERTAC Rail railyard inventory does well is provide a comprehensive overview of
where railyards are, who owns them, and gives a geographical allocation of switcher emissions
bounded by what is reported as nationwide switcher fuel usage by the Class I railroads. These
sources can be important for air quality management in nonattainment areas, as well as in
regional analysis and for future transportation planning. This inventory will be useful for
regional and some local modeling, helps identify where railyards need to be better characterized,
and provides a strong foundation for future development of a meaningful nationwide Class I
switcher emissions inventory.
There are important uncertainties associated with estimates from this method, including, but not
limited to, the use of tonnage hauled as an indicator of the amount of switching activity, and, for
a few of the railroads, how the amount of switcher fuel use was determined to be reported in the
R-l. The R-l reported values are currently under examination.
There is also likely significant variability in actual switching duty-cycles and, potentially, in the
number of switchers operating at some railyards at different times of the year. 'Road-switching',
or the use of what are considered switching locomotives to move between nearby yards, should
be addressed in either this or the ERTAC line-haul inventory.
It must be noted that freight-related rail activity is not always routine and no annual emissions
inventory will ever be able to capture the innate variability of the source. However, as other
large emission sources are reduced, and if rail activity increases as expected, it is important to
include our best estimates of these sources in air quality analysis. In the future, on-line data
loggers and other tracking technologies, combined with ambient studies and detailed modeling,
will hopefully provide more insight to the emissions of locomotives and other railyard sources.
References
1. E.H. Pechan & Associates, Inc., "Documentation for Aircraft, Commercial Marine
Vessel, Locomotive, and Other Nonroad Components of the National Emissions
Inventory, Volume I - Methodology"; EPA Contract No.: 68-D-02-063. Prepared for the
US EPA Emissions, Monitoring and Analysis Division, Sept. 30, 2005.
ftp://ftp.epa.gov/EmisInventorv/2002finalnei/documentation/mobile/20Q2nei mobile no
nroad methods.pdf. Related documents at
ftp://ftp.epa.gov/EmisInventory/2002finalnei/documentation/mobile/2002nei_mobile_no
nroad_train.pdf
C-8
-------
2. Sierra Research, Inc., "Revised Inventory Guidance For Locomotive Emissions"; Report
No. SR2004-06-01. Prepared for Southeastern States Air Resource Managers
(SESARM), June 2004. http://www.metro4-sesarm.org/pubs/railroad/FinalGuidance.pdf
3. Sierra Research, Inc., "Research Project: Development of Railroad Emission Inventory
Methodologies"; Report No. SR2004-06-02. Prepared for Southeastern States Air
Resource Managers (SESARM), June 2004. http://www.metro4-
sesarm.org/pubs/railroad/FinalMethodologies.pdf
14. Environ, "Draft LADCO 2005 Locomotive Emissions". Prepared for Lake Michigan Air
Director Consortium, Feb 2007.
http://www.ladco.org/reports/technical_support_document/references/ladco_2005_locom
otive_emissions.021406.pdf
15. Southern Research Institute, "NYSERDA Clean Diesel Technology: Non-Road Field
Demonstration Program, Development of the 2002 Locomotive Survey for New York
State"; Agreement Number 8958. Prepared for the New York State Energy Research And
Development Authority (NYSERDA), Feb. 09, 2007.
http://www.nvserda.org/publications/LocomotiveSurvevReportwithAppendices.pdf
16. M. Bergin; M. Harrell; J. McDill; M. Janssen; L. Driver; R. Fronczak; R. Nath,; and D.
Seep. "ERTAC Rail: A Collaborative Effort in Building a Railroad-Related Emissions
Inventory Between Eastern States Air Protection Agencies and Participation with the
Railroad Industry," 18th Annual International Emission Inventory Conference.
Baltimore, MD. April 14 - 17, 2009. Paper and presentation available at:
http://www.epa.gov/ttn/chief/conference/eil8/session6/bergin.pdf
17. Confidential database was provided with assistance from Raquel Wright of the Federal
Railroad Administration. Similar public data is available from the Bureau of
Transportation Statistics, in the NTAD 2009 shapefile data (data is representative for the
year 2007): http://www.bts.gov/publications/national transportation atlas database/2009
18. EPA Technical Highlights: Emission Factors for Locomotives, EPA Office of
Transportation and Air Quality, EPA-420-F-09-025, April 2009.
http://www.epa.gov/otaq/regs/nonroad/locomotv/420fD9025.pdf
Appendix: ERTAC Rail Data Workgroup
REPRESENTATIVE
ORGANIZATION
Matt Harrell
IL EPA
Michelle Bergin (Co-Chair) and Byeong Kim
GAEPD
Mark Janssen (Co-Chair)
LADCO
Julie McDill and Patrick Davis
MARAMA
Laurel Driver
US EPA OAQPS
Robert Fronczak
AAR
Steven Sullivan
ASLRRA
Rick Nath
CSX
C-9
-------
David Seep and Lyle Staley
BNSF
Ken Roberge
CPR
Carl Akins and Peter Conlon
KCS
Erika Akkerman
CN
M. John Germer
UP
Brent Mason and Richard Russell
NS
Joanne Maxwell
Amtrak
C-10
-------
United States Office of Air Quality Planning and Standards Publication No. EPA-454/B-20-016
Environmental Protection Air Quality Assessment Division May 2011
Agency Research Triangle Park, NC
------- |