Control of Hazardous Air Pollutants
from Mobile Sources
Regulatory Impact Analysis
United States
Environmental Protection
Agency
-------
Control of Hazardous Air Pollutants
from Mobile Sources
Regulatory Impact Analysis
Assessment and Standards Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency
v>EPA
United States EPA420-R-07-002
Environmental Protection February 2007
Agency
-------
Table of Contents
Executive Summary ES-1
Chapter 1: Mobile Source Air Toxics Health Information 1-1
Chapter 2: Emission Inventories 2-1
Chapter 3: Air Quality and Resulting Health and Welfare Effects of Air Pollution from Mobile
Sources 3-1
Chapter 4: Industry Characterization 4-1
Chapter 5: Vehicle Technological Feasibility 5-1
Chapter 6: Feasibility of the Benzene Control Program 6-1
Chapter 7: Portable Fuel Container Feasibility and Test Procedures 7-1
Chapter 8: Impact of New Requirements on Vehicle Costs 8-1
Chapter 9: Cost of the Gasoline Benzene Standard and Other Control Options
Considered 9-1
Chapter 10: Portable Fuel Container Costs 10-1
Chapter 11: Cost per Ton of Emissions Reduced 11-1
Chapter 12: Cost-Benefit Analysis 12-1
Chapter 13: Economic Impact Analysis 13-1
Chapter 14: Small-Business Flexibility Analysis 14-1
-------
List of Acronyms
ug/kg-day Micrograms per kilogram day
ug/m3 Microgram per cubic meter
AAM Alliance of Automobile Manufacturers
ABT Averaging, Banking, and Trading
ACS American Cancer Society
AEO Annual Energy Outlook (an EIA publication)
AGO Atmospheric Gas Oil (a refinery stream)
ALTP Absolute Level Trigger Point
AMS/EPA Regulatory Model AERMOD
ANS Alaska North Slope
APHEA Air Pollution and Health: A European Approach
API American Petroleum Institute
AQCD Air Quality Criteria Document
AQMTSD Air Quality Modeling Technical Support Document
ARB (California) Air Resources Board
ASPEN Assessment System for Population Exposure Nationwide
ASTM American Society of Testing Materials
ATB Atmospheric Tower Bottoms (a refinery stream)
ATSDR Agency for Toxic Substances and Disease Registry
ATV All-Terrain Vehicles
BBL Barrel
BC Black Carbon
BenMAP Environmental Benefits Mapping and Analysis Program
BPCD Barrels Per Calendar Day
BPD Barrels Per Day
BTEX Benzene, Toluene, Ethylbenzene, and Xylene isomers
BTX Benzene, Toluene, and Xylene isomers
BZ Benzene
C Celsius
C6 or C6 A hydrocarbon molecule with a specified number of carbon atoms, in this case 6 carbons
CA California
CAA Clean Air Act
CAIR Clean Air Interstate Rule
California EPA California Environmental Protection Agency
CAMR Clean Air Mercury Rule
CAMx Comprehensive Air Quality Model with Extension
CAND Clean Air Nonroad Diesel
CARD California Air Resources Board
CASAC Clean Air Science Advisory Committee
CAVR Clean Air Visibility Rule
CB Chronic Bronchitis
CD Criteria Document
CDC Center for Disease Control
CE10 Gasoline with 10 percent ethanol content
CEA Cost-Effectiveness Analysis
CFEIS Certification and Fuel Economy Information System
CFR Code of Federal Regulations
CG Conventional Gasoline
CHAD Consolidated Human Activity Database
CHD Coronary Heart Disease
CI Compression Ignition
CUT Chemical Industry Institute of Toxicology
CIMT Carotid Intima-Media Thickness
11
-------
CM 15 Gasoline with 15 percent methanol content
CMAI Chemical Market Associates Incorporated
CMAQ Community Multiscale Air Quality Model
CNG Compressed Natural Gas
CNS Central Nervous System
CO Carbon Monoxide
CO2 Carbon Dioxide
COI Cost-of-Illness
COPD Chronic Obstructive Pulmonary Disease
COV Coefficient of Variation
C-R Concentration-Response
CRC E-55/E-59 Coordinating Research Council Emission Test Program for Heavy Duty Trucks
CUA Cost-Utility Analysis
DCS Distributed Control System
DEOG Diesel Exhaust Organic Gases
DF Deterioration Factor
DOE Department of Energy
DPM Diesel Paniculate
E&C Engineering and Construction
E10 Gasoline Blend with Nominal 10 volume percent Ethanol
E200 Percent of Fuel Evaporated at 200 Degrees F (ASTM D 86)
E300 Percent of Fuel Evaporated at 300 Degrees F (ASTM D 86)
EAC Early Action Compact
EC/OC Elemental/Organic Carbon
ECM Engine Control Module
EGU Electrical Generating Utility
EIA Economic Impact Analysis
EIA Energy Information Administration (part of the U.S. Department of Energy)
EIM Economic Impact Model
EMS-HAP Emissions Modeling System for Hazardous Air Pollutants
EO Executive Order
EPA Environmental Protection Agency
EPEFE European Programme on Emissions, Fuels, and Engine Technology
EPAct Energy Policy Act of 2005
ETBE Ethyl Tertiary Butyl Ether
ETC Electronic Throttle Control
ETS Environmental Tobacco Smoke
EU European Union
EVOH Ethylene vinyl alcohol
F Fahrenheit
FACES Fresno Asthmatic Children's Environment Study
FBP Feed Boiling Point (also Final Boiling Point)
FCC Fluidized Catalytic Cracker
FCCU Fluidized Catalytic Cracking Unit
PEL Family Emission Level
FEV Functional Expiratory Volume
FHWA Federal Highway Administration
FOEB Fuel Oil Equivalent Barrel
FRM Final Rulemaking
FRTP Fixed Reduction Trigger Point
FTC Federal Trade Commission
FTP Federal Test Procedure
g/gal/day Grams per gallon per day
GDP Gross Domestic Product
GIS Geographic Information System
GM General Motors
ill
-------
GPA
GVW
GVWR
H2
HAD
HAP
HAPEM5
HC
HCO
HDN
HOPE
HDS
HOT
HEGO
HHC
HHDDT
HI
HLDT
HQ
HSR
HVGO
IBP
ICAO
ICD-9
ICI
IFF
IMO
IMPROVE
IRFA
IRIS
ISBL
ISC
ISCST
JAMA
K
KBBL
Kfoeb/day
KWH
LB
LCG
LCN
LCD
LDGT
LDGV
LOT
LDV
LEV I
LEV II
LEV
LHC
LLDT
LLE
LNS
LP
LPG
LRS
Geographic Phase-in Area
Gross Vehicle Weight
Gross Vehicle Weight Rating
Hydrogen gas
Diesel Health Assessment Document
Hazardous Air Pollutant
Hazardous Air Pollutant Exposure Model version 5
Hydrocarbon
Heavy Cycle Oil (a refinery stream)
Naphtha Hydrotreater (also Hydro-Denitrogenation Unit)
High density polyethylene
Hydro-Desulfurization Unit
Hydrotreater
(Heated) Exhaust Gas Oxygen
Heavy Hydrocrackate
Heavy Heavy-Duty Diesel Truck
Hazard Index
Heavy Light-Duty Truck
Hazard Quotient
Heavy Straight Run (a refinery stream)
Heavy Vacuum Gas Oil (a refinery stream)
Initial Boiling Point
International Civil Aviation Organization
International Classification of Diseases - Ninth Revision
Independent Commercial Importer
Institute Francais du Petrole
International Maritime Organization
Interagency Monitoring of Protected Visual Environments
Initial Regulatory Flexibility Analysis
Integrated Risk Information System
Inside Battery Limits
Integrated Source Complex
Industrial Source Complex Short Term
Journal of the American Medical Association
Thousand
Thousand barrels
Thousands of Fuel Oil Equivalent Barrels per Day
Kilowatt Hour
Pound
Light Cracked Gasoline
Light Coker Naphtha
Light Cycle Oil (a refinery stream)
Light Duty Gasoline Truck
Light Duty Gasoline Vehicle
Light-Duty Truck
Light-Duty Vehicle
Low Emission Vehicle I
Low Emission Vehicle II
Low Emission Vehicle
Light Hydrocrackate
Light Light-Duty Truck
Liquid-Liquid Extraction
Light Naphtha Splitter
Linear Programming (a type of refinery model)
Liquefied Petroleum Gas
Lower Respiratory Symptom
IV
-------
LSR
MACT
MAP
MAP
MC
MDPV
mg/m3
MHDDT
MHDT
MI
MILY
MLE
MM
MNCPES
MOBILE6.2
MON
MRAD
MRL
MSAT
MS ATI
MTBE
NAAQS
NAICS
NAS
NATA
NATTS
NCI
NCLAN
NEI
NEMS
NESCAUM
NESHAP
NHAPS
NHEXAS
NIOSH
NLEV
NMHC
NMIM
NMIM2005
NMMAPS
NMOG
NO2
NONROAD
NONROAD2005
NOX
NPRA
NPRM
NRC
NSTC
O&M
OBD
OEHHA
OGJ
OMB
OP
OSHA
Light Straight Run (a refinery stream)
Maximum Available Control Technology
(Engine) Mass Air Flow
(Engine) Manifold Absolute Pressure
Motorcycle
Medium-Duty Passenger Vehicle
Milligrams per cubic meter
Diesel-Fueled Medium Heavy-Duty Truck
Medium Heavy-Duty Truck
Myocardial Infarction
Morbidity Inclusive Life Years
Maximum Likelihood Estimate
Million
Minnesota Children's Pesticide Exposure Study
EPA's Highway Vehicle Emission Model
Motor Octane Number
Minor Restricted Activity Days
Minimum Risk Level
Mobile Source Air Toxics
2001 Mobile Source Air Toxics Rule
Methyl Tertiary-Butyl Ether
National Ambient Air Quality Standards
North American Industrial Classification System
National Academy of Science
National Scale Air Toxics Assessment
National Air Toxics Trends Sites
National Cancer Institute
National Crop Loss Assessment Network
National Emissions Inventory
National Energy Modeling System
Northeast States for Coordinated Air Use Management
National Emissions Standards for Hazardous Air Pollutants
National Human Activity Pattern Survey
National Human EXposure Assessment Survey
National Institute for Occupational Safety and Health
National Low Emission Vehicle
Non-Methane Hydrocarbons
National Mobile Inventory Model (EPA software tool)
National Mobile Inventory Model Released in 2005
National Mortality, Morbidity and Air Pollution
Non-Methane Organic Gases
Nitrogen Dioxide
EPA's Non-road Engine Emission Model
EPA's Non-road Engine Emission Model Released in 2005
Oxides of Nitrogen
National Petroleum Refiners Association
Notice of Proposed Rulemaking
National Research Council
National Science and Technology Council
Operating and maintenance
On-board Diagnostics
Office of Environmental Health Hazard Assessment
Oil and Gas Journal
Office of Management and Budget
Original Production
U.S. Department of Labor Occupational Safety and Health Organization
-------
OTAQ Office of Transportation and Air Quality
PADD Petroleum Administration for Defense District
PAH Polycyclic Aromatic Hydrocarbon
PC Particle Count
PC Passenger car
PFC Portable Fuel Containers
PFT perfluorocarbon tracer
PM Paniculate Matter
PM/NMHC Paniculate Matter to Non-Methane Hydrocarbon Ratio
PM10 Coarse Particle
PM2 5 Fine Particle
POM Polycyclic Organic Matter
PONA Paraffin, Olefin, Naphthene, Aromatic
PPM Parts Per Million
PRTP Percentage Reduction Trigger Point
PSI Pounds per Square Inch
PSR Power Systems Research
QALY Quality-Adjusted Life Year
R&D Research and Development
RAMS Regional Atmospheric Modeling System
RBOB Reformulated Blendstock for Oxygenate Blending
RECS Residential Energy Consumption Survey
REMSAD Regional Modeling System for Aerosols and Deposition
RfC Reference Concentration
RfD Oral reference dose
RFG Reformulated Gasoline
RFS Renewable Fuels Standard
RIA Regulatory Impact Analysis
RIOPA Relationship Between Indoor, Outdoor and Personal Air
ROI Return on Investment
RON Research Octane Number
RPM Revolutions Per Minute
RSM Response Surface Metamodel
RVP Reid Vapor Pressure
SAB Science Advisory Board
SAB-HES Science Advisory Board - Health Effects Subcommittee
S AE Society of Automotive Engineers
SBA Small Business Administration
SBAR Panel, or 'the Panel' Small Business Advocacy Review Panel
SBREFA Small Business Regulatory Enforcement Fairness Act of 1996
SCF Standard Cubic Foot
SECA SOX Emission Control Area
SER Small Entity Representative
SFTP Supplemental Federal Test Procedure
SHED Sealed Housing for Evaporative Determination
SI Spark Ignition
SIP State Implementation Plan
SOA Secondary Organic Aerosols
SPECIATE EPA's repository of Total Organic Compound (TOC) & Paniculate Matter (PM) speciated
profiles
SUV Sports-Utility Vehicle
SVM Small Vehicle Manufacturer
SVOC Semi-Volatile Organic Compound
SwRI Southwest Research Institute
TDM Travel Demand Model
TEACH Toxic Exposure Assessment - Columbia/Harvard
VI
-------
TEAM Total Exposure Assessment Methodology
THC Total Hydrocarbon
TMP 2,2,4-Trimethylpentane
TWA Time-weighted Average
TSP Total Suspended Paniculate Matter
TWC Three-Way Catalyst
UC Unified Cycle Emission Test Procedure from ARB
UCL Upper Confidence Limit
ULSD Ultra-Low Sulfur Diesel
URE Unit Risk Estimate
URS Upper Respiratory Symptom
UV Ultraviolet
UVb Ultraviolet-b
VGO Vacuum Gas Oil (a refinery stream)
VMT Vehicle Miles Traveled
VOC Volatile Organic Compound
VSL Value of a Statistical Life
VTB Vacuum Tower Bottoms (a refinery stream)
WLD Work Loss Days
WTP Willingness-to-Pay
Vll
-------
Final Regulatory Impact Analysis
Executive Summary
EPA is adopting new standards to reduce emissions of mobile source air toxics (MSATs)
including benzene and overall hydrocarbons from motor vehicles, motor vehicle fuels, and
portable fuel containers (PFCs). This Regulatory Impact Analysis provides technical, economic,
and environmental analyses of the new emission standards. The anticipated emission reductions
will significantly reduce exposure to harmful pollutants and also provide assistance to states and
regions facing ozone and particulate air quality problems that are causing a range of adverse
health effects, especially in terms of respiratory impairment and related illnesses.
Chapter 1 reviews information related to the health effects of mobile source air toxics.
Chapter 2 provides emissions inventory estimates, including estimates of anticipated emissions
reductions. Chapter 3 presents air quality, and resulting health and welfare effects, associated
with air toxics, ozone, and particulate matter (PM). Chapter 4 contains an overview of the
affected refiners and manufacturers, including a description of the range of products involved
and their place in the market. Chapters 5 through 7 summarize the available information
supporting the specific standards we are adopting, providing a technical justification for the
feasibility of the standards for vehicles, fuels, and PFCs, respectively. Chapters 8 throughlO
present cost estimates of complying with the new standards or vehicles, fuels, and PFCs,
respectively. Chapter 11 compares the costs and the emission reductions to generate an estimate
of the cost per ton of pollutant removed. Chapters 12 and 13 describe the estimated societal
costs and benefits of the rulemaking. Chapter 14 presents our Regulatory Flexibility Analysis, as
called for in the Regulatory Flexibility Act.
The following paragraphs briefly describe the standards that we are finalizing and the
estimated impacts.
Emissions Standards
Vehicles
We are adopting new standards for both exhaust and evaporative emissions from
passenger vehicles. The new exhaust emissions standards will significantly reduce non-methane
hydrocarbon (NMHC) emissions from passenger vehicles at cold temperatures. These
hydrocarbons include many mobile source air toxics (including benzene), as well as VOC.
The current NMHC standards are typically tested at 75° F, and recent research and
analysis indicates that these standards are not resulting in robust control of NMHC at lower
temperatures. (There is an existing cold temperature standard, but it applies only to CO.) We
believe that cold temperature NMHC control can be substantially improved using the same
technological approaches that are generally already being used in the Tier 2 vehicle fleet to meet
the stringent standards at 75° F. We project that these cold-temperature NMHC controls will also
result in lower direct PM emissions at cold temperatures.
Accordingly, we are requiring that light-duty vehicles, light-duty trucks, and medium-
duty passenger vehicles be subject to a new NMHC exhaust emissions standard at 20° F.
ES-1
-------
Final Regulatory Impact Analysis
Vehicles at or below 6,000 pounds gross vehicle weight rating (GVWR) will be subject to a
sales-weighted fleet average NMHC level of 0.3 grams/mile. Vehicles between 6,000 and 8,500
pounds GVWR and medium-duty passenger vehicles will be subject to a sales-weighted fleet
average NMHC level of 0.5 grams/mile. For lighter vehicles, the standard will phase in between
2010 and 2013. For heavier vehicles, the new standards will phase in between 2012 and 2015.
We are also adopting a credit program and other provisions designed to provide flexibility to
manufacturers, especially during the phase-in periods. These provisions are designed to allow
the earliest possible phase-in of standards and help minimize costs and ease the transition to new
standards.
We are also adopting a set of nominally more stringent evaporative emission standards
for all light-duty vehicles, light-duty trucks, and medium-duty passenger vehicles. The new
standards are equivalent to California's Low Emission Vehicle II (LEV II) standards, and they
reflect the evaporative emissions levels that are already being achieved nationwide. The
standards will codify the approach that manufacturers are already taking for 50-state evaporative
systems, and thus the standards will prevent backsliding in the future. The new evaporative
emission standards begin in 2009 for lighter vehicles and in 2010 for the heavier vehicles.
Gasoline Fuel Standards
We are requiring that beginning January 1, 2011, refiners and fuel importers will meet a
refinery average gasoline benzene content standard of 0.62% by volume on all their gasoline,
both reformulated and conventional (except for California, which is already covered by a similar
relatively stringent state program).
This new fuel standard will result in air toxics emissions reductions that are greater than
required under all existing gasoline toxics programs. As a result, EPA is establishing that upon
full implementation in 2011, the regulatory provisions for the benzene control program will
become the single regulatory mechanism used to implement the reformulated gasoline (RFG)
and Anti-dumping annual average toxics requirements. The current RFG and Anti-dumping
annual average provisions will be replaced by the new benzene control program. The MSAT2
benzene control program will also replace the MSAT1 requirements. In addition, the program
will satisfy certain fuel MS AT conditions of the Energy Policy Act of 2005. In all of these ways,
we will significantly consolidate and simplify the existing national fuel-related MSAT regulatory
program.
We are also allowing that refiners could generate benzene credits and use or transfer them
as a part of a nationwide averaging, banking, and trading (ABT) program. From 2007-2010
refiners can generate benzene credits by taking early steps to reduce gasoline benzene levels.
Beginning in 2011 and continuing indefinitely, refiners can generate credits by producing
gasoline with benzene levels below the 0.62 vol% refinery average standard. Refiners can apply
the credits towards company compliance, "bank" the credits for later use, or transfer ("trade")
them to other refiners nationwide (outside of California) under the new program. Under this
program, refiners can use credits to achieve compliance with the benzene content standard. In
addition, to the 0.62 vol% standards, refiners must also meet a maximum average benzene
ES-2
-------
Final Regulatory Impact Analysis
standard of 1.3 vol% beginning on July 1, 2012. A refinery's or importer's actual annual
average gasoline benzene levels may not exceed this maximum average standard.
Portable Fuel Container Controls
Portable fuel containers (PFCs) include gasoline containers (gas cans) and kerosene and
diesel containers. PFCs are consumer products used to refuel a wide variety of equipment,
including lawn and garden equipment, generators, heaters, recreational equipment, and passenger
vehicles that have run out of gas. We are adopting standards that will reduce hydrocarbon
emissions from evaporation, permeation, and spillage. These standards will significantly reduce
benzene and other toxics, as well as VOC more generally. VOC is an ozone precursor. We are
also applying the new requirements to kerosene and diesel containers, which are identical to gas
cans except for their color and could be used for gasoline.
We are adopting a performance-based standard of 0.3 grams per gallon per day of
hydrocarbons, based on the emissions from the can over a diurnal test cycle. The standard will
apply to PFCs manufactured on or after January 1, 2009. We are also adopting test procedures
and a certification and compliance program, in order to ensure that PFCs will meet the emission
standard over a range of in-use conditions. The new requirements will result in the best available
control technologies, such as durable permeation barriers, automatically closing spouts, and cans
that are well-sealed.
California implemented an emissions control program for PFCs in 2001, and since then,
several other states have adopted the program. In 2005, California adopted a revised program,
which will take effect July 1, 2007. The revised California program is very similar to the
program we are adopting. Although a few aspects of the program we are adopting are different,
we believe manufacturers will be able to meet both EPA and California requirements with the
same container designs.
Projected Impacts
The following paragraphs and tables summarize the projected emission reductions and
costs associated with the emission standards. See the detailed analysis later in this document for
further discussion of these estimates.
Emissions Reductions
Toxics
ES-3
-------
Final Regulatory Impact Analysis
sources as well due to lower benzene levels in gasoline. Annual benzene emissions from
Table 1: Estimated Reductions in Benzene Emissions from New Control Measures by
Sector, 2020 and 2030 (tons per year)
Fuels
Vehicles
PFCs
2020
17,618
27,097
718
2030
19,643
45,037
814
Total 42,760 61,035
Table 2: Estimated Reductions in MSAT Emissions from New Control Measures by
Sector, 2020 and 2030 (tons per year)
Fuels
Vehicles
PFCs
2020
17,618
177,007
18,553
2030
19,643
294,284
21,036
Total 210,303 330,844
voc
VOC emissions will be reduced by the hydrocarbon emission standards for both light-
duty vehicles and PFCs. Annual VOC emission reductions from these sources will be about 34%
lower in 2030 because of the new rule.
Table 3: Estimated Reductions in VOC Emissions from Light-Duty Gasoline Vehicles and
PFCs, 2020 and 2030 (tons per year)
Vehicles
PFCs
2020
529,363
216,294
2030
882,762
245,255
ES-4
-------
Final Regulatory Impact Analysis
Total
745,658
1,128,017
PM2.5
We expect that only the vehicle control will reduce emissions of direct PM2.5. As shown
in Table 4, we expect this control to reduce direct PM2.5 emissions by about 19,000 tons in 2030.
In addition, the VOC reductions from the vehicle and PFC standards will also reduce secondary
formation of PM2.5.
Table 4. Estimated National Reductions in Direct PMi.s Exhaust Emissions from Light-
Duty Gasoline Vehicles and Trucks, 2020 and 2030 (tons per year)
PM2 5 Reductions from Vehicle
Standards (tons)
2020
11,646
2030
19,421
Costs
Fuels
The refinery model estimates that the benzene standard will cost 0.27 cents per gallon,
averaged over the entire U.S. gasoline pool. (When averaged only over those refineries which
are assumed to take steps to reduce their benzene levels, the average cost will be 0.40 cents per
gallon.) This per-gallon cost will result from an industry-wide investment in capital equipment
of $1,110 million to reduce gasoline benzene levels. This will amount to an average of $14
million in capital investment in each refinery that adds such equipment. The aggregate costs for
the fuel program for 2020 and 2030 are provided in Table 5. The increase in costs is due to the
projected increase in gasoline usage.
Table 5. Estimated Aggregate Annual Cost for the Benzene Standard, 2020 and 2030
2020
2030
Fuels program
$398 million
$441 million
Vehicles
ES-5
-------
Final Regulatory Impact Analysis
We project that the average incremental costs associated with the new cold temperature
standards will be less than $1 per vehicle. We are not projecting changes to vehicle hardware as
a result of the new standard. Costs are associated with vehicle R&D and recalibration as well as
facilities upgrades to handle additional development testing under cold conditions. Also, we are
not anticipating additional costs for the new evaporative emissions standard. We expect that
manufacturers will continue to produce 50-state evaporative systems that meet LEV II standards.
Therefore, harmonizing with California's LEV-II evaporative emission standards will streamline
certification and be an "anti-backsliding" measure. It also will codify the approach
manufacturers have already indicated they are taking for 50-state evaporative systems.
We also estimated annual aggregate costs associated with the new cold temperature
emissions standards. These costs are projected to increase with the phase-in of standards and
peak in 2014 at about $13.4 million per year, then decrease as the fixed costs are fully amortized.
As shown in Table 6, we project the costs will be fully amortized by 2020.
Table 6. Estimated Aggregate Annual Cost for the Vehicle Standards, 2020 and 2030
2020
2030
Vehicles program
$0
$0
PFCs
Table 7 summarizes the projected near-term and long-term per unit average costs to meet
the new emission standards. Long-term impacts on PFCs are expected to decrease as
manufacturers fully amortize their fixed costs. The table also shows our projections of average
fuel savings over the life of the PFC when used with gasoline.
Table 7 Estimated Average PFC Costs and Lifetime Fuel Savings
Near-Term Costs
Long-Term Costs
Cost
$2.69
$1.52
Gasoline Savings (NPV)
$4.24
We have also estimated aggregate costs and gasoline fuel savings which are projected to
peak in 2013 at about $61 million and then drop to about $33 million once fixed costs are
recovered. The aggregate annual costs and gasoline savings estimates for 2020 and 2030 are
provided in Table 8.
Table 8. Estimated Aggregate Annual Cost and Gasoline Savings for the PFC Standards,
2020 and 2030
ES-6
-------
Final Regulatory Impact Analysis
PFC Costs
2020
$37,542,748
2030
$45,764,401
PFC Gasoline Savings $109,589,064 $124,264,434
Cost Per Ton
We have calculated the cost per ton of HC, benzene, total MSATs, and PM emissions
reductions associated with the new fuel, vehicle, and PFC programs. We have calculated the
costs per ton using the net present value of the annualized costs of the program, including PFC
gasoline fuel savings, from 2009 through 2030 and the net present value of the annual emission
reductions through 2030. We have also calculated the cost per ton of emissions reduced in the
year 2020 and 2030 using the annual costs and emissions reductions in that year alone. This
number represents the long-term cost per ton of emissions reduced. For fuels, the cost per ton
estimates include costs and emission reductions that will occur from all motor vehicles and
nonroad engines fueled with gasoline as well as PFCs and gasoline distribution.
We have not attempted to apportion costs across these various pollutants for purposes of
the cost per ton calculations since there is no distinction in the technologies, or associated costs,
used to control the pollutants. Instead, we have calculated costs per ton by assigning all costs to
each individual pollutant. If we apportioned costs among the pollutants, the costs per ton
presented here would be proportionally lowered depending on what portion of costs were
assigned to the various pollutants. The results of the analysis are provided in Tables 9 through
12.
The cost per ton estimates for each individual program are presented separately in the
tables below, and are part of the justification for each of the programs. For informational
purposes, we also present the cost per ton for the three programs combined.
ES-7
-------
Final Regulatory Impact Analysis
Table 9. HC Aggregate Cost per Ton and Long-Term Annual Cost Per Ton
($2003)
Vehicles
PFCs (without
fuel savings)
PFCs (with fuel
savings)
Combined (with
fuel savings)
Discounted
Lifetime
Cost per ton at 3%
$14
$240
$0
$0
Discounted
Lifetime
Cost per ton at 7%
$18
$270
$0
$0
Long-Term Cost
per Ton in 2020
$0
$170
$0
$0
Long-Term Cost
per Ton in 2030
$0
$190
$0
$0
Table 10. Benzene Aggregate Cost per Ton and Long-Term Annual Cost Per Ton
($2003)
Fuels
Vehicles
PFCs (without
fuel savings)
PFCs (with fuel
savings)
Combined (with
fuel savings)
Discounted
Lifetime
Cost per ton at 3%
$22,400
$270
$74,500
$0
$8,200
Discounted
Lifetime
Cost per ton at 7%
$23,100
$360
$82,900
$0
$8,600
Long-Term Cost
per Ton in 2020
$22,600
$0
$52,200
$0
$7,600
Long-Term Cost
per Ton in 2030
$22,500
$0
$56,200
$0
$5,900
ES-8
-------
Final Regulatory Impact Analysis
Table 11 MSAT Aggregate Cost per Ton and Long-Term Annual Cost Per Ton
($2003)
Fuels
Vehicles
PFCs (without
fuel savings)
PFCs (with fuel
savings)
Combined (with
fuel savings)
Discounted
Lifetime
Cost per ton at 3%
$22,400
$42
$2,800
$0
$1,700
Discounted
Lifetime
Cost per ton at 7%
$23,100
$54
$3,100
$0
$1,800
Long-Term Cost
per Ton in 2020
$22,600
$0
$2,000
$0
$1,600
Long-Term Cost
per Ton in 2030
$22,500
$0
$2,200
$0
$1,100
Table 12 Direct PM Aggregate Cost per Ton and Long-Term Annual Cost Per Ton
($2003)
Vehicles
Discounted
Lifetime
Cost per ton at 3%
$650
Discounted
Lifetime
Cost per ton at 7%
$870
Long-Term Cost
per Ton in 2020
$0
Long-Term Cost
per Ton in 2030
$0
Benefits
This analysis projects significant benefits throughout the period from initial
implementation of the new standards through 2030. When translating emission benefits to health
effects and monetized values, however, we only quantify the PM-related benefits associated with
the new cold temperature vehicle standards. The reductions in PM from the cold temperature
vehicle standards will result in significant reductions in premature deaths and other serious
human health effects, as well as other important public health and welfare effects. Table 13
provides the estimated monetized benefits of the cold temperature vehicle standards for 2020 and
2030. We estimate that in 2030, the benefits we are able to monetize are expected to be
approximately $6.3 billion using a 3 percent discount rate and $5.7 billion using a 7 percent
discount rate, assuming a background PM threshold of 3 ug/m3 in the calculation of PM
mortality. There are no compliance costs associated with the cold temperature vehicle program
after 2019; vehicle compliance costs are primarily research and development, and facility costs
are expected to be recovered by manufacturers over the first ten years of the program beginning
in 2010. Total costs of the entire MSAT rule, which include both the PFC, vehicle, and fuel
standards, are $400 million in 2030 (in 2003$, including fuel savings).
ES-9
-------
Final Regulatory Impact Analysis
The PM2.5 benefits are scaled based on relative changes in direct PM emissions between
this rule and the proposed Clean Air Nonroad Diesel (CAND) rule. As explained in Section
12.2.1 of the RIA, the PM2.5 benefits scaling approach is limited to those studies, health impacts,
and assumptions that were used in the proposed CAND analysis. As a result, PM-related
premature mortality is based on the updated analysis of the American Cancer Society cohort
(ACS; Pope et al., 2002). However, it is important to note that since the CAND rule, EPA's
Office of Air and Radiation (OAR) has adopted a different format for its benefits analysis in
which characterization of the uncertainty in the concentration-response function is integrated into
the main benefits analysis. Within this context, additional data sources are available, including a
recent expert elicitation and updated analysis of the Six-Cities Study cohort (Laden et al., 2006).
Please see the PM NAAQS RIA for an indication of the sensitivity of our results to use of
alternative concentration-response functions.
The analysis presented here assumes a PM threshold of 3 ug/m3, equivalent to
background. Through the RIA for the Clean Air Interstate Rule (CAIR), EPA's consistent
approach had been to model premature mortality associated with PM exposure as a nonthreshold
effect; that is, with harmful effects to exposed populations modeled regardless of the absolute
level of ambient PM concentrations. This approach had been supported by advice from EPA's
technical peer review panel, the Science Advisory Board's Health Effects Subcommittee (SAB-
HES). However, EPA's most recent PM2.s Criteria Document concludes that "the available
evidence does not either support or refute the existence of thresholds for the effects of PM on
mortality across the range of concentrations in the studies," (p. 9-44). Furthermore, in the RIA
for the PM NAAQS we used a threshold of 10 ug/m3 based on recommendations by the Clean
Air Scientific Advisory Committee (CAS AC) for the Staff Paper analysis. We consider the
impact of a potential, assumed threshold in the PM-mortality concentration response function in
Section 12.6.2.2 of the RIA
Table 13 Estimated Monetized PM-Related Health Benefits of the Mobile Source Air
Toxics Standards: Cold Temperature Controls
Using a 3% discount rate
Using a 7% discount rate
Total Benefits3' b'c (billions 2003$)
2020
$3.3 + B
$3.0 + B
2030
$6.3 + B
$5.7 + B
a Benefits include avoided cases of mortality, chronic illness, and other morbidity health endpoints. PM-related
mortality benefits estimated using an assumed PM threshold at background levels (3 ug/m3). There is
uncertainty about which threshold to use and this may impact the magnitude of the total benefits estimate. For a
more detailed discussion of this issue, please refer to Section 12.6 of the RIA.
b For notational purposes, unqualified benefits are indicated with a "B" to represent the sum of additional
monetary benefits and disbenefits. A detailed listing of unqualified health and welfare effects is provided in
Table 12.1-2 of the RIA.
0 Results reflect the use of two different discount rates: 3 and 7 percent, which are recommended by EPA's
Guidelines for Preparing Economic Analyses and OMB Circular A-4. Results are rounded to two significant
digits for ease of presentation and computation.
Economic Impact Analysis
ES-10
-------
Final Regulatory Impact Analysis
We prepared an Economic Impact Analysis (EIA) to estimate the economic impacts of
the emission control program on the PFC, gasoline fuel, and light-duty vehicle markets. Our
estimates of the net social costs of the program for 2020 and 2030 are provided in Table 14
below. These estimates reflect the estimated costs associated with the gasoline, PFC, and vehicle
controls and the expected gasoline fuel savings from better evaporative controls on PFCs. The
results of the economic impact modeling performed for the gasoline fuel and PFC control
programs suggest that the social costs of those two programs are expected to be about $440.1
million in 2020 with consumers of these products expected to bear about 58 percent of these
costs. We estimate fuel savings of about $80.7 million in 2020 that will accrue to consumers.
There are no social costs associated with the vehicle program in 2020.
Table 14 Net Social Costs Estimates for the Program (Millions of 2003$)
1
2020
2030 |
Net Social Costs $359.4 $400.0
Impact on Small Businesses
We prepared a Regulatory Flexibility Analysis, which evaluates the potential impacts of
new standards and fuel controls of this rule on small entities. As a part of this analysis, we
interacted with several small entities representing the various affected sectors and convened a
Small Business Advocacy Review Panel to gain feedback and advice from these representatives.
This feedback was used to develop regulatory alternatives to address the impacts of the rule on
small businesses. Small entities raised general concerns related to potential difficulties and costs
of meeting the upcoming standards.
The Panel consisted of members from EPA, the Office of Management and Budget, and
the Small Business Administration's Office of Advocacy. We are adopting most of the Panel's
recommendations. These provisions will reduce the burden on small entities that will be subject
to this rule's requirements. We have included provisions that give small light-duty vehicle
manufacturers, small gasoline refiners, and small PFC manufacturers several compliance options
aimed specifically at reducing the burden on these small entities. In general, for vehicles and
fuels, the options are similar to small entity provisions adopted in prior rulemakings where EPA
set vehicle and fuel standards. The options included for small PFC manufacturers are unique to
this rulemaking since we are adopting PFC standards for the first time.
ES-11
-------
Final Regulatory Impact Analysis
Chapter 1: Table of Contents
Chapter 1: Mobile Source Air Toxics Health Information 2
1.1. What Are MSATs? 2
1.1.1. Compounds Emitted by Mobile Sources and Identified in IRIS 2
1.1.2. Compounds Emitted by Mobile Sources and Included on Section 112(b) List of
Hazardous Air Pollutants 5
1.1.3. Other Sources of Information on Compounds with Potential Serious Adverse
Health Effects 6
1.1.4. Which Mobile Source Emissions Pose the Greatest Health Risk at Current Levels??
1.1.4.1. Risk Drivers in 1999 National-Scale Air Toxics Assessment 7
1.1.4.2. 1999 NATA Risk Drivers with Significant Mobile Source Contribution .... 10
1.2. Dose-Response and Agency Risk Assessment Practice 11
1.2.1. Cancer 11
1.2.2. Chronic Exposure and Noncancer Health Effects 13
1.2.3. Acute Exposure and Noncancer Health Effects 13
1.3. Summary of Air Toxic Health Effects 14
1.3.1. Benzene 14
1.3.2. 1,3-Butadiene 17
1.3.3. Formaldehyde 18
1.3.4. Acetaldehyde 19
1.3.5. Acrolein 20
1.3.6. Naphthalene 20
1.3.7. 2,2,4-Trimethylpentane 21
1.3.8. Ethylbenzene 21
1.3.9. n-Hexane 22
1.3.10. Methyl Tertiary Butyl Ether (MTBE) 23
1.3.11. Styrene 23
1.3.12. Toluene 23
1.3.13.Xylenes 24
1.3.14. Polycyclic Organic Matter (POM) 25
1.3.15. Diesel Exhaust 25
1.4. Emerging Issues 26
1.4.1. Gasoline PM 26
1.4.2. Metals 29
1-1
-------
Final Regulatory Impact Analysis
Chapter 1: Mobile Source Air Toxics Health Information
1.1. What Are MSATs?
Section 202(1) refers to "hazardous air pollutants from motor vehicles and motor vehicle
fuels." We use the term "mobile source air toxics (MSATs)" to refer to compounds that are
emitted by mobile sources and have the potential for serious adverse health effects. There are a
variety of ways in which to identify compounds that have the potential for serious adverse health
effects. For example, EPA's Integrated Risk Information System (IRIS) is EPA's database
containing information on human health effects that may result from exposure to various
chemicals in the environment. In addition, Clean Air Act section 112(b) contains a list of
hazardous air pollutants that EPA is required to control through regulatory standards; other
agencies or programs such as the Agency for Toxic Substances and Disease Registry and the
California EPA have developed health benchmark values for various compounds; and the
International Agency for Research on Cancer and the National Toxicology Program have
assembled evidence of substances that cause cancer in humans and issue judgments on the
strength of the evidence. Each source of information has its own strengths and limitations. For
example, there are inherent limitations on the number of compounds that have been investigated
sufficiently for EPA to conduct an IRIS assessment. There are some compounds that are not
listed or not quantitatively assessed in IRIS but are considered to be hazardous air pollutants
under Clean Air Act section 112(b) and are regulated by the Agency (e.g., propionaldehyde,
2,2,4-trimethylpentane).
1.1.1. Compounds Emitted by Mobile Sources and Identified in IRIS
In its 2001 MS AT rule, EPA identified a list of 21 MSATs. We listed a compound as an
MSAT if it was emitted from mobile sources, and if the Agency had concluded in IRIS that the
compound posed a potential cancer hazard and/or if IRIS contained an inhalation reference
concentration or ingestion reference dose for the compound. Since 2001, EPA has conducted an
extensive review of the literature to produce a list of the compounds identified in the exhaust or
evaporative emissions from onroad and nonroad equipment, using baseline as well as alternative
fuels (e.g., biodiesel, compressed natural gas).1 This list, the Master List of Compounds Emitted
by Mobile Sources ("Master List"), currently includes approximately 1,000 compounds. It is
available in the public docket for this rule and on the web (http://www.epa.gov/otaq/toxics.htm ).
Table l.l.-l lists those compounds from the Master List that currently meet those 2001 MSAT
criteria, based on the current IRIS.
Table l.l.-l identifies all of the compounds from the Master List that are present in IRIS
with (a) a cancer hazard identification of known, probable, or possible human carcinogens (under
the 1986 EPA cancer guidelines) or carcinogenic to humans, likely to be carcinogenic to humans,
or suggestive evidence of carcinogenic potential (under the 2005 EPA cancer guidelines); and/or
(b) an inhalation reference concentration or an ingestion reference dose. Although all these
compounds have been detected in emissions from mobile sources, many are emitted in trace
amounts and data are not adequate to develop an inventory. Those compounds for which we
have developed an emissions inventory are summarized in Chapter 2 Table 2.2.-1. There are
1-2
-------
Final Regulatory Impact Analysis
several compounds for which IRIS assessments are underway and therefore are not included in
Table l.l.-l. These compounds are: cerium, copper, ethanol, ethyl tertiary butyl ether (ETBE),
platinum, propionaldehyde, and 2,2,4-trimethylpentane.
The fact that a compound is listed in Table l.l.-l does not imply a risk to public health or
welfare at current levels, or that it is appropriate to adopt controls to limit the emissions of such a
compound from motor vehicles or their fuels. In conducting any such further evaluation,
pursuant to sections 202(a) or 21 l(c) of the Act, EPA would consider whether emissions of the
compound from motor vehicles cause or contribute to air pollution which may reasonably be
anticipated to endanger public health or welfare.
1-3
-------
Final Regulatory Impact Analysis
Table l.l.-l. Compounds Emitted by Mobile Sources That Are Listed in IRIS*
1,1,1 ,2-Tetrafluoroethane
1,1,1 -Trichloroethane
1,1-Biphenyl
1 ,2-Dibromoethane
1 ,2-Dichlorobenzene
1,3-Butadiene
2,4-Dinitrophenol
2-Methylnaphthalene
2-Methylphenol
4-Methylphenol
Acenaphthene
Acetaldehyde
Acetone
Acetophenone
Acrolein (2-propenal)
Ammonia
Anthracene
Antimony
Arsenic, inorganic
Barium and compounds
B enz [a] anthracene
Benzaldehyde
Benzene
Cadmium
Carbon disulfide
Carbon tetrachloride
Chlorine
Chlorobenzene
Chloroform
Chromium III
Chromium VI
Chrysene
Crotonaldehyde
Cumene (isopropyl benzene)
Cyclohexane
Cyclohexanone
Di(2-ethylhexyl)phthalate
Dib enz [a,h] anthracene
Dibutyl phthalate
Di chl oromethane
Diesel PM and Diesel exhaust
organic gases
Di ethyl phthalate
Ethylbenzene
Ethylene glycol monobutyl
ether
Fluoranthene
Fluorene
Manganese
Mercury, elemental
Methanol
Methyl chloride
Methyl ethyl ketone
(MEK)
Methyl isobutyl ketone
(MIBK)
Methyl tert-butyl ether
(MTBE)
Molybdenum
Naphthalene
Nickel
Nitrate
N-Nitrosodiethylamine
N-Nitrosodimethylamine
N-Nitroso-di-n-
butylamine
N-Nitrosodi-N-
propylamine
N-Nitrosopyrrolidine
Pentachlorophenol
Phenol
Phosphorus
Phthalic anhydride
Pyrene
Selenium and compounds
Silver
1-4
-------
Final Regulatory Impact Analysis
Benzo[a]pyrene (BaP)
B enzo [b ]fluoranthene
Benzo[k]fluoranthene
Benzole acid
Beryllium and compounds
Boron (Boron and Borates
only)
Bromomethane
Butyl benzyl phthalate
Formaldehyde
Furfural
Hexachl orodib enzo-p-di oxin,
mixture (dioxin/furans)
n-Hexane
Hydrogen cyanide
Hydrogen sulfide
Indeno[ 1 ,2,3 -cd]pyrene
Lead and compounds
(inorganic)
Strontium
Styrene
Tetrachloroethylene
Toluene
Tri chl orofluoromethane
Vanadium
Xylenes
Zinc and compounds
*Compounds listed in IRIS as known, probable, or possible human carcinogens and/or
pollutants for which the Agency has calculated a reference concentration or reference dose.
1.1.2. Compounds Emitted by Mobile Sources and Included on Section 112(b) List of
Hazardous Air Pollutants
Clean Air Act section 112(b) contains a list of hazardous air pollutants that EPA is
required to control through regulatory standards. As discussed above, there are some compounds
emitted by mobile sources that are not listed in IRIS but are considered to be hazardous air
pollutants under Clean Air Act section 112(b) and are regulated by the Agency such as
propionaldehyde and 2,2,4-trimethylpentane. Compounds emitted by mobile sources that are
Clean Air Act section 112(b) hazardous air pollutants are listed in Table 1.1.-2. Although all
these compounds have been detected in emissions from mobile sources, many are emitted in
trace amounts and data are not adequate to develop an inventory. Those compounds for which
we have developed an emissions inventory are summarized in Table 2.2.-1.
Table 1.1.-2. Compounds Emitted by Mobile Sources That Are Listed in CAA Section
1 , 1 ,2-Trichloroethane
1 ,2-Dibromoethane
1,3 -Butadiene
2,2,4-Trimethylpentane
2,3,7,8-Tetrachlorodibenzo-p-
dioxin
2,4-Dinitrophenol
Carbon disulfide
Carbon tetrachloride
Chlorine
Chlorobenzene
Chloroform
Chromium (III and VI)
Methyl ethyl ketone
Methyl tert-butyl ether
Methyl chloride
Naphthalene
Nickel compounds
N-Nitrosodimethylamine
1-5
-------
Final Regulatory Impact Analysis
2-Methylphenol (o-cresol)
4-Methylphenol (p-cresol)
Acetaldehyde
Acetophenone
Acrolein
Antimony compounds
Arsenic compounds
Benzene
Beryllium
Biphenyl
Bromomethane
Cadmium compounds
Cumene
Di(2-ethylhexyl)phthalate
(DEHP)
Dibutylphthalate
Dichloromethane
Ethyl benzene
Formaldehyde
Hexane
Hydrogen cyanide
("Cyanide compounds in
Section 112(b))
Lead compounds
Manganese
Mercury compounds
Methanol
Pentachlorophenol
Phenol
Phosphorus
Phthalic anhydride
Polycyclic organic matter*
Propionaldehyde
Selenium compounds
Styrene
Tetrachl oroethy 1 ene
Toluene
Xylenes (isomers and mixture)
*Includes organic compounds with more than one benzene ring, and which have a boiling point greater than or equal
to 100.5 C.
1.1.3. Other Sources of Information on Compounds with Potential Serious Adverse
Health Effects
Additional sources of information are available to characterize the potential for cancer or
noncancer health effects from toxic air pollutants. These include the Agency for Toxic
Substances and Disease Registry list of minimal risk levels (http://www.atsdr.cdc.gov/mrls.html),
California EPA list of Reference Exposure Levels
(http://www.oehha.ca.gov/risk/ChemicalDB/index.asp), International Agency for Research on
Cancer lists of carcinogenic compounds (http://www.iarc.fr/ENG/Databases/index.php), the
National Toxicology Program list of carcinogenic compounds (http://ntp-server.niehs.nih.gov/),
and the U.S. EPA Emergency Planning and Community Right-to-Know Act list of extremely
hazardous substances (http://yosemite.epa.gov/oswer/ceppoehs.nsf/content/BackGround). EPA
relies on these sources of information, as appropriate, for certain types of analyses.2
1-6
-------
Final Regulatory Impact Analysis
1.1.4. Which Mobile Source Emissions Pose the Greatest Health Risk at Current Levels?
The 1999 National-Scale Air Toxics Assessment (NAT A) provides some perspective on
which mobile source emissions pose the greatest risk at current estimated ambient levels.A We
also conducted a national-scale assessment for future years, which is discussed more fully in
Chapters 2 and 3 of the RIA. The limitations and uncertainties associated with NAT A are
discussed in Section 3.2.1.3 of the RIA. Our understanding of what emissions pose the greatest
risk will evolve over time, based on our understanding of the ambient levels and health effects
associated with the compounds.8
1.1.4.1. Risk Drivers in 1999 National-Scale Air Toxics Assessment
The 1999 NATA evaluates 177 hazardous air pollutants currently listed under CAA
section 112(b), as well as diesel PM. NATA is described in greater detail in Chapters 2 and 3 of
this RIA. Additional information can also be obtained from the NATA website
(http://www.epa.gov/ttn/atw/natal999). Based on the assessment of inhalation exposures
associated with outdoor sources of these hazardous air pollutants, NATA has identified cancer
and noncancer risk drivers on a national and regional scale (Table 1.1.-3). A cancer risk driver
on a national scale is a hazardous air pollutant for which at least 25 million people are exposed to
risk greater than ten in one million. Benzene is the only compound identified in the 1999 NATA
as a national cancer risk driver.0 A cancer risk driver on a regional scale is a hazardous air
pollutant for which at least one million people are exposed to risk greater than ten in one million
or at least 10,000 people are exposed to risk greater than 100 in one million. Twelve compounds
(or groups of compounds in the case of POM) were identified as regional cancer risk drivers.
The 1999 NATA concludes that diesel particulate matter is among the substances that pose the
greatest relative risk, although the cancer risk cannot be quantified.
A noncancer risk driver at the national scale is a hazardous air pollutant for which at least
25 million people are exposed at a concentration greater than the inhalation reference
concentration. The RfC is an estimate (with uncertainty spanning perhaps an order of
magnitude) of a daily exposure to the human population (including sensitive subgroups) that is
likely to be without appreciable risk of deleterious effects during a lifetime. Acrolein is the only
compound identified in the 1999 NATA as a national noncancer risk driver.0 A noncancer risk
driver on a regional scale is defined as a hazardous air pollutant for which at least 10,000 people
are exposed to an ambient concentration greater than the inhalation reference concentration.
A It is, of course, not necessary for EPA to show that a compound is a national or regional risk driver to
show that its emission from motor vehicles may reasonably cause or contribute to endangerment of public health or
welfare. A showing that motor vehicles contribute some non-trivial percentage of the inventory of a compound
known to be associated with adverse health effects would normally be sufficient. Cf. Bluewater Network v. EPA.
370 F. 3d 1, 15 (D.C. Cir. 2004).
B The discussion here considers risks other than those attributed to ambient levels of criteria pollutants.
c Benzene was assigned an overall confidence level of "higher" based on consideration of the combined
uncertainties from the modeling estimates.
D Acrolein was assigned an overall confidence level of "lower" based on consideration of the combined
uncertainties from the modeling estimates.
1-7
-------
Final Regulatory Impact Analysis
Sixteen regional-scale noncancer risk drivers were identified in the 1999 NATA (see Table 1.1.-
3.).
-------
Final Regulatory Impact Analysis
Table 1.1.-3. National and Regional Cancer and Noncancer Risk Drivers in 1999 NATA
Cancer l
National drivers 2
Benzene11
Regional drivers 3
Arsenic compounds1"
BenzidineL
l,3-ButadieneL
Cadmium compounds1"
Carbon tetrachloride11
Chromium VIL
Coke ovenM
Ethylene oxide11
HydrazineM
NaphthaleneM
Perchl oroethy 1 eneM
Polycyclic organic matterM
Noncancer
National drivers 4
AcroleinL
Regional drivers 5
Antimony11
Arsenic compounds1"
l,3-ButadieneL
Cadmium compounds1"
Chlorine1"
Chromium VIL
Diesel PMM
FormaldehydeM
Hexamethylene l-6-diisocyanateM
Hydrazine11
Hydrochloric acidL
Maleic anhydride1"
Manganese compounds1"
Nickel compounds1"
2,4-Toluene diisocyanate1"
Triethylamine1"
:The list of cancer risk drivers does not include diesel paniculate matter. However, the 1999 NATA
concluded that it was one of the pollutants that posed the greatest relative cancer risk.
2 At least 25 million people exposed to risk >10 in 1 million
3 At least 1 million people exposed to risk >10 in 1 million or at least 10,000 people exposed to risk >100
in 1 million
4 At least 25 million people exposed to a hazard quotient > 1.0
5 At least 10,000 people exposed to a hazard quotient > 1
EPA has assigned an overall confidence level for each pollutant in NATA based on consideration of the
combined uncertainties from emissions estimation, ambient concentration modeling, and exposure
modeling. These judgments refer to the relative confidence between two air toxics compounds. A
judgment of "Higher" (H) means the confidence is higher for this compound than for compounds
assigned a "Medium" (M) or "Lower" (L).
1-9
-------
Final Regulatory Impact Analysis
It should be noted that varying levels of confidence are associated with risk estimates for
individual pollutants, based on the quality of the data used to estimate emissions, ambient
concentrations and exposure. For the pollutants included in NAT A, EPA rated its confidence
inrisk estimates, based on the quality of the data used for emissions, air quality, and exposure
modeling, as high, medium, or lower. EPA has a high level of confidence in the data for benzene,
medium confidence in the data for formaldehyde, but lower confidence in data for 1,3-butadiene
and acrolein.
1.1.4.2.
1999 NATA Risk Drivers with Significant Mobile Source Contribution
Among the national and regional-scale cancer and noncancer risk drivers identified in the
1999 NATA, seven compounds have significant contributions from mobile sources: benzene,
1,3-butadiene, formaldehyde, acrolein, polycyclic organic matter (POM), naphthalene, and diesel
particulate matter and diesel exhaust organic gases (Table 1.1.-4.). For example, mobile sources
contribute 68% of the national benzene inventory, with 49% from on-road sources and 19% from
nonroad sources based on 1999 NATA data.
Table 1.1.-4. Mobile Source Contribution to 1999 NATA Risk Drivers
1999 NATA Risk Drivers
Benzene11
l,3-ButadieneL
FormaldehydeM
AcroleinL
Polycyclic organic matter*M
NaphthaleneM
Diesel PM and Diesel
exhaust organic gasesM
Percent
Contribution
from All Mobile
Sources
68%
58%
47%
25%
6%
27%
100%
Percent
Contribution
from On-road
Mobile Sources
49%
41%
27%
14%
3%
21%
38%
This POM inventory includes the 15 POM compounds: benzo[b]fluoranthene, benz[a]anthracene,
indeno(l,2,3-c,d)pyrene, benzo[k]fluoranthene, chrysene, benzo[a]pyrene, dibenz(a,h)anthracene,
anthracene, pyrene, benzo(g,h,i)perylene, fluoranthene, acenaphthylene, phenanthrene, fluorene, and
acenaphthene.
EPA has assigned an overall confidence level for each pollutant in NATA based on consideration of the
combined uncertainties from emissions estimation, ambient concentration modeling, and exposure
modeling. These judgments refer to the relative confidence between two air toxics compounds. A judgment
of "Higher" (H) means the confidence is higher for this compound than for compounds assigned a
"Medium" (M) or "Lower" (L).
1-10
-------
Final Regulatory Impact Analysis
1.2. Dose-Response and Agency Risk Assessment Practice
This section describes EPA's formal process for conducting risk assessment. The
EPA framework for assessing and managing risks reflects the risk assessment and risk
management paradigm set forth by the National Academy of Sciences in 19833 which
was incorporated into the 1986 EPA risk guidance4 and revised in 2005 in the EPA
Guidelines for Carcinogen Risk Assessment and Supplemental Guidance for Assessing
Susceptibility from Early-Life Exposure to Carcinogens.5 The paradigm divides the risk
assessment and management process into four general phases. The first three phases
(exposure assessment, dose-response assessment, and risk characterization) comprise risk
assessment. The fourth phase, risk management, involves evaluation of information
provided by the risk assessment to the environmental manager who makes a risk
management decision.
An exposure assessment is the quantitative or qualitative evaluation of contact to
a specific pollutant and includes such characteristics as intensity, frequency, and duration
of contact. The numerical output of an exposure assessment may be either exposure or
dose, depending on the purpose of the evaluation and available data.
The dose-response assessment produces two sequential analyses. The first
analysis is the hazard identification, which identifies contaminants that are suspected to
pose health hazards, describes the specific forms of toxicity (e.g., neurotoxicity,
carcinogenicity, etc.) that they may cause, and evaluates the conditions under which these
forms of toxicity might be expressed in exposed humans. The types of effects that are
relevant to a particular chemical (e.g., cancer, noncancer) are determined as part of the
hazard identification.
The second analysis is the human health dose-response assessment, which
generally describes the characterization of the relationship between the concentration,
exposure, or dose of a pollutant and the resultant health effects. Dose-response
assessment methods generally consist of two parts. First is the evaluation of the
experimentally observed relationship between health effects and the concentration,
exposure and/or dose of a particular compound, and second is the extrapolation from the
observed range to lower doses and risks.
1.2.1. Cancer
The term 'cancer' is used to describe a group of related diseases that affect a
variety of organs and tissues. Cancer results from a combination of genetic damage and
nongenetic factors that favor the growth of damaged cells. The EPA document,
Guidelines for Carcinogen Risk Assessment6 (2005) provides guidance on hazard
identification for carcinogens. The approach recognizes three broad categories of data:
(1) human data (primarily, epidemiological); (2) results of long-term experimental animal
bioassays; and (3) supporting data, including a variety of short-term tests for genotoxicity
and other relevant properties. The 2005 Guidelines for hazard identification recommend
that an agent's human carcinogenic potential be described in a weight-of-evidence
1-11
-------
Final Regulatory Impact Analysis
narrative. The narrative summarizes the full range of available evidence and describes
any conditions associated with conclusions about an agent's hazard potential (e.g.,
carcinogenic by some routes of exposure and not others). To provide additional clarity
and consistency in weight-of-evidence narratives, the Guidelines suggest a set of weight-
of-evidence descriptors to accompany the narratives. The five descriptors are:
Carcinogenic to Humans, Likely to be Carcinogenic to Humans, Suggestive Evidence of
Carcinogenic Potential, Inadequate Information to Assess Carcinogenic Potential, and
Not Likely to be Carcinogenic to Humans. These descriptors replace those based on the
EPA 1986 Risk Assessment Guidelines which classified a compound as Group A:
Carcinogenic to Humans, Group B: Probably Carcinogenic to Humans, Group C:
Possibly Carcinogenic to Humans, Group D: Not Classifiable as to Human
Carcinogenicity, or Group E: Evidence of Noncarcinogenicity for Humans.
A quantitative assessment is performed depending on the weight-of-evidence and
the suitability of the available information regarding a relationship between the dose of a
compound and the effect it causes (dose-response data). Dose-response models are used
to calculate unit risk estimates (URE). Inhalation cancer risks are quantified by EPA
using the unit risk, which represent the excess lifetime cancer risk estimated to result
from continuous lifetime exposure to an agent at a concentration of 1 |ig/m3 in air. These
unit risks are typically upper-bound estimates, although where there are adequate
epidemiological data, the unit risk may be based on a maximum likelihood estimate
(MLE). Except for benzene and chromium, where risks are based on maximum
likelihood dose-response values, risks from mobile source air toxics should all be
considered upper-bound values. This means they are plausible upper limits to risks. True
risks could be greater, but are likely to be lower, and could be zero. A discussion of the
confidence in a quantitative cancer risk estimate is provided in the IRIS file for each
compound. The discussion of the confidence in the cancer risk estimate includes an
assessment of the source of the data (human or animal), uncertainties in dose estimates,
choice of the model used to fit the exposure and response data and how uncertainties and
potential confounders are handled.
The 2005 Guidelines include Supplemental Guidance for Assessing Susceptibility
from Early-Life Exposure to Carcinogens.7 The Supplemental Guidance is part of EPA's
response to the recommendation of the National Research Council (1994) that "EPA
should assess risks to infants and children whenever it appears that their risks might be
greater than those of adults." For several potential carcinogens, there is some evidence of
higher cancer risks following early-life exposure. Accordingly, the Supplemental
Guidance describes the approaches that EPA could use in assessing cancer risks
following early-life exposures. The 1999 NATA does not include default adjustments for
early life exposures recently recommended in the Supplemental Guidance. Incorporation
of such adjustments, if needed, would lead to higher estimates of lifetime risk.
1-12
-------
Final Regulatory Impact Analysis
1.2.2. Chronic Exposure and Noncancer Health Effects
Noncancer effects resulting from chronic exposures include a wide range of
effects in many organ systems, e.g., respiratory, cardiovascular, immune, kidney. Hazard
identification procedures for chronic noncancer effects are described in EPA guidelines.
The EPA has published guidelines for assessing several specific types of noncancer
effects, including mutagenicity,8 developmental toxicity,9 neurotoxicity10; and
reproductive toxicity.u For identification of hazards resulting from long-term (chronic)
exposures, EPA reviews available data on different health endpoints and target organs
and describes the range of effects observed and the related dose/exposure levels. EPA
focuses particular attention to effects that occur at relatively low doses or that may have
particular relevance to human populations. The inhalation reference concentration (RfC)
and oral reference dose (RfD) are the Agency consensus quantitative toxicity values for
use in chronic noncancer risk assessment. The RfC or RfD is defined as an estimate, with
uncertainty spanning perhaps an order of magnitude, of an inhalation exposure/oral dose
to the human population (including sensitive subgroups) that is likely to be without
appreciable risks of deleterious effects during a lifetime. The RfC or RfD is derived using
1) a thorough review of the health effects database for an individual chemical and 2) the
most sensitive and relevant endpoint and the principal study(ies) demonstrating that
endpoint. RfCs for inhalation are derived according to the Agency's 1994 guidance.12 A
statement regarding the confidence in the RfC and/or RfD is developed to reflect the
confidence in the principal study or studies on which the RfC or RfD are based and the
confidence in the underlying database. Factors that affect the confidence in the principal
study include how well the study was designed, conducted and reported. Factors that
affect the confidence in the database include an assessment of the availability of
information regarding identification of the critical effect, potentially susceptible
populations and exposure scenarios relevant to assessment of risk. In 2002 an EPA
RfC/RfD Technical Panel prepared several recommendations for preparation of
noncancer reference values.13
1.2.3. Acute Exposure and Noncancer Health Effects
Noncancer health impacts resulting from acute (short-term) exposures have been
assessed for many compounds in the occupational setting. EPA currently does not have
acute exposures reference values in IRIS comparable to the RfC described above. EPA's
Office of Research and Development proposed an Acute Reference Exposure (ARE)
approach for evaluating short term exposure effects in 1998.14 In 2002 EPA completed a
review document which summarizes recommendations of the EPA RfC/RfD Technical
Panel for preparation of noncancer reference values including acute exposure values.15
In response to the EPA Science Advisory Board review of the Acute Reference Exposure
methodology and recommendations from EPA's RfC/RfD Technical Panel, ORD is
currently developing an advanced acute inhalation reference concentration (acute RfC)
methodology. As part of this new methodology, acute inhalation assessments are being
developed for a few selected compounds including acrolein and hydrogen sulfide.
1-13
-------
Final Regulatory Impact Analysis
1.3. Summary of Air Toxic Health Effects
From a public health perspective, it is important to assess the emission
contributions to atmospheric levels of various air toxics (including diesel PM and exhaust
organic gases) emitted by motor vehicle engines, including their physical properties,
sources of potential exposure, and health hazards. In this section, we describe the cancer
and noncancer health effects attributed to chronic exposure to various mobile source air
toxics as well as any acute exposure health effects, where data are available. We focus
here on the air toxics that are identified in the NATA as risk drivers (see Section 1.1) and
that account for a significant share of mobile sources emissions. We also consider
compounds for which we expect emission reductions from today's proposed rule. We are
also including diesel paniculate matter and diesel exhaust organic gases in this discussion.
EPA has concluded that diesel exhaust ranks with the other substances that the national-
scale assessment suggests pose the greatest relative risk.
1.3.1. Benzene
Benzene is an aromatic hydrocarbon that is present as a gas in both exhaust and
evaporative emissions from mobile sources. Inhalation is the major source of human
exposure to benzene in the occupational and non-occupational setting.
The EPA's IRIS database lists benzene as a known human carcinogen (causing
leukemia) by all routes of exposure.16 A number of adverse noncancer health effects
including blood disorders and immunotoxicity, have also been associated with long-term
occupational exposure to benzene.
Long-term occupational inhalation exposure to benzene has been shown to cause
cancers of the hematopoetic (blood cell) system in adults. Among these are acute
nonlymphocytic leukemia,E and chronic lymphocytic leukemia.17'18 A doubling of risk
for acute nonlymphocytic leukemia and myelodysplastic syndrome was found at average
exposure levels under 10 ppm (32 mg/m3).19 EPA has not formally evaluated this study
as part of the IRIS review process. Leukemias, lymphomas, and other tumor types have
been observed in experimental animals exposed to benzene by inhalation or oral
administration. Exposure to benzene and/or its metabolites has also been linked with
E Leukemia is a blood disease in which the white blood cells are abnormal in type or number.
Leukemia may be divided into nonlymphocytic (granulocytic) leukemias and lymphocytic leukemias.
Nonlymphocytic leukemia generally involves the types of white blood cells (leukocytes) that are involved
in engulfing, killing, and digesting bacteria and other parasites (phagocytosis) as well as releasing
chemicals involved in allergic and immune responses. This type of leukemia may also involve
erythroblastic cell types (immature red blood cells). Lymphocytic leukemia involves the lymphocyte type
of white blood cell that is responsible for antibody and cell-mediated immune responses. Both
nonlymphocytic and lymphocytic leukemia may, in turn, be separated into acute (rapid and fatal) and
chronic (lingering, lasting) forms. For example; in acute myeloid leukemia there is diminished production
of normal red blood cells (erythrocytes), granulocytes, and platelets (control clotting), which leads to death
by anemia, infection, or hemorrhage. These events can be rapid. In chronic myeloid leukemia (CML) the
leukemic cells retain the ability to differentiate (i.e., be responsive to stimulatory factors) and perform
function; later there is a loss of the ability to respond.
1-14
-------
Final Regulatory Impact Analysis
20, 21
chromosomal changes in humans and animals ' and increased proliferation of mouse
fjrj ry^
bone marrow cells. '
The latest assessment by EPA estimates the excess risk of developing leukemia
from inhalation exposure to benzene at 2.2 x 10"6 to 7.8 x 10"6 per |ig/m3. In other words,
there is an estimated risk of about two to eight excess leukemia cases in one million
people exposed to 1 |ig/m3 of benzene over a lifetime.24 This range of unit risks reflects
the MLEs calculated from different exposure assumptions and dose-response models that
are linear at low doses. At present, the true cancer risk from exposure to benzene cannot
be ascertained, even though dose-response data are used in the quantitative cancer risk
analysis, because of uncertainties in the low-dose exposure scenarios and lack of clear
understanding of the mode of action. A range of estimates of risk is recommended, each
having equal scientific plausibility. There are confidence intervals associated with the
MLE range that reflect random variation of the observed data. For the upper end of the
MLE range, the 5th and 95th percentile values are about a factor of 5 lower and higher
than the best fit value. The upper end of the MLE range (7.8 x 10"6 per |ig/m3) was used
in the 1999NATA.
It should be noted that not enough information is known to determine the
slope of the dose-response curve at environmental levels of exposure and to provide a
sound scientific basis to choose any particular extrapolation/exposure model to estimate
human cancer risk at low doses. EPA risk assessment guidelines suggest using an
assumption of linearity of dose response when (1) there is an absence of sufficient
information on modes of action or (2) the mode of action information indicates that the
dose-response curve at low dose is or is expected to be linear.25 Data that were
considered by EPA in its carcinogenic update suggested that the dose-response
relationship at doses below those examined in the studies reviewed in EPA's most recent
benzene assessment may be supralinear. This relationship could support the inference
that cancer risks are as high, or higher than the estimates provided in the existing EPA
assessment.26 However, since the mode of action for benzene carcinogenicity is
unknown, the current cancer unit risk estimate assumes linearity of the low-dose response.
Data discussed in the EPA IRIS assessment suggest that genetic abnormalities occur at
low exposure in humans, and the formation of toxic metabolites plateaus above 25 ppm
(80,000 jig/m3).27 More recent data on benzene adducts in humans, published after the
most recent IRIS assessment, suggest that the enzymes involved in benzene metabolism
start to saturate at exposure levels as low as 1 ppm.28'29'30 These data highlight the
importance of ambient exposure levels and their contribution to benzene-related adducts.
Because there is a transition from linear to saturable metabolism below 1 ppm, the
assumption of low-dose linearity extrapolated from much higher exposures could lead to
substantial underestimation of leukemia risks. This is consistent with recent
epidemiological data which also suggest a supralinear exposure-response relationship and
which "[extend] evidence for hematopoietic cancer risks to levels substantially lower than
had previously been established".31'32'33 These data are from the largest cohort study
done to date with individual worker exposure estimates. However, these data have not
yet been formally evaluated by EPA as part of the IRIS review process, and it is not clear
how they might influence low-dose risk estimates. A better understanding of the
1-15
-------
Final Regulatory Impact Analysis
biological mechanism of benzene-induced leukemia is needed.
Children may represent a subpopulation at increased risk from benzene exposure,
due to factors that could increase their susceptibility. Children may have a higher unit
body weight exposure because of their heightened activity patterns which can increase
their exposures, as well as different ventilation tidal volumes and frequencies, factors that
influence uptake. This could entail a greater risk of leukemia and other toxic effects to
children if they are exposed to benzene at similar levels as adults. There is limited
information from two studies regarding an increased risk to children whose parents have
been occupationally exposed to benzene.34'35 Data from animal studies have shown
benzene exposures result in damage to the hematopoietic (blood cell formation) system
during development.36'37'38 Also, key changes related to the development of childhood
leukemia occur in the developing fetus.39 Several studies have reported that genetic
changes related to eventual leukemia development occur before birth. For example, there
is one study of genetic changes in twins who developed T cell leukemia at 9 years of
age.40 An association between traffic volume, residential proximity to busy roads and
occurrence of childhood leukemia has also been identified in some studies, although
some studies show no association. These studies are discussed later in Chapter 3.
A number of adverse noncancer health effects, including blood disorders such as
preleukemia and aplastic anemia, have also been associated with long-term exposure to
benzene.41'42 People with long-term occupational exposure to benzene have experienced
harmful effects on the blood-forming tissues, especially in the bone marrow. These
effects can disrupt normal blood production and suppress the production of important
blood components, such as red and white blood cells and blood platelets, leading to
anemia (a reduction in the number of red blood cells), leukopenia (a reduction in the
number of white blood cells), or thrombocytopenia (a reduction in the number of blood
platelets, thus reducing the ability of blood to clot). Chronic inhalation exposure to
benzene in humans and animals results in pancytopenia,F a condition characterized by
decreased numbers of circulating erythrocytes (red blood cells), leukocytes (white blood
cells), and thrombocytes (blood platelets).43'44 Individuals that develop pancytopenia and
have continued exposure to benzene may develop aplastic anemia, whereas others exhibit
both pancytopenia and bone marrow hyperplasia (excessive cell formation), a condition
that may indicate a preleukemic state.45' 46 The most sensitive noncancer effect observed
in humans, based on current data, is the depression of the absolute lymphocyte count in
blood.47' 48
EPA's inhalation reference concentration (RfC) for benzene is 30 |ig/m3. The
overall confidence in this RfC is medium. The RfC is based on suppressed absolute
lymphocyte counts seen in humans under occupational exposure conditions. Since
F Pancytopenia is the reduction in the number of all three major types of blood cells (erythrocytes,
or red blood cells, thrombocytes, or platelets, and leukocytes, or white blood cells). In adults, all three
major types of blood cells are produced in the bone marrow of the skeletal system. The bone marrow
contains immature cells, known as multipotent myeloid stem cells, that later differentiate into the various
mature blood cells. Pancytopenia results from a reduction in the ability of the red bone marrow to produce
adequate numbers of these mature blood cells.
1-16
-------
Final Regulatory Impact Analysis
development of this RfC, there have appeared reports in the medical literature of
benzene's hematotoxic effects in humans that provide data suggesting a wide range of
hematological endpoints that are triggered at occupational exposures of less than 5 ppm
(about 16 mg/m3)49 and, more significantly, at air levels of 1 ppm (about 3 mg/m3) or less
among genetically susceptible populations.50 These studies had large sample sizes and
extensive individual exposure monitoring. One recent study found benzene metabolites
in mouse liver and bone marrow at environmental doses, indicating that even
concentrations in urban air may elicit a biochemical response in rodents that indicates
toxicity.51 EPA has not formally evaluated these recent studies as part of the IRIS review
process to determine whether or not they will lead to a change in the current RfC. EPA
does not currently have an acute reference concentration for benzene. The Agency for
Toxic Substances and Disease Registry Minimal Risk Level for acute exposure to
benzene is 160 |ig/m3 for 1-14 days exposure.
1.3.2. 1,3-Butadiene
1,3-butadiene is formed in engine exhaust by the incomplete combustion of fuel.
It is not present in engine evaporative emissions because it is not generally present in an
appreciable amount in vehicle fuels.
EPA has characterized 1,3-butadiene as a leukemogen, carcinogenic to humans by
rrj r--)
inhalation. ' The specific mechanisms of 1,3-butadiene-induced carcinogenesis are
unknown however, it is virtually certain that the carcinogenic effects are mediated by
genotoxic metabolites of 1,3-butadiene. Animal data suggest that females may be more
sensitive than males for cancer effects; nevertheless, there are insufficient data from
which to draw any conclusions on potentially sensitive subpopulations. The upper bound
cancer unit risk estimate is 0.08 per ppm or 3xlO"5 per |ig/m3 (based primarily on linear
modeling and extrapolation of human data). In other words, it is estimated that
approximately 30 persons in one million exposed to 1 |ig/m3 of 1,3-butadiene
continuously for their lifetime would develop cancer as a result of this exposure. The
human incremental lifetime unit cancer risk estimate is based on extrapolation from
leukemias observed in an occupational epidemiologic study.54'55'56 This estimate
includes a two-fold adjustment to the epidemiologic-based unit cancer risk applied to
reflect evidence from the rodent bioassays suggesting that the epidemiologic-based
estimate (from males) may underestimate total cancer risk from 1,3-butadiene exposure
in the general population, particularly for breast cancer in females.57
A recent study extended the investigation of 1,3-butadiene exposure and leukemia
among synthetic rubber industry workers.58 The results of this study strengthen the
evidence for the relationship between 1,3-butadiene exposure and lymphohematopoietic
cancer. This relationship was found to persist after controlling for exposure to other
toxics in this work environment.
1,3-Butadiene also causes a variety of reproductive and developmental effects in
mice; no human data on these effects are available. The most sensitive effect was ovarian
atrophy observed in a lifetime bioassay of female mice.59 Based on this critical effect
1-17
-------
Final Regulatory Impact Analysis
and the benchmark concentration methodology, an RfC for chronic health effects was
calculated at 0.9 ppb (approximately 2 jig/m3). Confidence in the inhalation RfC is
medium.
1.3.3. Formaldehyde
Formaldehyde is the most prevalent aldehyde in engine exhaust. It is formed as a
result of incomplete fuel combustion in both gasoline and diesel engines, although
formaldehyde accounts for a smaller quantity of total exhaust hydrocarbons from
gasoline engines. Formaldehyde emissions can vary substantially by engine duty cycle,
emission control system and composition of fuel. Formaldehyde is not a component of
evaporative emissions but it can be formed photochemically in the atmosphere.
Since 1987, EPA has classified formaldehyde as a probable human carcinogen
based on evidence in humans and in rats, mice, hamsters, and monkeys.60 EPA's
current IRIS summary provides an upper bound cancer unit risk estimate of 1.3xlO"5 per
|ig/m3.G In other words, there is an estimated risk of about thirteen excess leukemia
cases in one million people exposed to 1 |ig/m3 of formaldehyde over a lifetime. EPA is
currently reviewing recently published epidemiological data. For instance, research
conducted by the National Cancer Institute (NCI) found an increased risk of
nasopharyngeal cancer and lymphohematopoietic malignancies such as leukemia among
workers exposed to formaldehyde.61' 62 NCI is currently performing an update of these
studies. A recent National Institute of Occupational Safety and Health (NIOSH) study of
garment workers also found increased risk of death due to leukemia among workers
exposed to formaldehyde.63 Extended follow-up of a cohort of British chemical workers
did not find evidence of an increase in nasopharyngeal or lymphohematopoeitic cancers,
but a continuing statistically significant excess in lung cancers was reported.64
Based on the developments of the last decade, in 2004, the working group of the
International Agency for Research on Cancer concluded that formaldehyde is
carcinogenic to humans (Group 1 classification), on the basis of sufficient evidence in
humans and sufficient evidence in experimental animals—a higher classification than
previous IARC evaluations. In addition, the National Institute of Environmental Health
Sciences recently nominated formaldehyde for reconsideration as a known human
carcinogen under the National Toxicology Program. Since 1981 it has been listed as a
"reasonably anticipated human carcinogen." Recently the German Federal Institute for
Risk Assessment determined that formaldehyde is a known human carcinogen.65
In the past 15 years there has been substantial research on the inhalation
dosimetry for formaldehyde in rodents and primates by the CUT Centers for Health
Research (formerly the Chemical Industry Institute of Toxicology), with a focus on use of
rodent data for refinement of the quantitative cancer dose-response assessment.66'67'68
CIIT's risk assessment of formaldehyde incorporated mechanistic and dosimetric
information on formaldehyde. The risk assessment analyzed carcinogenic risk from
G U.S. EPA (1989). Integrated Risk Information System File for Formaldehyde. This material is
available electronically at http://www.epa.gov/iris/subst/0419.htm.
1-18
-------
Final Regulatory Impact Analysis
inhaled formaldehyde using approaches that are consistent with EPA's draft guidelines
for carcinogenic risk assessment. In 2001, Environment Canada relied on this cancer
dose-response assessment in their assessment of formaldehyde.69 In 2004, EPA also
relied on this cancer unit risk estimate during the development of the plywood and
composite wood products national emissions standards for hazardous air pollutants
(NESHAPs).70 In these rules, EPA concluded that the CUT work represented the best
available application of the available mechanistic and dosimetric science on the dose-
response for portal of entry cancers due to formaldehyde exposures. EPA is reviewing
the recent work cited above from the NCI and NIOSH, as well as the analysis by the CUT
Centers for Health Research and other studies, as part of a reassessment of the human
hazard and dose-response associated with formaldehyde.
Noncancer effects of formaldehyde have been observed in humans and several
animal species and include irritation to eye, nose and throat tissues in conjunction with
increased mucous secretions.71
1.3.4. Acetaldehyde
Acetaldehyde is formed as a result of incomplete fuel combustion in both gasoline
and diesel engines, although acetaldehyde accounts for a smaller quantity of total exhaust
hydrocarbons from gasoline engines. Acetaldehyde emissions can vary substantially by
engine duty cycle, emission control system and composition of fuel. Acetaldehyde is not
a component of evaporative emissions but it can be formed photochemically in the
atmosphere.
Acetaldehyde is classified in EPA's IRIS database as a probable human
carcinogen and is considered toxic by inhalation.72 Based on nasal tumors in rodents, the
upper confidence limit estimate of a lifetime extra cancer risk from continuous
acetaldehyde exposure is about 2.2xlO"6 per |ig/m3. In other words, it is estimated that
about 2 persons in one million exposed to 1 |ig/m3 acetaldehyde continuously for their
lifetime (70 years) would develop cancer as a result of their exposure although the risk
could be as low as zero.
In short-term (4 week) rat studies, compound-related histopathological changes
were observed only in the respiratory system at various concentration levels of
exposure.73'74 Data from these studies showing degeneration of the olfactory epithelium
were found to be sufficient for EPA to develop an RfC for acetaldehyde of 9 |ig/m3.
Confidence in the principal study is medium and confidence in the database is low, due to
the lack of chronic data establishing a no observed adverse effect level and due to the
lack of reproductive and developmental toxicity data. Therefore, there is low confidence
in the RfC.75 The Agency is currently conducting a reassessment of risk from inhalation
exposure to acetaldehyde.
The primary acute effect of exposure to acetaldehyde vapors is irritation of the
eyes, skin, and respiratory tract.76 Some asthmatics have been shown to be a sensitive
1-19
-------
Final Regulatory Impact Analysis
subpopulation to decrements in functional expiratory volume (FEV1 test) and broncho-
constriction upon acetaldehyde inhalation.77
1.3.5. Acrolein
Acrolein is found in vehicle exhaust and is formed as a result of incomplete
combustion of both gasoline and diesel fuel. It is not a component of evaporative
emissions but it can be formed photochemically from 1,3-butadiene in the atmosphere.
EPA determined in 2003 using the 1999 draft cancer guidelines that the human
carcinogenic potential of acrolein could not be determined because the available data
were inadequate. No information was available on the carcinogenic effects of acrolein in
humans and the animal data provided inadequate evidence of carcinogenicity.
Acrolein is an extremely volatile organic compound which possesses considerable
water solubility.78 As such, it readily absorbs into airway fluids in the respiratory tract
when inhaled. The toxicological data base demonstrating the highly irritating nature of
this vapor has been consistent, regardless of test species. Acrolein is intensely irritating
to humans when inhaled, with acute exposure resulting in upper respiratory tract irritation,
mucus hypersecretion and congestion.
Lesions to the lungs and upper respiratory tract of rats, rabbits, and hamsters
exposed to acrolein formed the basis of the reference concentrations for inhalation (RfC)
developed in 2003.79 The Agency has developed an RfC for acrolein of 0.02 |ig/m3 and
an RfD of 0.5 ug/kg-day.80 The overall confidence in the RfC assessment is judged to be
medium and the confidence in the RfD is medium to high.
The Agency is currently in the process of conducting an assessment of acute
exposure effects for acrolein. The intense irritancy of this carbonyl has been
demonstrated during controlled tests in human subjects who suffer intolerable eye and
nasal mucosal sensory reactions within minutes of exposure.81
1.3.6. Naphthalene
Naphthalene is found in small quantities in gasoline and diesel fuels.
Naphthalene emissions have been measured in larger quantities in both gasoline and
diesel exhaust and evaporative emissions from mobile sources.
In 2004, EPA released an external review draft of a reassessment of the inhalation
carcinogenicity of naphthalene.82 The draft reassessment (External Review Draft, IRIS
Reassessment of the Inhalation Carcinogenicity of Naphthalene) completed external peer
review in 2004 by Oak Ridge Institute for Science and Education.83 Based on external
comments, additional analyses are being considered. California EPA has released a new
risk assessment for naphthalene with a cancer unit risk estimate of 3xlO"5 per |ig/m3.84
The California EPA value was used in the 1999 NATA and in the analyses done for this
rule. In addition, IARC has reevaluated naphthalene and re-classified it as Group 2B:
1-20
-------
Final Regulatory Impact Analysis
possibly carcinogenic to humans.85 Current risk estimates for naphthalene are based on
extrapolations from rodent studies conducted at higher doses. At present, human data are
inadequate for developing estimates.
The current EPA IRIS assessment includes noncancer data on hyperplasia and
metaplasia in nasal tissue that form the basis of the inhalation RfC of 3 |ig/m3.86 The
principal study was given medium confidence because adequate numbers of animals were
used, and the severity of nasal effects increased at the higher exposure concentration.
However, the study produced high mortality and hematological evaluation was not
conducted beyond 14 days. The database was given a low-to-medium confidence rating
because there are no chronic or subchronic inhalation studies in other animal species, and
there are no reproductive or developmental studies for inhalation exposure. In the
absence of human or primate toxicity data, the assumption is made that nasal responses in
mice to inhaled naphthalene are relevant to humans; however, it cannot be said with
certainty that this RfC for naphthalene based on nasal effects will be protective for
hemolytic anemia and cataracts, the more well-known human effects from naphthalene
exposure. As a result, we have medium confidence in the RfC.
1.3.7. 2,2,4-Trimethylpentane
2,2,4-Trimethylpentane is a colorless liquid hydrocarbon also known as isooctane,
isobutyltrimethylmethane, and TMP. Automotive exhaust and automotive evaporative
emissions are important sources of 2,2,4-trimethylpentane in the atmosphere.
EPA is in the process of assembling a review draft of a reassessment of its 1991
2,2,4-TMP health effects assessment in EPA's IRIS database. The earlier document
found little conclusive evidence of specific health effects associated with 2,2,4-TMP
exposures in humans.87 Overall, there was "inadequate information to assess
carcinogenic potential," in accordance with EPA's Guidelines for Carcinogen Risk
Assessment (U.S. EPA, 1986), for 2,2,4-trimethylpentane. No chronic bioassay studies
were available that assessed the carcinogenic effects of 2,2,4-trimethylpentane in humans.
Oral studies existed linking 2,2,4-TMP with male rat kidney toxicity and an
increase in alpha2U-globulin protein and hyaline droplet accumulation in the proximal
tubules of the kidneys.88 These effects were not seen in the female rat test subjects.
These renal effects, specific to the male rat, are not thought to be relevant to humans.
Inhalation studies in animals had been performed but none were adequate to calculate an
inhalation RfC for the compound.
1.3.8. Ethylbenzene
Ethylbenzene is present as in both gasoline and diesel exhaust and in evaporative
emissions from gasoline-powered vehicles.89 Limited information is available on the
carcinogenic effects of ethylbenzene in humans and animals. Under the 1987 Cancer
Guidelines, EPA has classified ethylbenzene as a Group D carcinogen, meaning it is not
classifiable as to human carcinogenicity. This classification is the result of inadequate
1-21
-------
Final Regulatory Impact Analysis
data from animal bioassays and human studies
90
Chronic (long-term) exposure to ethylbenzene by inhalation in humans may result
in effects on the blood, kidney and liver. No information is available on the
developmental or reproductive effects of ethylbenzene in humans, although animal
studies have reported developmental effects via inhalation. The data from these studies
were found to be sufficient for EPA to develop an RfC of 1x103 ug/m3 for ethylbenzene
exposure. Confidence in the RfC is considered low because higher study exposure levels
might have been more informative and no chronic studies or multi-generational
developmental studies were available at the time. Animal studies have reported effects on
the blood, liver, and kidneys from ingestion exposure to ethylbenzene. The data from
these studies were found to be sufficient for EPA to develop an RfD for ethylbenzene
exposure of 100 ug/kg-day. Confidence in this RfD is considered low because rats of
only one sex were tested, no chronic studies were then available, and no other oral
toxicity data were found. Ethylbenzene is currently undergoing an IRIS update for both
cancer and noncancer effects, based on new data.
Acute (short-term) exposure to ethylbenzene in humans results in noncancer
respiratory effects, such as throat irritation and chest constriction, irritation of the eyes,
and neurological effects such as dizziness.91
1.3.9. n-Hexane
n-Hexane is a component of gasoline and is also found in exhaust and evaporative
emissions from motor vehicles. Monitoring data indicate that n-hexane occurs widely in
the atmosphere.92
Under the 2005 Guidelines for Carcinogen Risk Assessment, there is inadequate
information to assess the carcinogenic potential of n-hexane.93 Chronic exposure to n-
hexane in air is associated with polyneuropathy in humans, with numbness in the
extremities, muscular weakness, blurred vision, headache, and fatigue observed.
Neurotoxic effects have also been exhibited in rats. Mild inflammatory and degenerative
lesions in the nasal cavity have been observed in rodents chronically exposed by
inhalation. Limited information is available on the reproductive or developmental effects
of n-hexane; one study reported testicular damage in rats exposed to n-hexane through
inhalation. Birth defects have not been observed in the offspring of rats chronically
exposed via inhalation in several studies. The data from a study of peripheral neuropathy
was used to develop an RfC of 700 ug/m3 for n-hexane exposure.94 This RfC has been
given a confidence rating of medium due to medium confidence in the underlying study
and medium confidence in the database. The database lacks chronic exposure
information on the pure compound via any route of exposure, a multigenerational
developmental and reproductive toxicity study and a developmental neurotoxicity study.
Acute inhalation exposure of humans to high levels of n-hexane causes mild
central nervous system (CNS) depression and irritation of the skin. Nervous system
effects include dizziness, giddiness, slight nausea, and headache in humans.95
1-22
-------
Final Regulatory Impact Analysis
1.3.10. Methyl Tertiary Butyl Ether (MTBE)
Methyl tert-buty\ ether (MTBE) has been used in the United States since the late-
1970's as an octane-enhancing agent in gasoline.
In 1994, EPA's Office of Research and Development concluded that, under the
1986 EPA cancer risk assessment guidelines, inhalation cancer test results support
placing MTBE in Group C as a "possible human carcinogen."96 An Interagency
Assessment of Oxygenated Fuels similarly concluded that "While there are no studies on
the carcinogenicity of MTBE in humans, there is sufficient evidence to indicate that
MTBE is an animal carcinogen and to regard MTBE as having a human hazard potential.
However, estimates of human risk from MTBE contain large uncertainties in both human
exposure and cancer potency."97 The Agency is currently conducting a reassessment of
MTBE.
By the inhalation route, MTBE has been found to cause increases in liver and
kidney weights and increased severity of spontaneous kidney lesions, as well as swelling
around the eyes and increased prostration in laboratory rats98. These effects are cited as
the basis for EPA's current inhalation reference concentration (RfC) of 3 mg/m3 for
MTBE. The RfC has a medium to high confidence rating.
1.3.11. Styrene
Styrene is found in the exhaust from both gasoline- and diesel-powered engines.
Several epidemiologic studies suggest that there may be an association between styrene
exposure and an increased risk of leukemia and lymphoma. However, the evidence is
inconclusive due to confounding factors. Animal studies have produced both negative
and positive results. EPA is currently assessing the potential of styrene to cause cancer.
Chronic exposure of humans to styrene results in effects on the central nervous
system (CNS), such as headache, fatigue, weakness, depression, peripheral neuropathy,
minor effects on some kidney enzyme functions and on the blood. Human studies are
inconclusive on the reproductive and developmental effects of styrene. The data from
human studies looking at central nervous system effects was found to be sufficient for
EPA to develop an RfC of 1 mg/m3 for styrene exposure. The RfC is assigned an overall
confidence rating of medium. Data from animal oral exposure studies was found to be
sufficient for EPA to also develop an RfD of 200 ug/kg-day for styrene oral exposure.
The RfD is assigned an overall confidence rating of medium.
Acute exposure to styrene results in mucous membrane and eye irritation, and
central nervous system effects in humans. 99' 10°
1.3.12. Toluene
Toluene is found in evaporative as well as exhaust emissions from motor vehicles.
1-23
-------
Final Regulatory Impact Analysis
Under the 2005 Guidelines for Carcinogen Risk Assessment, there is inadequate
information to assess the carcinogenic potential of toluene because studies of humans
chronically exposed to toluene are inconclusive, toluene was not carcinogenic in adequate
inhalation cancer bioassays of rats and mice exposed for life, and increased incidences of
mammary cancer and leukemia were reported in a lifetime rat oral bioassay.101
The central nervous system (CNS) is the primary target for toluene toxicity in
both humans and animals for acute and chronic exposures. CNS dysfunction (which is
often reversible) and narcosis have been frequently observed in humans acutely exposed
to low or moderate levels of toluene by inhalation; symptoms include fatigue, sleepiness,
headaches, and nausea. Central nervous system depression has been reported to occur in
chronic abusers exposed to high levels of toluene. Symptoms include ataxia, tremors,
cerebral atrophy, nystagmus (involuntary eye movements), and impaired speech, hearing,
and vision. Chronic inhalation exposure of humans to toluene also causes irritation of the
upper respiratory tract, eye irritation, dizziness, headaches, and difficulty with sleep.102
Human studies have also reported developmental effects, such as CNS
dysfunction, attention deficits, and minor craniofacial and limb anomalies, in the children
of women who abused toluene during pregnancy. A substantial database examining the
effects of toluene in subchronic and chronic occupationally exposed humans exists. The
weight of evidence from these studies indicates neurological effects (i.e., impaired color
vision, impaired hearing, decreased performance in neurobehavioral analysis, changes in
motor and sensory nerve conduction velocity, headache, dizziness) as the most sensitive
endpoint. The data from these human studies was found to be sufficient for EPA to
develop an RfC of 5 mg/m3 for toluene exposure. The overall confidence in this RfC is
high. Additional data from animal oral exposure studies was found to be sufficient for
EPA to also develop an RfD of 80 ug/kg-day for toluene oral exposure.103 The overall
confidence in the RfD is medium.
1.3.13. Xylenes
Mixed xylenes are blended into gasoline and are present in diesel fuels. Xylenes
are emitted in the exhaust emissions and evaporative emissions of both gasoline- and
diesel-powered engines.
Inadequate information is available on the carcinogenic effects of mixed xylenes
in humans, and animal studies have been inconclusive. Under the 1999 Draft Revised
Guidelines for Carcinogen Risk Assessment, data are inadequate for an assessment of the
carcinogenic potential of xylenes.104
Chronic inhalation exposure in humans to mixed xylenes results primarily in
central nervous system effects, such as headache, nausea, fatigue and also included eye
and nose irritation and sore throat.105 Animal studies have reported developmental
effects, such as an increased incidence of skeletal variations in fetuses, and fetal
resorptions via inhalation. EPA developed an RfC of 100 ug/m3 for xylenes based on
impaired motor coordination in rats. The confidence rating assigned to the RfC for
1-24
-------
Final Regulatory Impact Analysis
xylenes is medium. Data from animal oral exposure studies, looking at decreased body
weight and increased mortality were found to be sufficient for EPA to develop an RfD of
200 ug/kg-day for oral xylene exposure. The RfD was assigned an overall confidence
rating of medium.106
Acute inhalation exposure to mixed xylenes in humans results in irritation of the
nose and throat, gastrointestinal effects such as nausea, vomiting, and gastric irritation,
mild transient eye irritation, and neurological effects.107
1.3.14. Poly cyclic Organic Matter (POM)
POM is a class of chemicals consisting of organic compounds having multiple
benzene rings and boiling points in excess of 100 degrees Celsius. POM is a byproduct
of the incomplete combustion of fossil fuels and, as such, is a component of diesel and
gasoline engine emissions. At least eight of the compounds included in the class of
compounds known as POM are classified by EPA as probable human carcinogens based
on animal data. These include acenaphthene, benzo(a)anthracene, benzo(b)fluoranthene,
benzo(k)fluoranthene, benzo(a)pyrene, chrysene, dibenz(a,h)anthracene, and
indeno(l,2,3-cd)pyrene. One POM, naphthalene, is discussed separately in this section.
Recent studies have found that maternal exposures to polyaromatic hydrocarbons
(PAHs), a subclass of POM, in a population of pregnant women were associated with
several adverse birth outcomes, including low birth weight and reduced length at birth.108
These studies are discussed later in Chapter 3.
1.3.15. Diesel Exhaust
In EPA's Diesel Health Assessment Document (HAD),109 diesel exhaust was
classified as likely to be carcinogenic to humans by inhalation at environmental
exposures, in accordance with the revised draft 1996/1999 EPA cancer guidelines. A
number of other agencies (National Institute for Occupational Safety and Health, the
International Agency for Research on Cancer, the World Health Organization, California
EPA, and the U.S. Department of Health and Human Services) have made similar
classifications. EPA concluded in the Diesel HAD that it is not possible currently to
calculate a cancer unit risk for diesel exhaust due to a variety of factors that limit the
current studies, such as limited quantitative exposure histories in occupational groups
investigated for lung cancer.
However, in the absence of a cancer unit risk, the EPA Diesel HAD sought to
provide additional insight into the significance of the cancer hazard by estimating
possible ranges of risk that might be present in the population. An exploratory analysis
was used to characterize a possible risk range by comparing a typical environmental
exposure level for highway diesel sources to a selected range of occupational exposure
levels. The occupationally observed risks were then proportionally scaled according to
the exposure ratios to obtain an estimate of the possible environmental risk. A number of
calculations are needed to accomplish this, and these can be seen in the EPA Diesel HAD.
1-25
-------
Final Regulatory Impact Analysis
The outcome was that environmental risks from diesel exhaust exposure could range
from a low of 10"4 to 10"5 to as high as 10"3, reflecting the range of occupational
exposures that could be associated with the relative and absolute risk levels observed in
the occupational studies. Because of uncertainties, the analysis acknowledged that the
risks could be lower than 10"4 or 10"5, and a zero risk from diesel exhaust exposure was
not ruled out.
Noncancer health effects of acute and chronic exposure to diesel exhaust
emissions are also of concern to the Agency. EPA derived an RfC from consideration of
four well-conducted chronic rat inhalation studies showing adverse pulmonary effects.110'
in, 112, 113 The Rfc ig 5 ^m3 for diegel exhaust as measurecj by diesel PM. This RfC
does not consider allergenic effects such as those associated with asthma or immunologic
effects. There is growing evidence, discussed in the Diesel HAD, that diesel exhaust can
exacerbate these effects, but the exposure-response data are presently lacking to derive an
RfC. The EPA Diesel HAD states, "With DPM [diesel paniculate matter] being a
ubiquitous component of ambient PM, there is an uncertainty about the adequacy of the
existing DE [diesel exhaust] noncancer database to identify all of the pertinent DE-caused
noncancer health hazards" (p. 9-19).
The Diesel HAD also briefly summarizes health effects associated with ambient
PM and discusses the EPA's annual National Ambient Air Quality Standard (NAAQS) of
15 |ig/m3. There is a much more extensive body of human data showing a wide spectrum
of adverse health effects associated with exposure to ambient PM, of which diesel
exhaust is an important component. The PM2.s NAAQS is designed to provide protection
from the noncancer and premature mortality effects of PM2.5 as a whole, of which diesel
PM is a constituent.
1.4. Emerging Issues
Beyond the specific areas of quantifiable risk discussed above in Chapter 1.1.2,
EPA is interested in emerging mobile source toxics issues that might require action in the
future. The emerging issues currently under investigation by EPA are gasoline PM and
metals.
1.4.1. Gasoline PM
Gasoline exhaust is a complex mixture that has not been evaluated in EPA's IRIS.
Gasoline exhaust is a ubiquitous source of particulate matter, contributing to the health
effects observed for ambient PM which is discussed extensively in the EPA Particulate
Matter Criteria Document.114 The PM Criteria Document notes that the PM components
of gasoline and diesel engine exhaust are hypothesized, important contributors to the
observed increases in lung cancer incidence and mortality associated with ambient
PM2.s.115 Gasoline PM is also a component of near-roadway emissions that may be
contributing to the health effects observed in people who live near roadways (see Chapter
3.1.3.1). There is also emerging evidence for the mutagenicity and cytotoxicity of
gasoline exhaust and gasoline PM. Seagrave et al. investigated the combined parti culate
1-26
-------
Final Regulatory Impact Analysis
and semivolatile organic fractions of gasoline and diesel engine emissions in various
animal and bioassay tests.116 The authors suggest that emissions from gasoline engines
(including both the semi-volatile organic compounds and the particulate matter) are
mutagenic and can induce inflammation and have cytotoxic effects.
EPA is working to improve the understanding of PM emissions from gasoline
engines, including the potential range of emissions and factors that influence emissions.
EPA led a large cooperative test program that recently completed testing approximately
500 randomly procured vehicles in the Kansas City metropolitan area. The purpose of
this study was to determine the distribution of gasoline PM emissions from the in-use
light-duty fleet. Results from this study are expected to be available shortly. This work
shows how PM emissions vary for light-duty gasoline vehicles (automobiles and light-
duty trucks) for different model years. It also shows how colder temperatures increase
gasoline PM emissions. The data from this program are being evaluated. Some source
apportionment studies in various areas of the country, including Denver and California,
show gasoline and diesel PM can result in larger contributions to ambient PM than
predicted by EPA emission inventories.117'118 These source apportionment studies were
one impetus behind the Kansas City study.
Another issue related to gasoline PM is the effect of mobile source on ambient
PM, especially secondary PM. Ambient PM is composed of primary PM emitted directly
into the atmosphere and secondary PM is formed in the atmosphere from chemical
reactions in the atmosphere. Sulfates and nitrates are major examples of inorganic
secondary PM, both of which have been well studied and quantified. Carbonaceous PM,
from both primary PM emissions and secondary PM formed in the atmosphere, is a major
source of PM, especially in urban areas. Various studies show that carbonaceous PM
specifically from mobile sources is a major PM constituent in many urban areas over
many portions of the country (including urban areas in the Northeast, Southeast, Midwest,
and California/Washington portions of the United States). This information is included
in EPA reports and various source apportionment studies.119'120'121'122'123'124'125
Primary carbonaceous mobile source emissions can be evaluated from emission
inventories. The ambient PM levels from these emissions and secondary PM formed in
the atmosphere from mobile sources can then be estimated by air quality modeling
studies using the CMAQ (Community Multi-scale Air Quality) model. In addition to
primary carbonaceous (organic aerosol) emissions, some specific compounds contribute
to atmospheric PM loadings via formation of secondary organic aerosols (SOA). These
compounds include monoterpenes and possibly isoprene and sesquiterpenes, as well as
anthropogenic aromatic hydrocarbons such as toluene (and probably higher molecular
weight non-aromatic hydrocarbons).
Smog chamber studies show that benzene forms SOA possibly through reactions
with NOx. Prior smog chamber work126 suggested benzene might be relatively inert in
forming SOA, although this early study may not be conclusive. However, the more
recent work shows that benzene does form SOA in smog chambers. This new smog
chamber work shows that benzene can be oxidized in the presence of NOx to form SOA
1-27
-------
Final Regulatory Impact Analysis
with maximum mass of SOA being 8-25% of the mass of benzene.127 Work is needed to
determine if a tracer compound can be found for benzene SOA which might indicate how
much of ambient SOA comes from benzene.
Upon release into the atmosphere, these numerous compounds can react with free
radicals in the atmosphere to form SOA. While SOA formation from many reactive
hydrocarbons has been investigated in the laboratory, there is relatively little information
available on the chemical composition of SOA compounds from specific hydrocarbon
precursors. This lack of information is largely due to having few reliable methods for
measuring the polar, high molecular weight compounds that are thought to make up much
of ambient SOA. The absence of compositional data has largely prevented identifying
aromatically-derived SOA in ambient samples which, in turn, has prevented observation-
based measurements of the aromatic and other SOA contributions to ambient PM levels.
Recently EPA has taken the first step in addressing these issues by developing a
tracer-based method for detecting SOA precursors in ambient samples. The method
consists of irradiating the SOA precursor of interest in a smog chamber in the presence of
NOx, collecting the SOA produced on filters, and then analyzing the samples for highly
polar compounds using advanced organic chemistry methods. Employing this method,
candidate tracers have been identified for several hydrocarbon compounds which are
emitted in significant quantities and known to produce SOA in the atmosphere. Some of
these compounds forming SOA that have been investigated in the current study are
toluene, a variety of monoterpenes, isoprene, and /?-caryophyllene, the latter three of
which are emitted by vegetation. 128' 129'130' 131; 132>133 The tracers provide a means to
identify the hydrocarbon SOA precursors present in ambient PM2.5 samples and show
promise for estimating their contributions to the organic carbon concentrations.
The results of a recent EPA field study, to be published in the peer-reviewed
literature, suggest aromatic hydrocarbon emissions, including toluene and possibly
xylenes, contribute to SOA in Research Triangle Park, North Carolina, with initial
estimates as high as 0.7 ug/m3 during smog events in July/August. The level of toluene-
derived SOA is the lowest in the November-February time frame (about 0.2 ug/m3) with
intermediate levels in the other months. Currently, EPA is conducting similar analyses of
ambient PM2.5 samples in Cincinnati, OH, Northbrook, IL, Detroit, MI, Bondville, IL,
and St. Louis, MO, the results of which will be available by the end of 2006. After
acceptance of the EPA field study results in the peer-reviewed literature, they will be
used to assess whether current treatment of aromatic SOA in the EPA CMAQ model need
to be modified. Along with most of the other state-of-the-science air quality models,
CMAQ predicts low levels of aromatic SOA.
One caveat regarding this work is that a large number of gaseous hydrocarbons
emitted into the atmosphere having the potential to form SOA have not yet been studied
in this way. It is possible that hydrocarbons which have not yet been studied produce
some of SOA species which are being used as tracers for other gaseous hydrocarbons.
This means that the present work could overestimate the amount of SOA in the
atmosphere to the gaseous hydrocarbons studied to date.
1-28
-------
Final Regulatory Impact Analysis
The issue of SOA formation from aromatic precursors is an important one to
which EPA and others are paying significant attention. Due to the large contribution of
mobile source emissions to overall aromatic levels in the atmosphere, this issue is a
crucial one for assessing what further reductions are possible in mobile source PM.
1.4.2. Metals
The emission of metals to the environment is receiving increasing attention.
Metals comprise a complex class of elements, some of which are toxic at very low
exposure levels. The chemical form in which a metal or metal compound is emitted often
determines the potential toxicity and ultimate fate of the element in the environment.
Research in recent years suggests that some metals (e.g., transition metals) play an
important role in the toxicity of ambient PM, and inhalation as well as ingestion of metals
is known to cause a diverse array of cancer and noncancer effects in mammals. Since
metals do not degrade in the environment, concerns arise regarding their accumulation in
plants, animals, soil and water. The emission of metals from mobile sources is an
emerging area of interest since the emissions are in the breathing zone and are distributed
in a concentrated fashion in the roadway environment.
Emission of metals from mobile sources occurs as the result of metallic impurities
in lubricating oil and fuel, catalyst wear, engine wear, brake wear, and tire wear.
Emission rates of most metals from mobile sources are quite low, presenting challenges
for many common measurement methods. In recent years, improvements in analytical
chemistry allow both the quantification of very low levels of metals in mobile source
exhaust as well as some characterization of the form of the metals emitted.134 Currently,
there are many gaps in our understanding of the quantity, chemical form and size
distribution of metals in exhaust or from tire and brake wear. Application of state-of-the-
art measurement techniques to mobile source metal emissions is just beginning. For
example, EPA is currently conducting an emissions characterization program to
understand the emission rate and chemical form of mercury in motor vehicle exhaust and
the total mercury concentration in gasoline, diesel fuel, lubricating oil, and brake wear
emissions. This work will help us understand the potential sources of motor vehicle
mercury emissions, and the contribution of motor vehicles relative to other sources of
mercury emissions. This information is necessary for any future consideration of control
options. Other metals are also being evaluated in various studies.
Metals can also be emitted from mobile sources as a result of their use as an
additive to gasoline and/or diesel fuel. As discussed in Chapter III.G of the preamble,
Clean Air Act section 211 provides EPA with the authority to require a fuel additive
manufacturer to collect necessary data to enable EPA to make a determination about the
potential for risk to public health.
1-29
-------
Final Regulatory Impact Analysis
References for Chapter 1
1 Rao, S.; Pollack, A.; Lindhjem, C. 2004 Expanding and updating the master list of
compounds emitted by mobile sources - Phase III Final Report. Environ International
Corporation.
2 www.epa.gov/ttn/atw/toxsource/summary.html. Tables of dose-response values on this
website, used in EPA Office of Air Quality Planning and Standards risk assessments, are
available in Docket EPA-HQ-OAR-2005-0036.
3 National Academy of Sciences. 1983. Risk Assessment in the Federal Government:
Managing the Process. Committee on the Institutional Means for Assessment of Risks to
Public Health, National Research Council. Found in various EPA library collections
through http://www.epa.gov/natlibra/ols.htm by its OCLC catalog no.09374015.
4 EPA. 1986. Guidelines for carcinogen risk assessment. Federal Register 51:33992-
34003. September 24.
5 EPA. 2005. Guidelines for carcinogen risk assessment and Supplemental Guidance for
Assessing Susceptibility from Early-Life Exposure to Carcinogens. EPA/630/P-03/001F.
6 EPA. 1986 Guidelines for carcinogen risk assessment. Federal Register 51:33992-
34003. September 24.
7 U. S. EPA. 2005 Supplemental Guidance for Assessing Susceptibility from Early-Life
Exposure to Carcinogens. Report No. EPA/630/R-03/003F.
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid= 116283
8 EPA. 1986. Guidelines for mutagenicity risk assessment. Federal Register 51:34006-
34012. Sept. 24.
9 EPA. 1991. Guidelines for developmental toxicity risk assessment. Federal Register
56:63798-63826.
10 EPA. 1998. Guidelines for neurotoxicity risk assessment. Federal Register 63:26926.
May 14. http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=12479.
11 EPA. 1996. Guidelines for reproductive toxicity risk assessment. EPA/630/R-96/009.
Federal Register 56274-56322, 31 October 1996.
http: //cfpub. epa. gov/ncea/cfm/recordi spl ay. cfm? dei d=2 838.
12 EPA. 1994. Methods for derivation of inhalation reference concentrations and
application of inhalation dosimetry. Washington D.C. EPA/600/8-90/066F.
13 EPA (2002) A review of the reference dose and reference concentration processes.
EPA/630/P-02/002F.
1-30
-------
Final Regulatory Impact Analysis
14 EPA (1998) Methods for exposure-response analysis for acute inhalation exposure to
chemicals: development of the acute reference exposure. Review draft. Office of
Research and Development, Washington, D.C. EPA/600/R-98/051.
15 EPA (2002) A review of the reference dose and reference concentration processes.
EPA/630/P-02/002F.
16 EPA 2005 "Full IRIS Summary for Benzene (CASRN 71-43-2)" Environmental
Protection Agency, Integrated Risk Information System (IRIS), Office of Health and
Environmental Assessment, Environmental Criteria and Assessment Office, Cincinnati,
OH http://www.epa.gov/iris/subst/0276.htm.
17 U.S. EPA (1985) Environmental Protection Agency, Interim quantitative cancer unit
risk estimates due to inhalation of benzene, prepared by the Office of Health and
Environmental Assessment, Carcinogen Assessment Group, Washington, DC. for the
Office of Air Quality Planning and Standards, Washington, DC., 1985. Document no.
EPA600-X-85-022.
18 Clement Associates, Inc. (1991) Motor vehicle air toxics health information, for U.S.
EPA Office of Mobile Sources, Ann Arbor, MI, September 1991. Available in Docket
EPA-HQ-OAR-2005-0036.
19 Hayes, R. B., S. N. Yin, M. S. Dosemici, et al. (1997) Benzene and the dose-related
incidence of hematological neoplasms in China. J. Nat. Cancer Inst. 89:1065-1071.
20 International Agency for Research on Cancer (IARC) (1982) IARC monographs on the
evaluation of carcinogenic risk of chemicals to humans, Volume 29, Some industrial
chemicals and dyestuffs, International Agency for Research on Cancer, World Health
Organization, Lyon, France, p. 345-389.
21 U.S. EPA (1998) Carcinogenic Effects of Benzene: An Update, National Center for
Environmental Assessment, Washington, DC. EPA600-P-97-001F. Enter report number
at the following search page,
http://yosemite.epa.gOv/ncepihom/nsCatalog.nsf//SearchPubs7Openform.
22 Irons, R.D., W.S. Stillman, D.B. Colagiovanni, and V.A. Henry (1992) Synergistic
action of the benzene metabolite hydroquinone on myelopoietic stimulating activity of
granulocyte/macrophage colony-stimulating factor in vitro, Proc. Natl. Acad. Sci.
89:3691-3695.
23 U.S. EPA (1998) Carcinogenic Effects of Benzene: An Update, National Center for
Environmental Assessment, Washington, DC. EPA600-P-97-001F. Enter report number
at the following search page,
http://vosemite.epa.gOv/ncepihom/nsCatalog.nsf//SearchPubs7Openform.
1-31
-------
Final Regulatory Impact Analysis
24 U.S. EPA (1998) Carcinogenic Effects of Benzene: An Update, National Center for
Environmental Assessment, Washington, DC. EPA600-P-97-001F. Enter report number
at the following search page,
http://yosemite.epa.gOv/ncepihom/nsCatalog.nsf//SearchPubs7Openform.
25 U. S. EPA (2005) Guidelines for Carcinogen Risk Assessment. Report No.
EPA/630/P-03/001F. http://cfpub.epa.gov/ncea/raf/recordisplav.cfm?deid=l 16283.
26 U.S. EPA (1998) Carcinogenic Effects of Benzene: An Update. EPA/600/P-97/001F. .
Enter report number at the following search page,
http://vosemite.epa.gOv/ncepihom/nsCatalog.nsf//SearchPubs7Openform.
27 Rothman, N; Li, GL; Dosemeci, M; et al. (1996) Hematotoxicity among Chinese
workers heavily exposed to benzene. Am J Ind Med 29:236-246.
28 Rappaport, S.M.; Waidyanatha, S.; Qu, Q.; Shore, R.; Jin, X.; Cohen, B.; Chen, L.;
Melikian, A.; Li, G; Yin, S.; Yan, H.; Xu, B.; Mu, R.; Li, Y.; Zhang, X.; and Li, K.
(2002) Albumin adducts of benzene oxide and 1,4-benzoquinone as measures of human
benzene metabolism. Cancer Research. 62:1330-1337.
29 Rappaport, S.M.; Waidyanatha, S.; Qu, Q.; Yeowell-O'Connell, K.; Rothman, N.;
Smith M.T.; Zhang, L.; Qu, Q.; Shore, R.; Li, G.; Yin, S. (2005) Protein adducts as
biomarkers of human enzene metabolism. Chem Biol Interact. 153-154:103-109.
30 Lin, Y-S., Vermeulen, R., Tsai, C.H., Suramya, W., Lan, Q., Rothman, N., Smith,
M.T., Zhang, L., Shen, M., Songnian, Y., Kim, S., Rappaport, S.M. (2006) Albumin
adducts of electrophilic benzene metabolitesin benzene-exposed and control workers.
Environ Health Perspec.
31 Hayes, R.B.; Yin, S.; Dosemeci, M.; Li, G.; Wacholder, S.; Travis, L.B.; Li, C.;
Rothman, N.; Hoover, R.N.; and Linet, M.S. (1997) Benzene and the dose-related
incidence of hematologic neoplasms in China. J. Nat. Cancer Inst. 89:1065-1071.
32 Hayes, R.B.; Songnian, Y.; Dosemeci, M.; and Linet, M. (2001) Benzene and
lymphohematopoietic malignancies in humans. Am J Indust Med, 40:117-126.
33 Lan, Q.;, Zhang, L., Li, G., Vermeulen, R., et al. (2004). Hematotoxicicity in Workers
Exposed to Low Levels of Benzene. Science 306: 1774-1776.
34 Shu, X.O,; Gao, Y.T.; Brinton, L.A.; et al. (1988) A population-based case-control
study of childhood leukemia in Shanghai. Cancer 62:635-644.
35 McKinney P.A.; Alexander, F.E.; Cartwright, R.A.; et al. (1991) Parental occupations
of children with leukemia in west Cumbria, north Humberside, and Gateshead, Br Med J
302:681-686.
1-32
-------
Final Regulatory Impact Analysis
36 Keller, KA; Snyder, CA. (1986) Mice exposed in utero to low concentrations of
benzene exhibit enduring changes in their colony forming hematopoietic cells.
Toxicology 42:171-181.
37 Keller, KA; Snyder, CA. (1988) Mice exposed in utero to 20 ppm benzene exhibit
altered numbers of recognizable hematopoietic cells up to seven weeks after exposure.
Fundam Appl Toxicol 10:224-232.
38 Corti, M; Snyder, CA. (1996) Influences of gender, development, pregnancy and
ethanol consumption on the hematotoxicity of inhaled 10 ppm benzene. Arch Toxicol
70:209-217.
39 U. S. EPA. (2002). Toxicological Review of Benzene (Noncancer Effects). National
Center for Environmental Assessment, Washington, DC. Report No. EPA/635/R-
02/00IF. http://www.epa.gov/iris/toxreviews/0276-tr.pdf.
40 Ford, AM; Pombo-de-Oliveira, MS; McCarthy, KP; MacLean, JM; Carrico, KC;
Vincent, RF; Greaves, M. (1997) Monoclonal origin of concordant T-cell malignancy in
identical twins. Blood 89:281-285.
41 Aksoy, M. (1989) Hematotoxicity and carcinogenicity of benzene. Environ. Health
Perspect. 82: 193-197.
42 Goldstein, B.D. (1988) Benzene toxicity. Occupational medicine. State of the Art
Reviews. 3: 541-554.
43 Aksoy, M. (1991) Hematotoxicity, leukemogenicity and carcinogenicity of chronic
exposure to benzene. In: Arinc, E.; Schenkman, J.B.; Hodgson, E., Eds. Molecular
Aspects of Monooxygenases and Bioactivation of Toxic Compounds. New York:
Plenum Press, pp. 415-434.
44 Goldstein, B.D. (1988) Benzene toxicity. Occupational medicine. State of the Art
Reviews. 3: 541-554.
45 Aksoy, M., S. Erdem, and G. Dincol. (1974) Leukemia in shoe-workers exposed
chronically to benzene. Blood 44:837.
46 Aksoy, M. and K. Erdem. (1978) A follow-up study on the mortality and the
development of leukemia in 44 pancytopenic patients associated with long-term exposure
to benzene. Blood 52: 285-292.
47 Rothman, N., G.L. Li, M. Dosemeci, W.E. Bechtold, G.E. Marti, Y.Z. Wang, M. Linet,
L.Q. Xi, W. Lu, M.T. Smith, N. Titenko-Holland, L.P. Zhang, W. Blot, S.N. Yin, and
R.B. Hayes (1996) Hematotoxicity among Chinese workers heavily exposed to benzene.
Am. J. Ind. Med. 29: 236-246.
1-33
-------
Final Regulatory Impact Analysis
48 EPA 2005 "Full IRIS Summary for Benzene (CASRN 71-43-2)" Environmental
Protection Agency, Integrated Risk Information System (IRIS), Office of Health and
Environmental Assessment, Environmental Criteria and Assessment Office, Cincinnati,
OH http://www.epa.gov/iris/subst/0276.htm.
49 Qu, Q., R. Shore, G. Li, X. Jin, L.C. Chen, B. Cohen, et al. (2002). Hematological
changes among Chinese workers with a broad range of benzene exposures. Am. J.
Industr. Med. 42: 275-285.
50 Lan, Q.;, Zhang, L., Li, G., Vermeulen, R., et al. (2004). Hematotoxicity in Workers
Exposed to Low Levels of Benzene. Science 306: 1774-1776.
51
Turtletaub, K.W. and Mani, C. (2003). Benzene metabolism in rodents at doses
relevant to human exposure from urban air. Health Effects Inst. Research Report No. 113.
52 U.S. EPA. (2002). Health Assessment of 1,3-Butadiene. Office of Research and
Development, National Center for Environmental Assessment, Washington Office,
Washington, DC. Report No. EPA600-P-98-001F at
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=54499.
53 EPA 2005 "Full IRIS Summary for 1,3-butadiene (CASRN 106-99-0)" Environmental
Protection Agency, Integrated Risk Information System (IRIS), Office of Health and
Environmental Assessment, Environmental Criteria and Assessment Office, Cincinnati,
OH http://www.epa.gov/iris/subst/0139.htm.
54 Delzell, E., Sathiakumar, N., Hovinga, M., Macaluso, M., Julian, J., Larson, R., Cole,
P. and Muir, D.C.F., 1996. A Follow-up Study of Synthetic Rubber Workers. Toxicology,
113, 182-189.
55 U.S. EPA. (2002). Health Assessment of 1,3-Butadiene. Office of Research and
Development, National Center for Environmental Assessment, Washington Office,
Washington, DC. Report No. EPA600-P-98-001F at
http://cfpub.epa.gov/ncea/cfm/recordisplav.cfm?deid=54499.
56 EPA 2005 "Full IRIS Summary for 1,3-butadiene (CASRN 106-99-0)" Environmental
Protection Agency, Integrated Risk Information System (IRIS), Office of Health and
Environmental Assessment, Environmental Criteria and Assessment Office, Cincinnati,
OH http://www.epa.gov/iris/subst/0139.htm.
57 EPA 2005 "Full IRIS Summary for 1,3-butadiene (CASRN 106-99-0)" Environmental
Protection Agency, Integrated Risk Information System (IRIS), Office of Health and
Environmental Assessment, Environmental Criteria and Assessment Office, Cincinnati,
OH http://www.epa.gov/iris/subst/0139.htm.
1-34
-------
Final Regulatory Impact Analysis
58 Delzell, E., Sathiakumar, N., Graff, I, Macaluso, M., Maldonado, G., Matthews, R.
(2006) An updated study of mortality among North American synthetic rubber industry
workers. Health Effects Institute Report Number 132.
59 Bevan, C.; Stadler, J.C.; Elliot, G.S.; et al. (1996) Subchronic toxicity of 4-
vinylcyclohexene in rats and mice by inhalation. Fundam. Appl. Toxicol. 32:1-10.
60 U.S. EPA (1987) Assessment of Health Risks to Garment Workers and Certain Home
Residents from Exposure to Formaldehyde, Office of Pesticides and Toxic Substances,
April 1987. Found in various EPA library collections through
http://www.epa.gov/natlibra/ols.htm by its OCLC catalog no. 16989049.
61 Hauptmann, M..; Lubin, J. H.; Stewart, P. A.; Hayes, R. B.; Blair, A. 2003. Mortality
from lymphohematopoetic malignancies among workers in formaldehyde industries.
Journal of the National Cancer Institute 95: 1615-1623.
62 Hauptmann, M..; Lubin, J. H.; Stewart, P. A.; Hayes, R. B.; Blair, A. 2004. Mortality
from solid cancers among workers in formaldehyde industries. American Journal of
Epidemiology 159: 1117-1130.
63 Pinkerton, L. E. 2004. Mortality among a cohort of garment workers exposed to
formaldehyde: an update. Occup. Environ. Med. 61: 193-200.
64 Coggon, D, EC Harris, J Poole, KT Palmer. 2003. Extended follow-up of a cohort of
British chemical workers exposed to formaldehyde. J National Cancer Inst. 95:1608-
1615.
65 Bundesinstitut fur Risikobewertung (BfR) Toxicological Assessment of Formaldehyde.
Opinion of BfR No. 023/2006 of 30 March 2006.
www.bfr.bund. de/cm/290/toxicological_assessment_of_formal dehyde.pdf
66 Conolly, RB, JS Kimbell, D Janszen, PM Schlosser, D Kalisak, J Preston, and FJ
Miller. 2003. Biologically motivated computational modeling of formaldehyde
carcinogenicity in the F344 rat. Tox Sci 75: 432-447.
67 Conolly, RB, JS Kimbell, D Janszen, PM Schlosser, D Kalisak, J Preston, and FJ
Miller. 2004. Human respiratory tract cancer risks of inhaled formaldehyde: Dose-
response predictions derived from biologically-motivated computational modeling of a
combined rodent and human dataset. Tox Sci 82: 279-296.
68 Chemical Industry Institute of Toxicology (CUT). 1999. Formaldehyde: Hazard
characterization and dose-response assessment for carcinogenicity by the route of
inhalation. CUT, September 28, 1999. Research Triangle Park, NC.
69 Health Canada (2001) Priority Substances List Assessment Report. Formaldehyde.
Environment Canada, Health Canada, February 2001. The document may be accessed at
1-35
-------
Final Regulatory Impact Analysis
http://www.hc-sc.gc.ca/ewh-setnt/pubs/contatninants/psl2-
Isp2/formaldehyde/index_e.html .
70 U.S. EPA (2004) National Emission Standards for Hazardous Air Pollutants for
Plywood and Composite Wood Products Manufacture: Final Rule. (69 FR 45943,
7/30/04)
71 Agency for Toxic Substances and Disease Registry (ATSDR). Toxicological Profile
for Formaldehyde. Public Health Service, U.S. Department of Health and Human
Services, Atlanta, GA. Available at http://www.atsdr.cdc.gov/toxprofiles/tpl 11 .html.
72 EPA 2005 "Full IRIS Summary for Acetaldehyde (CASRN 75-07-0)" Environmental
Protection Agency, Integrated Risk Information System (IRIS), Office of Health and
Environmental Assessment, Environmental Criteria and Assessment Office, Cincinnati,
OH http://www.epa.gov/iris/subst/0290.htm.
73 Appleman, L. M., R. A. Woutersen, V. J. Feron, R. N. Hooftman, and W. R. F. Notten.
(1986). Effects of the variable versus fixed exposure levels on the toxicity of
acetaldehyde in rats. J. Appl. Toxicol. 6: 331-336.
74 Appleman, L.M., R.A. Woutersen, and VJ. Feron. (1982). Inhalation toxicity of
acetaldehyde in rats. I. Acute and subacute studies. Toxicology. 23: 293-297.
75 EPA 2005 "Full IRIS Summary for Acetaldehyde (CASRN 75-07-0)" Environmental
Protection Agency, Integrated Risk Information System (IRIS), Office of Health and
Environmental Assessment, Environmental Criteria and Assessment Office, Cincinnati,
OH http://www.epa.gov/iris/subst/0290.htm.
76 EPA 2005 "Full IRIS Summary for Acetaldehyde (CASRN 75-07-0)" Environmental
Protection Agency, Integrated Risk Information System (IRIS), Office of Health and
Environmental Assessment, Environmental Criteria and Assessment Office, Cincinnati,
OH http://www.epa.gov/iris/subst/0290.htm.
77 Myou, S.; Fujimura, M.; Nishi K.; Ohka, T.; and Matsuda, T. (1993) Aerosolized
acetaldehyde induces histamine-mediated bronchoconstriction in asthmatics. Am Rev
RespirDis 148(4 Pt 1): 940-3.
78 Agency for Toxic Substances and Disease Registry (ATSDR). Toxicological Profile
for Acrolein. Public Health Service, U.S. Department of Health and Human Services,
Atlanta, GA. 2005. Publication #PB/91/180307/AS at
http://www.atsdr.cdc. gov/toxprofiles/tp 124. html
79 EPA 2005 "Full IRIS Summary for Acrolein (CASRN 107-02-8)" Environmental
Protection Agency, Integrated Risk Information System (IRIS), Office of Health and
Environmental Assessment, Environmental Criteria and Assessment Office, Cincinnati,
OH http://www.epa.gov/iris/subst/0364.htm.
1-36
-------
Final Regulatory Impact Analysis
80 EPA 2005 "Full IRIS Summary for Acrolein (CASRN 107-02-8)" Environmental
Protection Agency, Integrated Risk Information System (IRIS), Office of Health and
Environmental Assessment, Environmental Criteria and Assessment Office, Cincinnati,
OH http://www.epa.gov/iris/subst/0364.htm.
81 Agency for Toxic Substances and Disease Registry (ATSDR). Draft Toxicological
Profile for Acrolein. 2005 Public Health Service, U.S. Department of Health and Human
Services, Atlanta, GA. Available at http://www.atsdr.cdc.gov/toxprofiles/.
82US EPA .2004. Toxicological Review of Naphthalene. (Reassessment of inhalation
cancer risk), Environmental Protection Agency, Integrated Risk Information System
(IRIS), Research and Development, National Center for Environmental Assessment,
Washington, DC http://www.epa.gov/iris/subst/0436.htm
83 Oak Ridge Institute for Science and Education. (2004) External Peer Review for the
IRIS Reassessment of the Inhalation Carcinogenicity of Naphthalene. August 2004.
http://cfpub2.epa.gov/ncea/cfm/recordisplay.cfm?deid=86019
84 California EPA (2004) Long Term Health Effects of Exposure to Naphthalene. Office
of Environmental Health Hazard Assessment at
http://www. oehha. ca. gov/air/toxi c contaminants/draftnaphth.html
85 International Agency for Research on Cancer (IARC) (2002) Monographs on the
Evaluation of the Carcinogenic Risk of Chemicals for Humans. Vol. 82. Lyon, France.
86 USEPA. 1998. Toxicological Review of Naphthalene. Environmental Protection
Agency, Integrated Risk Information System (IRIS), Research and Development,
National Center for Environmental Assessment, Washington, DC
http://www.epa.gov/iris/subst/0436.htm.
87 EPA 2005 "Full IRIS Summary for 2,2,4-Trimethylpentane (CASRN 540-84-1)"
Environmental Protection Agency, Integrated Risk Information System (IRIS), Office of
Health and Environmental Assessment, Environmental Criteria and Assessment Office,
Cincinnati, OH http://www.epa.gov/iris/subst/0614.htm.
88 Hazardous Substances Data Bank (HSDB) on Isooctane 2005. National Library of
Medicine Bethesda, MD found at http://toxnet.nlm.nih.gov/index.html. Enter "isooctane"
at search screen.
89 ATSDR (1999) Toxicological Profile for Ethylbenzene (update). USDHHS, PHS,
ATSDR. Publication PB/99/166647 at http://www.atsdr.cdc.gov/toxprofiles/tp 110.html.
90 EPA 2005 "Full IRIS Summary for Ethylbenzene (CASRN 100-41-4)" Environmental
Protection Agency, Integrated Risk Information System (IRIS), Office of Health and
Environmental Assessment, Environmental Criteria and Assessment Office, Cincinnati,
1-37
-------
Final Regulatory Impact Analysis
OH http://www.epa.gov/iris/subst/0051 .htm.
91 Agency for Toxic Substances and Disease Registry (ATSDR). Toxicological Profile for
Ethylbenzene. 1999 Public Health Service, U.S. Department of Health and Human
Services, Atlanta, GA. Available at http://www.atsdr.cdc.gov/toxprofiles/.
92 ATSDR. 1999. Toxicological Profile for n-Hexane. USDHHS, PHS, ATSDR.
Publication#PB/99/166688. http://www.atsdr.cdc.gov/toxprofiles/tpll3.html.
93 EPA 2005 "Full IRIS Summary for n-Hexane (CASRN 110-54-3)" Environmental
Protection Agency, Integrated Risk Information System (IRIS), Office of Health and
Environmental Assessment, Environmental Criteria and Assessment Office, Cincinnati,
OH http://www.epa.gov/iris/subst/0486.htm.
94 EPA 2005 "Full IRIS Summary for n-Hexane (CASRN 110-54-3)" Environmental
Protection Agency, Integrated Risk Information System (IRIS), Office of Health and
Environmental Assessment, Environmental Criteria and Assessment Office, Cincinnati,
OH http://www.epa.gov/iris/subst/0486.htm.
95 Agency for Toxic Substances and Disease Registry (ATSDR). Toxicological Profile for
n-Hexane. 1999 Public Health Service, U.S. Department of Health and Human Services,
Atlanta, GA. Available at http://www.atsdr.cdc.gov/toxprofiles/.
96 EPA. 1994. Health risk perspectives on fuel oxygenates. Washington, DC: Office of
Research and Development; Report No. EPA 600/R-94/217.
97 Interagency Oxygenated Fuels Assessment Steering Committee. 1997. Interagency
assessment of oxygenated fuels. Washington, DC: National Science and Technology
Council, Committee on Environment and Natural Resources and Office of Science and
Technology Policy, http://www.epa.gov/otaq/regs/fuels/ostpfm.pdf.
98 EPA 2005 "Full IRIS Summary for Methyl-tertiary-butyl Ether (MTBE, CASRN 1634-
04-4)" Environmental Protection Agency, Integrated Risk Information System (IRIS),
Office of Health and Environmental Assessment, Environmental Criteria and Assessment
Office, Cincinnati, OH http://www.epa.gov/iris/subst/0545.htm.
99 EPA 2005 "Full IRIS Summary for Styrene (CASRN 100-42-5)" Environmental
Protection Agency, Integrated Risk Information System (IRIS), Office of Health and
Environmental Assessment, Environmental Criteria and Assessment Office, Cincinnati,
OH http://www.epa.gov/iris/subst/0104.htm.
100 Agency for Toxic Substances Disease Registry (1992) Toxicological profile for
styrene. Atlanta: ATSDR. http://www.atsdr.cdc.gov/toxprofiles/tp53.html.
101 EPA 2005 "Full IRIS Summary for Toluene (CASRN 108-88-3)" Environmental
Protection Agency, Integrated Risk Information System (IRIS), Office of Health and
1-38
-------
Final Regulatory Impact Analysis
Environmental Assessment, Environmental Criteria and Assessment Office, Cincinnati,
OH http://www.epa.gov/iris/subst/0118.htm.
102 EPA 2005 "Full IRIS Summary for Toluene (CASRN 108-88-3)" Environmental
Protection Agency, Integrated Risk Information System (IRIS), Office of Health and
Environmental Assessment, Environmental Criteria and Assessment Office, Cincinnati,
OH http://www.epa.gov/iris/subst/0118.htm.
103 EPA 2005 "Full IRIS Summary for Toluene (CASRN 108-88-3)" Environmental
Protection Agency, Integrated Risk Information System (IRIS), Office of Health and
Environmental Assessment, Environmental Criteria and Assessment Office, Cincinnati,
OH http://www.epa.gov/iris/subst/0118.htm.
104 EPA 2005 "Full IRIS Summary for Xylenes (CASRN 1330-20-7)" Environmental
Protection Agency, Integrated Risk Information System (IRIS), Office of Health and
Environmental Assessment, Environmental Criteria and Assessment Office, Cincinnati,
OH http://www.epa.gov/iris/subst/0270.htm.
105 EPA Toxicological Review of Xylenes. January 2003. EPA 635/R-03/001.
106 EPA 2005 "Full IRIS Summary for Xylenes (CASRN 1330-20-7)" Environmental
Protection Agency, Integrated Risk Information System (IRIS), Office of Health and
Environmental Assessment, Environmental Criteria and Assessment Office, Cincinnati,
OH http://www.epa.gov/iris/subst/0270.htm.
107 Agency for Toxic Substances and Disease Registry (ATSDR). Toxicological Profile
for Xylene. 2005 Public Health Service, U.S. Department of Health and Human
Services, Atlanta, GA. Available at http://www.atsdr.cdc.gov/toxprofiles/.
108 Perera, F.P.; Rauh, V.; Tsai, W-Y.; et al. (2002) Effect of transplacental exposure to
environmental pollutants on birth outcomes in a multiethnic population. Environ Health
Perspect 111: 201-205.
109 U.S. EPA (2002) Health Assessment Document for Diesel Engine Exhaust.
EPA/600/8-90/057F Office of Research and Development, Washington DC. This
document is available electronically at
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=29060 .
110 Ishinishi, N; Kuwabara, N; Takaki, Y; et al. (1988) Long-term inhalation experiments
on diesel exhaust. In: Diesel exhaust and health risks. Results of the HERP studies.
Ibaraki, Japan: Research Committee for HERP Studies; pp. 11-84.
111 Heinrich, U; Fuhst, R; Rittinghausen, S; et al. (1995) Chronic inhalation exposure of
Wistar rats and two different strains of mice to diesel engine exhaust, carbon black, and
titanium dioxide. Inhal Toxicol 7:553-556.
112 Mauderly, JL; Jones, RK; Griffith, WC; et al. (1987) Diesel exhaust is a pulmonary
carcinogen in rats exposed chronically by inhalation. Fundam Appl Toxicol 9:208-221.
113
Nikula, KJ; Snipes, MB; Barr, EB; et al. (1995) Comparative pulmonary toxicities and
1-39
-------
Final Regulatory Impact Analysis
carcinogenicities of chronically inhaled diesel exhaust and carbon black in F344 rats.
Fundam Appl Toxicol 25:80-94.
114 U.S. Environmental Protection Agency (2004) Air Quality Criteria for Particulate
Matter. Research Triangle Park, NC: National Center for Environmental Assessment -
RTF Office; Report No. EPA/600/P-99/002aF.
115 U.S. Environmental Protection Agency (2004) Air Quality Criteria for Particulate
Matter. Research Triangle Park, NC: National Center for Environmental Assessment -
RTF Office; Report No. EPA/600/P-99/002aF, p. 8-318.
116 Seagrave, J.; McDonald, J.D.; Gigliotti, A.P.; Nikula, K.J.; Seilkop, S.K.; Gurevich,
M. and Mauderly, J.L. (2002) Mutagenicity and in Vivo Toxicity of Combined
Particulate and Semivolatile Organic Fractions of Gasoline and Diesel Engine Emissions.
Toxicological Sciences 70:212-226.
117 Watson,!., Fujita,E., Chow,J., Zielinska,B., Richards,L., Neff,W., Dietrich,D.
Northern Front Range Air Quality Study Final Report: Volume 1. June 30, 1998. For
Colorado State University, Cooperative Institute for Research in the Atmosphere, by
Desert Research Institute, Reno, NV. This document is available in the EPA Docket as
EPA-HQ-OAR-2005-0036-1179.
118 Schauer, J.J.; Rogge, W.F.; Hildemann, L.M.; et al. (1996) Source apportionment of
airborne particulate matter using organic compounds as tracers. Atmos Environ
30(22):3837-3855.
119 EPA, 2001, "National Air Quality and Emission Trends Report, 1999," EPA 454/R-
01-004.
120 Watson,!., Fujita,E., Chow,J., Zielinska,B., Richards,L., Neff,W., Dietrich,D.
Northern Front Range Air Quality Study Final Report: Volume 1. June 30, 1998. For
Colorado State University, Cooperative Institute for Research in the Atmosphere, by
Desert Research Institute, Reno, NV. This document is available in the EPA Docket as
EPA-HQ-OAR-2005-0036-1179.
121 Schauer, James J., Wolfgang F. Rogge, Lynn M. Hildemann, Monica A. Mazurek,
Glen R. Cass, and Bernd Simoneit, 1996, "Source Apportionment of Airborne Particulate
Matter Using Organic Compounds as Tracers," Atmospheric Environment, 30, 3837-
3855.
122 Zheng, Mei, Glen R. Cass, James J. Schauer, and Eric S. Edgerton, 2002, "Source
Apportionment of PM2.5 in the Southeastern United States Using Solvent-Extractable
Organic Compounds as Tracers," Environmental Science and Technology, 36, 2631-2371.
1-40
-------
Final Regulatory Impact Analysis
123 Hannigan, Michael P., William F. Busby, Jr., Glen R. Cass, 2005, "Source
Contributions to the Mutagenicity of Urban Particulate Air Pollution," Journal of the Air
and Waste Management Association, 55, 399-410.
124 Kleeman, Michael J.., Glen R. Cass, 1999, "Identifying the Effect of Individual
Emission Sources on Particulate Air Quality Within a Photochemical Aerosol Processes
Trajectory Model," Atmospheric Environment, 33, 4597-4613.
125 Schauer, James J., Glen R. Cass, 2000, "Source Apportionment of Wintertime Gas-
Phase and Particle-Phase Air Pollutants Using Organic Compounds as Tracers,
Environmental Science and Technology, 1821-1832.
126 Izumi, L. and T. Fukuyama. 1990, Photochemical aerosol formation from aromatic
hydrocarbons in the presence of NOx., Atmospheric Environment, 24A, 1433.
127 Martin-Reviego, M. and K. Wirtz, 2005. "Is benzene a precursor for secondary
organic aerosol?" Environmental Science and Technology, 39, 1045-1054.
1 98
Kleindienst, T.E., T.S. Conver, C.D. Mclver, and E.G. Edney. 2004. Determination of
secondary organic aerosol products from the photooxidation of toluene and their implications in
ambient PM2 5. J. Atmos. Chem. 47, 79-100.
Edney. E.O., T.E. Kleindienst, M. Jaoui, M. Lewandowski, J.H. Offenberg, W. Wang, M.
Claeys. 2005. Formation of 2-methyl tetrols and 2-methylglyceric acid in secondary organic
J J J O J JO
aerosol from laboratory irradiated isoprene/NOx/SO2/air mixtures and their detection in ambient
PM25 samples collected in the Eastern United States. Atmospheric Environment 39: 5281-5289.
Claeys, M., R. Szmigielski, I. Kourtchev, P. Van der Veken, R. Vermeylen, W. Maenhaut, M.
Jaoui, T.E. Kleindienst, M. Lewandowski, J.H. Offenberg, E.O. Edney. 2007.
Hydroxydicarboxylic acids: Markers for secondary organic aerosol from the photooxidation of ot-
pinene. Environ. Sci. Technol. (Web Edition, 01/23/2007).
Jaoui, M., T.E. Kleindienst, M. Lewandowski, J.H. Offenberg, E.O. Edney. 2005.
Identification and quantification of aerosol polar oxygenated compounds bearing carboxylic or
hydroxyl groups. 2. Organic tracer compounds from monoterpenes. Environ. Sci.
Technol. 39:5661-5673.
132 Edney. E.G., T.E. Kleindienst, T.S. Conver, C.D. Mclver, and W.S. Weathers. 2003. Polar
organic oxygenates in PM2 5 at a southeastern site in the United States. Atmospheric Environment
37, 3947-3965.
Lewandowski, M., M. Jaoui, T.E. Kleindienst, J.H. Offenberg, E.O. Edney. 2007.
Composition of PM25 during the summer of 2003 in Research Triangle Park, North Carolina.
Atmos. Environ, (in press; available on line 01/30/2007).
1-41
-------
Final Regulatory Impact Analysis
134 Schauer, James J., Glynis C Lough, Martin M Shafer, William F Christensen, Michael
F Arndt, Jeffrey T DeMinter and June-Soo Park. 2006. Characterization of metals
emitted from motor vehicles. Health Effects Institute Research Report Number 133.
1-42
-------
Final Regulatory Impact Analysis
Chapter 2: Table of Contents
Chapter 2: Emission Inventories 3
2.1 Criteria Pollutants 3
2.1.1 Methods 3
2.1.1.1 Highway Vehicles 4
2.1.1.2 Portable Fuel Containers 9
2.1.2 Emission Reductions of Criteria Pollutants Resulting From Controls 10
2.1.3 Strengths and Limitations of Criteria Pollutant Inventories 20
2.2 Air Toxics 22
2.2.1 Emission Inventories Used in Air Quality Modeling 22
2.2.1.1.1 Highway Vehicles 23
2.2.1.1.2 Nonroad Equipment in theNonroad Model 31
2.2.1.1.3 Commercial Marine Vessels, Locomotives and Aircraft 32
2.2.1.1.4. Portable Fuel Containers 33
2.2.1.1.5. Gasoline Distribution 35
2.2.1.1.6. Other Stationary Sources 36
2.2.1.1.7 Precursor Emissions 38
2.2.1.2 Trends in Air Toxic Emissions 42
2.2.1.2.1 Emission Trends Without Controls 42
2.2.1.2.2 Impact on Inventory of Controls 58
2.2.2 Emission Reductions from Controls 64
2.2.2.1 Methodology Changes from Air Quality Inventories 64
2.2.2.1.1 Highway Vehicles 64
2.2.2.1.2 Nonroad Equipment 64
2.2.2.1.3 Portable Fuel Containers 65
2.2.2.1.4 Gasoline Distribution 65
2.2.2.2 Estimated Reductions for Air Toxic Pollutants of Greatest Concern.... 65
2.2.2.2.1 Fuel Benzene Standard 66
2.2.2.2.2 Cold Temperature VOC Emission Control 75
2.2.2.2.3 Portable Fuel Container Control 79
2.2.2.2.4 Cumulative Reductions of Controls 82
2.3 Potential Implications of New Emissions Data for Inventories 89
2.3.1 Newer Technology Light Duty Vehicles 89
2.3.2 Heavy-Duty Vehicles (CRC E-55/E-59) 90
2.3.3 Small Spark Ignition Engines 91
2.3.4 Nonroad CI engines 93
2.4 Description of Current Mobile Source Emissions Control Programs that Reduce
MSATs 94
2.4.1 Fuels Programs 94
2.4.1.1 RFG 94
2.4.1.2 Anti-dumping 95
2.4.1.3 2001 Mobile Source Air Toxics Rule (MSATI) 95
2.4.1.4 Gasoline Sulfur 96
2.4.1.5 Gasoline Volatility 96
2.4.1.6 Diesel Fuel 96
2-1
-------
Final Regulatory Impact Analysis
2.4.1.7 Phase-Out of Lead in Gasoline 97
2.4.2 Highway Vehicle and Engine Programs 97
2.4.3 Nonroad Engine Programs 99
2.4.3.1 Land-based Diesel Engines 100
2.4.3.2 Land-Based SI Engines 100
2.4.3.2.1 Large Land-Based SI Engines 100
2.4.3.2.2 Recreational Vehicles 101
2.4.3.2.3 Small Land-Based SI Engines 101
2.4.3.3 Marine Engines 101
2.4.3.3.1 Marine SI Engines 102
2.4.3.3.2 Marine Diesel Engines 102
2.43 A Locomotives 103
2.4.3.5 Aircraft 103
2.4.4 Voluntary Programs 103
2-2
-------
Final Regulatory Impact Analysis
Chapter 2: Emission Inventories
This chapter describes the methods used to develop inventories for air quality
modeling, estimation of emission benefits, and calculation of cost-effectiveness for this
rule. The chapter also presents and discusses these inventories. MSAT inventories for
air quality modeling were developed well in advance of final rule promulgation, because
of the lead time required to conduct air quality, exposure, and risk analyses. Thus, these
inventories do not include revised estimates of emissions using new fuel quality estimates
developed for the Renewable Fuel Standard Program, as discussed below. Therefore, the
chapter has separate sections discussing MSAT inventories used for modeling, and
revised inventories used to estimate emission benefits of the rule and cost-effectiveness.
2.1 Criteria Pollutants
2.1.1 Methods
For the final rule, we have revised the emission inventories to reflect conditions
anticipated under the Renewable Fuel Standard (RFS) program. The RFS program was
mandated by the Energy Policy Act of 2005 in order to increase national consumption of
renewable fuels. In September 2006, EPA issued a proposed rule to implement the RFS
program for 2007 and beyond.1 The RFS proposal analyzed several different scenarios of
increased ethanol use and developed county-level fuel properties specific to each
scenario.
In one particular RFS scenario, we estimated county-level fuel properties by
allocating the Energy Information Agency's forecast of 9.6 billion gallons of national
ethanol consumption in 2012, attributing as much as possible for use as an oxygenate in
reformulated gasoline. For purposes of this rule, we have selected this scenario as the
most likely ethanol volume and distribution in 2012, and have therefore adopted those
fuel properties as the new baseline fuel for MSAT inventories used to evaluate the cost-
effectiveness of the standards being finalized in this rule. In the discussion that follows,
the new MSAT baseline fuel is referred to as the "RFS fuel". The RFS Draft Regulatory
Impact Assessment (DRIA) contains a detailed discussion of the effects of ethanol fuel
on gasoline properties and the methods by which we derived RFS county-level fuel
properties.2
Though cost-effectiveness inventories in both the RFS proposal and the MSAT
final rule reflect RFS fuels, there are slight differences in other methodologies used to
estimate the emissions inventories. However, the differences are minor and have little
impact on emission reductions used to evaluate cost-effectiveness.
2-3
-------
Final Regulatory Impact Analysis
2.1.1.1 Highway Vehicles
Highway vehicle hydrocarbon (HC) emission inventories were calculated by
using vehicle emission rates produced from the emission model MOBILE6.2 multiplied
by vehicle miles traveled (VMT) using the National Mobile Inventory Model (NMIM).3
MOBILE6.2 uses emission factors obtained through the analysis of emissions data
collected from vehicle emission research.4 The emission factors reflect impacts of
vehicle standards as well as current and planned inspection and maintenance programs.
They also reflect impacts of changes in properties of gasoline and diesel fuels. Impacts
of alternative fueled vehicles and engines (e.g. liquid propane, compressed natural gas,
methanol) are negligible in NMIM. The VMT used by NMIM was estimated for base
years using historical data from the Federal Highway Administration, allocated to
counties using the methodology documented for the National Emissions Inventory, and
projected to future years using the Energy Information Administration's National Energy
Modeling System (NEMS) Transportation Model. NEMS projects VMT for personal
travel based on demographic effects and economic influences such as estimated fuel costs
and disposable income, and projects commercial truck travel based on economic factors
such as industrial output and demand. This is the same approach used in the Clean Air
Interstate Air Quality (CAIR) rule.5 As mentioned above, county-level fuel properties
contained in the public release version of NMIM were revised to RFS fuel.
Analysis of vehicle emission certification data submitted by vehicle
manufacturers to EPA as part of requirements to comply with requirements for cold
temperature carbon monoxide (CO) standards, as well as surveillance testing data from
the California Air Resources Board, indicated that MOBILE6.2 was substantially
underestimating start emissions at cold temperatures for Tier 1 and later vehicles. This
data was supplemented with test data collected by EPA at Southwest Research Institute
(SwRI)6 and was then used to adjust the temperature and engine start emission factors in
MOBILE6.2 to provide inputs to NMIM, which calculates county-level national
inventories.7
EPA cold CO certification data was paired as 20 °F versus 75 °F tests per engine
family to calculate the additional hydrocarbon (HC) emissions due to lower temperature.
Available bag emission data indicated that at 20 °F, as in the standard Federal Test
Procedure (FTP) at 75 °F, the majority of HC emissions occur during vehicle start and
that lower vehicle soak and start temperatures result in higher HC emissions. Table 2.1 .-
1 indicates the trends found in the EPA Cold CO program certification data.
The state of California has a 50 °F emission standard requirement and that data,
also supplied by manufacturers, reflects the same trend over the smaller temperature
difference (Table 2.1.-2).
The EPA testing at SwRI was performed on four Tier 2 vehicles to confirm the
effects seen in the certification data and to extend the range of soak temperature to 0 °F.
A summary of the hydrocarbon data is found in Table 2.1.-3.
2-4
-------
Final Regulatory Impact Analysis
Table 2.1.-1. FTP HC Data From Federal Certified Vehicles
(grams per mile)
Emission Standard
Tier 1
TLEV
LEV
ULEV
LEV2
2004 Tier 2
2005 Tier 2
2006 Tier 2
Sample Size
410
64
695
132
119
172
190
90
75°
Mean
0.1190
0.0804
0.0501
0.0335
0.0296
0.0406
0.0415
0.0408
Std. Dev.
0.0553
0.0286
0.0209
0.0214
0.0123
0.0169
0.0203
0.0239
20°
Mean
0.8630
0.6996
0.6402
0.4675
0.5035
0.5641
0.5651
0.5502
Std. Dev.
0.7269
0.2778
0.3723
0.2727
0.2549
0.3269
0.3247
0.3107
Table 2.1.-2. FTP HC Emissions Data from California Certified
Vehicles
(grams per mile)
Emission
Standard
LEV
ULEV
LEV2
Sample
Size
53
14
21
75°
Mean
0.0397
0.0162
0.0346
Std.
Dev.
0.0259
0.0043
0.0097
50°
Mean
0.0988
0.0403
0.0843
Std.
Dev.
0.0631
0.0176
0.0310
Ratio of
Averages
2.49
2.48
2.44
2-5
-------
Final Regulatory Impact Analysis
Table 2.1.-3. SwRI FTP (Bag 1 Only) Emissions from Four
Tier 2 Vehicles
Temperature in °F
Number of Observations
Average THC (gm/mile)
Standard deviation
Ratio to 75 °F
75
4
0.115
0.072
1
20
8
1.658
0.780
14.446
0
4
3.752
2.117
32.699
MOBILE6.2 currently has engine start emission factors based on 75° emission
test data on 1981 and newer vehicles. These engine start emissions are the difference, in
grams, between the emissions from phase 1 of the FTP after a 12-hour engine soak and
the emissions of the same driving fully warm and without the engine start. Temperature
effects on HC emissions are estimated using a multiplier that depends on ambient
temperature. This process is described in the MOBILE6.2 documentation.8 The current
engine start adjustments in MOBILE6.2 are not as large for Tier 1 and later vehicles as
what is indicated in the certification and SwRI test data. A method of correcting the
emission factors was developed using the test data. Those methods are covered in detail
in EPA technical report no. EPA420-D-06-001, "Cold Temperature Effects on Vehicle
HC Emissions."
Based on our analysis from Tier 1 and newer vehicles, it was decided that additive
values would be applied to 75 °F start emission factors based on temperature and vehicle
technology (i.e., Tier 1, NLEV, Tier 2, etc). Additive values can more closely
approximate the additional hydrocarbon emissions caused strictly by the start and warm-
up of the engine and/or the exhaust aftertreatment at the different temperatures than
multiplicative values. These values were obtained from subtracting the FTP emissions at
0, 20, and 50 °F from the FTP emissions at 75 °F using the certification and SwRI test
data. For emissions at temperature points where data was not available (i.e., 50 °F for
Tier 2 vehicles), linear interpolation between the 0°, 20° and 75 °F test data was used. All
of the difference in emissions is attributed to the increase in engine start emissions. The
values used for inputs for start adjustments are found in Table 2.1 .-4.
It is not clear what impact this phenomenon has on HC emissions in
malfunctioning or deteriorated vehicles. Emissions could go up proportionally to
properly operating vehicles or could go up at a lower rate. Properly operating vehicles
are very clean due to their emissions technology. Vehicle starts represent a period of
operation where the vehicle's emissions equipment is not fully operational and the
oxidation of fuel to carbon dioxide and water is not optimal. This situation is similar to
the conditions found in a deteriorated or improperly maintained vehicle except that the
condition is temporary in a normal vehicle. While MOBILE currently uses a multiplier
to account for temperature effects, doing so in the case of Tier 2 high emitting vehicles
results in extremely high and unrealistic emission rates. Therefore we have used the
MOBILE6.2 estimate of FTP emissions at 20 °F for model year 2005 high-emitting
vehicles in calendar year 2005 to develop the additive factor for all Tier 2 high-emitting
2-6
-------
Final Regulatory Impact Analysis
vehicles. Those values are found in Table 2.1 .-5. We are not changing high-emitting
vehicle emission factors for Tier 1 and older vehicles.
Table 2.1.-4. Increase in Engine Start Hydrocarbon Emissions
Over the 75 °F Baseline at Low Temperatures
(grams per engine start after a 12 hour soak)
Index
1
2
3
4
5
6
7
8
9
10
11
Index
1
2
3
4
5
6
7
8
9
10
11
12
Description
Tier 0 (not used)
Intermediate Tier 1
Tierl
Tier 2 (not used)
Intermediate Transitional Low Emission Vehicle
Transitional Low Emission Vehicle
Intermediate Low Emission Vehicle
Low Emission Vehicle (LEV)
Transitional Ultra Low Emission Vehicle
Ultra Low Emission Vehicle (ULEV)
Zero Emission Vehicle (ZEV) (not used)
Tier 2 (All Cars & Trucks) By Model Year
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
"F
0
25.96
25.96
25.96
18.26
21.60
21.60
20.59
20.59
15.14
15.14
0.00
0
18.26
18.27
17.77
17.77
17.77
17.77
17.77
17.77
17.77
17.77
17.77
17.77
20
12.98
12.98
12.98
9.13
10.80
10.80
10.29
10.29
7.57
7.57
0.00
20
9.13
9.13
8.88
8.88
8.88
8.88
8.88
8.88
8.88
8.88
8.88
8.88
50
3.09
3.09
3.09
3.27
2.09
2.09
1.30
1.30
0.87
0.87
0.00
50
3.27
3.27
3.27
3.27
3.27
3.27
3.27
3.27
3.27
3.27
3.27
3.27
Table 2.1.-5. Tier 2 High Emitter HC Adjustment
Based on 2005 Model Year MOBILE6.2 Results in Calendar Year 2005
Temperature °F
Engine start grams without
adjustment
Additional grams
0
63.335
50.522
20
41.360
28.547
50
21.821
9.008
75
12.813
N/A
2-7
-------
Final Regulatory Impact Analysis
The above tables and the new emission standard were used to determine the
effects of the cold temperature emission standard on start emission factors. The predicted
reductions were applied to Tier 2 vehicles over the phase-in period of the standards.
Those values are found in Table 2.1 .-6. No reductions beyond those found for normally-
emitting Tier 2 vehicles are applied for Tier 2 high-emitting vehicles.
With the appropriate HC start emission temperature adjustment factors, we can
provide the necessary emission factors required as inputs to NMIM to project pre-control
and control inventories for this rule. With the exception of using RFS fuel, no
modification to any other components of NMEVI is needed to calculate these inventories.
The inventories are presented in Chapter 2.1.2.
Table 2.1.-6. Adjustments to Engine Start Hydrocarbon Emissions
Over the 75 °F Baseline at Low Temperatures
For MSAT Rule
(grams per engine start after a 12 hour soak)
Index
1
2
o
J
4
5
6
7
8
9
10
11
12
Index
1
2
3
4
5
6
7
8
9
10
11
12
Tier 2 Cars & Light Trucks <6,000 Ibs GVWR
By Model Year
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
Tier 2 Light Trucks >6,000 Ibs GVWR By Model
Year
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
"F
0
18.26
18.27
17.77
17.77
17.77
17.77
6.66
6.66
6.66
6.66
6.66
6.66
20
9.13
9.13
8.88
8.88
8.88
8.88
3.3
3.3
3.3
3.3
3.3
3.3
50
3.27
3.27
3.27
3.27
3.27
3.27
1.215
1.215
1.215
1.215
1.215
1.215
"F
0
18.26
18.27
17.77
17.77
17.77
17.77
17.77
17.77
11.0
11.0
11.0
11.0
20
9.13
9.13
8.88
8.88
8.88
8.88
8.88
8.88
5.5
5.5
5.5
5.5
50
3.27
3.27
3.27
3.27
3.27
3.27
3.27
3.27
2.025
2.025
2.025
2.025
Phase In
Fraction
0
0
0
0
0
0
0.25
0.50
0.75
1.00
1.00
1.00
Phase In
Fraction
0
0
0
0
0
0
0
0
0.25
0.50
0.75
1.00
2-8
-------
Final Regulatory Impact Analysis
2.1.1.2 Portable Fuel Containers
In 1999, California's Air Resources Board (ARE) proposed a methodology to
estimate annual VOC emissions from portable fuel containers (PFCs) within California.
Their approach relied on survey data to first estimate the number of PFCs, and then to
combine those estimates with results from testing PFCs to develop a statewide annual
inventory.
EPA has modified California's approach. We first used our NONROAD2005
emissions model to estimate (for each month of the year and for each state) the quantity
of gasoline dispensed from PFCs that was used to fuel nonroad equipment. Then using
some of the California survey data on the amount of gasoline stored in each PFC, EPA
estimated the number of PFCs in use (each season) with gasoline in each state. These
estimated counts of PFCs were similar (but not identical) to the California estimates.
EPA also adjusted the California emission estimates to account for daily temperature
variations and seasonal RVP variations. (The estimated RVPs for future years include
the effects of the Renewable Fuels Standard.) EPA then combined its state-by-state
estimates of PFC usage with its adjusted emission rates to obtain seasonal VOC inventory
estimates for each state.9 This analysis does not consider usage of PFCs with diesel or
kerosene fuels, as these fuels contribute minimally to evaporation emissions due to the
very low volatilities of these fuels.
For each of the 50 states plus the District of Columbia, this EPA approach
produced the estimates for calendar year 1990 given in Table 2.1 .-7. Assuming no
changes (i.e., no controls), each of these estimates will increase by approximately 1.21
percent annually due to the increase in gasoline consumption predicted by the
NONROAD model.
Twelve states plus the District of Columbia (California, Connecticut, Delaware,
Maine, Maryland, New Hampshire, New Jersey, New York, Ohio, Pennsylvania, Texas,
Virginia, and Washington DC) already have or will implement controls on the design of
PFCs that will reduce HC emissions. Additionally, three other states (Massachusetts,
Rhode Island, and Vermont) are also planning to adopt the California PFC program.
Inventories include the impacts of these programs, as described in a technical support
document (EPA, 2006, Report No. EPA420-R-07-001).
Additionally, California has begun to adopt more stringent emission standards
that will require each PFC to emit (permeation plus evaporation) no more than 0.3 grams
of VOC per day for each gallon of capacity. This requirement will be effective July 1,
2007. Assuming that PFCs have a typical life of about five years on average, the "new"
versions of the PFCs should replace virtually all of the earlier versions by 2013. As these
state programs result in replacing the existing PFCs with lower-emitting PFCs, the
estimated national inventory of VOCs associated with PFCs will drop by about 20
percent.
2-9
-------
Final Regulatory Impact Analysis
To estimate the VOC emissions from PFCs upon implementation of the final rule,
we made the following three changes to our inventory estimates:
1. Since the final rule makes it unlikely for a newly designed PFC to be left in the
"open" position, we altered the distribution of the cans (from the California
survey) to 100 percent "closed." This change reduced the VOC emissions from
both evaporation as well as spillage during transport. (Note, the 15 states plus the
District of Columbia that are adopting the California PFC rules already had this
change applied. So, this affected the VOC emissions from only PFCs in the other
35 states.)
2. This final rule also produces changes to the design of the individual PFCs that are
expected to reduce the spillage by 50 percent when these PFCs are used to refuel
individual pieces of equipment. Again, this emission reduction was already
included in the base case for those states that are adopting the California rules.
Therefore, only the PFCs in the remaining 35 states contributed to our estimated
reductions of spillage.
3. Finally, the final rule includes a maximum emission rate of 0.3 grams per gallon
per day for the new PFCs. We used this emission standard to estimate the total
permeation plus evaporative emissions from each newly designed PFC. Only
California has adopted this requirement. Thus, the effect of this final national
requirement applies to the remaining 49 states.
The change in VOC emissions was then calculated by subtracting the emissions
(on a state-by-state basis) estimated using these preceding three changes from our base
estimates. The national estimate was simply the sum of the 50 individual state (plus DC)
estimates. The national pre- and post-control inventories are presented in Chapter 2.1.2
below.
2.1.2 Emission Reductions of Criteria Pollutants Resulting From Controls
2.1.2.1 Light-Duty Gasoline Vehicles
We are finalizing as proposed a 20° F FTP emission standard for non-methane
hydrocarbon (NMHC) emissions from spark ignition vehicles of 0.3 grams per mile for
light-duty vehicles and trucks that weigh 6000 pounds or less and a 0.5 gram per mile
standard for vehicles that weigh more than 6000 pounds (including medium-duty
passenger vehicles; i.e., "MDPVs"). The standard will be applied to a manufacturer on a
sales-weighted fleet-wide basis. Furthermore, the standards will be phased in over a
period of time following the schedule found in Table 2.1.-8.
The resulting reductions were modeled based upon the above standard and the
phase-in period. This was done as outlined in Section 2.1.1.1 with an external data file
provided as input to MOBILE6.2 that altered MOBILE6.2 start emission factors for Tier
2-10
-------
Final Regulatory Impact Analysis
2 vehicles only. MOBILE6.2 was then used with NMIM (using fuel properties which
reflect RFS, as described in Section 2.1.1) to generate county and nationwide inventories
for the control case. When the standard is fully phased in we expect a 60% reduction in
start emissions in gasoline-fueled vehicles that have a gross vehicle weight rating
(GVWR) of less than or equal to 6000 Ibs and a 30% reduction for gasoline-fueled
vehicles that have a GVWR greater than 6000 Ibs. The impact on future nationwide
VOC inventories is found in Table 2.1 .-9. Table 2.1 .-10 shows the impacts on a state-by-
state basis in year 2030.
2-11
-------
Final Regulatory Impact Analysis
Table 2.1-7. PFC Emissions (Tons VOC per Year) by Source Type (for 1990)
State
AK
AL
AR
AZ
CA
CO
CT
DC
DE
FL
GA
HI
IA
ID
IL
IN
KS
KY
LA
MA
MD
ME
Ml
MN
MO
MS
MT
NC
ND
NE
NH
NJ
NM
NV
NY
OH
OK
OR
PA
Rl
SC
SD
TN
TX
UT
VA
VT
WA
Wl
WV
WY
50-State
Refilling PFC at Pump
Vapor Displ
224.8
24.8
279.1
105.7
1,532.1
202.7
123.2
36.6
7.6
933.1
390.9
58.1
50.6
405.1
241.4
99.6
93.5
129.1
168.9
40.7
226.0
199.0
316.9
181.4
97.0
212.6
26.4
55.0
123.4
44.1
332.9
58.0
517.1
407.9
17.2
507.3
139.6
133.8
419.5
28.3
207.8
20.9
237.0
875.0
70.8
18.7
309.8
225.6
65.4
166.3
14.8
11,403.3
Spillage
15.0
1.9
22.9
8.5
133.9
18.9
12.0
3.1
0.7
72.5
32.4
4.0
4.9
36.3
19.8
8.3
8.5
11.1
12.1
4.3
21.1
19.1
29.6
15.5
7.4
19.0
2.6
5.2
10.6
4.5
30.0
5.2
47.5
31.5
1.8
41.1
12.1
12.8
38.5
2.7
15.1
2.0
18.6
67.6
6.7
1.9
27.6
20.5
5.4
16.4
1.5
972.1
Spillage
During
Transport
447.1
60.1
647.9
262.7
3,760.8
536.5
342.7
89.1
25.0
2,055.5
930.8
112.9
146.3
1,058.5
578.2
248.5
247.9
340.2
370.7
130.7
597.8
556.1
886.3
463.1
230.6
560.9
81.9
154.0
295.8
131.0
857.4
155.6
1,414.3
911.8
54.1
1,188.5
352.5
373.6
1,132.0
80.8
438.3
62.1
553.4
1,954.5
201.4
57.6
786.6
595.1
170.6
488.5
48.1
28,226.3
Refueling Equipment
Vapor
Displ
224.8
24.8
279.1
105.7
1,532.1
202.7
123.2
36.6
7.6
933.1
390.9
58.1
50.6
405.1
241.4
99.6
93.5
129.1
168.9
40.7
226.0
199.0
316.9
181.4
97.0
212.6
26.4
55.0
123.4
44.1
332.9
58.0
517.1
407.9
17.2
507.3
139.6
133.8
419.5
28.3
207.8
20.9
237.0
875.0
70.8
18.7
309.8
225.6
65.4
166.3
14.8
11,403.3
Spillage
1,010.8
103.2
1,630.1
533.4
9,284.9
1,319.4
837.2
217.9
56.6
5,050.7
2,234.7
285.3
316.0
2,458.1
1,353.8
541.9
567.2
727.8
771.4
297.7
1,520.6
1,322.3
1,966.1
992.3
476.0
1,271.6
160.5
336.4
759.6
299.6
2,041.2
358.7
3,095.2
2,179.0
103.9
2,843.0
824.4
864.5
2,644.5
188.9
1,066.9
124.8
1 ,245.3
4,645.6
418.4
127.6
1,986.7
1,399.7
345.8
1,089.3
92.7
66,389.1
Permeation Plus
Evaporation
4,286.7
776.6
3,936.1
2,813.4
19,682.1
2,137.2
1,422.5
514.9
235.1
14,664.5
5,918.5
1 ,208.2
780.1
5,764.9
3,914.8
1 ,886.4
1 ,457.6
2,914.7
5,178.9
620.4
2,528.1
2,561.3
5,253.7
3,281.1
2,997.4
3,427.2
506.1
911.4
1 ,362.6
572.4
4,049.9
1 ,050.8
8,473.6
6,950.0
302.1
7,500.9
2,322.6
1 ,889.7
6,498.5
422.5
3,981.0
398.1
4,944.1
15,730.9
1,208.1
296.3
3,853.6
3,174.0
1 ,700.5
2,512.5
265.7
181,040.0
Totals by
State
6,209.2
991.3
6,795.1
3,829.4
35,925.8
4,417.5
2,860.8
898.2
332.7
23,709.5
9,898.3
1,726.6
1,348.5
10,127.9
6,349.3
2,884.3
2,468.2
4,252.1
6,670.9
1,134.5
5,119.6
4,856.7
8,769.4
5,114.9
3,905.5
5,704.1
803.9
1,516.8
2,675.3
1,095.5
7,644.2
1,686.2
14,064.8
10,888.3
496.2
12,588.0
3,790.6
3,408.1
11,152.6
751.5
5,916.9
628.9
7,235.5
24,148.7
1,976.3
520.9
7,274.1
5,640.6
2,353.1
4,439.2
437.7
299,434.1
2-12
-------
Final Regulatory Impact Analysis
Table 2.1.-8. Phase-in Schedule for 20°F Standard by Model Year
Vehicle GVWR
(Category)
< 6000 Ibs
(LDV/LLDT)
> 60001bs HLDT
(and MDPV)
2010
25%
2011
50%
2012
75%
25%
2013
100%
50%
2014
75%
2015
100%
Table 2.1.-9. Impact on Nationwide VOC Emissions from Light Duty Vehicles and
Trucks of a 20 °F FTP Emission Standard for Non-Methane Hydrocarbons.
Year
1999
2010
2015
2020
2030
Tons Without Standard
4,899,891
2,990,760
2,614,987
2,538,664
2,878,836
Tons With Standard
N. A.
2,839,012
2,293,703
2,009,301
1,996,074
Reduction
N.A.
151,748
321,284
529,363
882,762
2-13
-------
Final Regulatory Impact Analysis
Table 2.1.-10. Impacts on State Light Duty Vehicle and Truck VOC Emissions of
20 °F FTP Emission Standard for Non-Methane Hydrocarbons in 2030.
State
AL
AK
AZ
AR
CA
CO
CT
DE
DC
FL
GA
HI
ID
IL
IN
IA
KS
KY
LA
ME
MD
MA
MI
MN
MS
MO
MT
NE
NV
NH
NJ
NM
NY
NC
ND
OH
OK
OR
PA
RI
SC
SD
TN
TX
UT
VT
VI
WA
WV
WI
WY
Reference Case
Tons
52,985
11,605
50,655
30,893
259,253
61,855
28,766
7,213
3,229
123,002
100,284
7,835
21,439
107,467
85,144
38,982
31,740
48,011
36,806
16,942
45,754
44,407
133,830
86,476
25,290
71,439
16,326
22,897
28,102
15,434
54,869
31,625
112,589
94,614
11,222
115,095
46,290
66,957
105,046
9,036
47,950
11,920
70,526
159,952
36,024
9,873
80,579
108,386
16,993
64,663
10,566
Control Case
Tons
41,636
6,299
39,988
23,185
185,702
40,187
17,706
4,639
2,146
110,498
75,270
7,626
13,588
67,221
57,529
25,254
22,190
32,867
30,134
10,247
29,230
25,717
86,171
51,148
19,642
49,467
10,015
15,077
20,771
9,413
35,834
22,152
67,387
69,429
6,752
73,824
34,712
46,226
67,864
5,641
36,058
7,443
51,999
126,799
24,050
5,906
53,729
74,481
10,833
37,816
6,574
Reduction
in Tons (a)
11,349
5,306
10,667
7,708
73,551
21,667
11,059
2,574
1,082
12,504
25,014
209
7,851
40,245
27,615
13,729
9,550
15,144
6,672
6,695
16,525
18,690
47,659
35,328
5,648
21,972
6,311
7,819
7,330
6,022
19,035
9,473
45,202
25,185
4,470
41,271
11,578
20,731
37,183
3,395
11,892
4,476
18,528
33,154
11,974
3,967
26,850
33,905
6,160
26,847
3,992
Percent
Reduction
21
46
21
25
28
35
38
36
34
10
25
3
37
37
32
35
30
32
18
40
36
42
36
41
22
31
39
34
26
39
35
30
40
27
40
36
25
31
35
38
25
38
26
21
33
40
33
31
36
42
38
(a) Values calculated prior to rounding reference and control values.
2-14
-------
Final Regulatory Impact Analysis
Test data show that the controls on cold temperature hydrocarbon emissions will
have the ancillary benefit of reducing PM emissions as well. Emissions generated during
cold temperature starts tend to be elevated due to a combination of a cold catalyst and
excess fuel in the combustion chamber. These factors increase emissions of benzene and
other hydrocarbons, and at the same time allow for unburned or pyrolized fuel to be
emitted.
A number of source apportionment studies have indicated previously that
emissions from vehicles starting at cold temperatures contribute disproportionately to
ambient PM2 5. For instance, the Northern Front Range Air Quality study conducted in
the Denver, CO area during the winter of 1997 estimated that, on average, 12% of
ambient PM2.5 could be attributed to cold start light-duty gasoline vehicle emissions.10
At this point, the PM emission factors in MOBILE6.2 for PM from light-duty
gasoline vehicles are not sensitive to temperatures. However, as outlined above, the
emission factors for hydrocarbons and gaseous toxics are temperature-dependent.
In order to estimate the expected emission reductions in PM as a result of the cold
temperature standards, we evaluated the relationship between PM and NMHC in Tier 2
vehicles operating at different temperatures. All emissions benefits of the cold
temperature standard are expected to affect only the cold temperature starting emissions.
As such, all analyses were restricted to Bag 1. However, similar results were obtained
when using full weighted FTP results.
First, data from the only extant testing program of Tier 2 vehicles at multiple
temperatures was obtained from Southwest Research Institute.n Figure 2.1 .-1 shows the
PM emission factors as a function of temperature. Like NMHC, PM emission factors
increase exponentially with lower temperatures through the entire range of testing.
2-15
-------
Final Regulatory Impact Analysis
Figure 2.1.-1. FTP Bag 1 PM Emissions vs. Temperature, Tier 2 Vehicles
I I I I
0 20 40 60
ill l l l l
E
O)
CL -2 -
-4 -
as
m
-6 -
-10 -
Vehicle 4
0
8
0
o
Vehicle 1
8
8 o
0 9
o
o
Vehicle 5
0
8
Vehicle 2
o
o
o
Vehicle 6
8.
o o
o
o o
Vehicles
o
©
o
o
i i i r
- -2
- -6
- -10
i i i r
0 20 40 60 0 20 40 60
Temperature (F)
Figure 2.1.-2 illustrates the relationship between FTP Bag 1 NMHC and PM emission
factors in this test program. Lower temperature tests are found to the upper right corner,
corresponding to elevated emissions of both NMHC and PM. The symbol used for each
data point represents the different vehicles in the test program. As shown, there is a
clear, linear association. Thus, we concluded that estimated reductions in PM as a result
of the hydrocarbon emission controls in this rule could be estimated by applying a PM to
NMHC ratio to the estimated reduction in NMHC.
2-16
-------
Final Regulatory Impact Analysis
Figure 2.1.-2. FTP Bag 1 PM and FTP Bag 1 NMHC for Various Tier 2 Vehicles
"0:1
O)
ra
ffi
oo —
A
$ ^<
.»' Ą *
v 4-Q v
V 0 A + V
V •%> o
V v
X A o
1 1 1 1
-3 -2 -1 0
A +
I I
1 2
Bag1 NHMC-ln(g/mi)
Plot Icons are Vehicle-Specific
In order to determine an appropriate PM/NMHC ratio for calculating PM
reductions from NMHC reductions during cold start conditions, we employed mixed
models with random vehicle terms.12 We fit several models to the data, treating the
PM/NMHC ratio as a dependent variable. In summary, the model fit to the data was:
Y=n+T+b+e
Here, Y is a matrix of dependent variables (emission factors);
|l is the intercept term or "grand mean";
b is the change in emission factor associated with discrete testing temperatures;
T is the vehicle effect, normally distributed around zero;
e is the random error term (normally distributed).
Tests in which temperature was treated as a continuous variable were also employed.
Overall, the b term was found to be significant only at 75° testing, and this may have
been due to random measurement errors in the PM/NMHC ratio as a result of very low
emissions at 75°. The b term became insignificant when it was allowed to vary randomly
by vehicle. In addition, because the standards apply only to cold starting conditions, the
effect on the ratio at 75° is not relevant to changes in overall emissions. Therefore, we
used the mean PM/NMHC ratio of 0.022 to calculate the expected ancillary reductions in
PM. The 95% confidence interval for the mean was 0.020 - 0.024.
2-17
-------
Final Regulatory Impact Analysis
Using this number, the expected reductions in PM from this rule are estimated to
be 7,068 tons in 2015, 11,646 tons in 2020 and 19,421 tons in 2030. These calculations
provide initial evidence that the potential public health impacts of this final rule are
substantial.
In subsequent test programs demonstrating the feasibility of the NMHC standards
in this final rule, the test vehicles exhibited substantial reductions in PM emissions as
well. The test results from the two feasibility vehicles fall within the range of those
derived from the SwRI test program. These PM emission reductions at 20° F were of
similar magnitude as those predicted by the above calculation. Furthermore, examining
the feasibility demonstration results, the PM/NMHC ratio of the emission reductions
were both close to the value of 0.022 used in the above calculation, spanning either side
of the original number (0.010-0.025).A In the first feasibility test program, the vehicle
reflected a unique control technology that requires careful coordination among the engine
air-fuel ratio and secondary air injection timing and air volume to provide the maximum
emission benefits. That feasibility program was a "proof of concept" study that did not
have the ability to fully explore ideal control coordination and sizing of the emission
control system. In the second feasibility study, the vehicle only received recalibration to
achieve emission reductions, which is likely to be more representative of the emission
control technologies that will be employed for the majority of vehicles. Despite different
technologies being used in the feasibility tests, the six current unmodified production
vehicles tested in the SwRI test program are considered to be more representative of
emission control technologies found throughout the fleet.
Several factors are not accounted for in the emission reduction estimation
procedures, which adds uncertainty to the level of emission reductions reported here.
First, if manufacturers employ control technologies that differ substantially from those in
the two feasibility test programs, actual emission reductions could differ from the
estimates here. Second, actual PM reductions may be affected by the extent to which
different vehicle or engine technologies penetrate into the vehicle market (such as hybrid
electric drivetrains and direct injection gasoline engines).
2.1.2.2 Portable Fuel Containers
The PFC controls in this final rule will also reduce emissions of hydrocarbons. As
noted in Section 2.1.1.2, fifteen states plus the District of Columbia have adopted
controls on PFCs independent of the controls being finalized in this rule. In Figure 2.1 .-
3, we have graphed the estimated annual national VOC emissions (in tons) associated
with PFCs for the following three scenarios:
— a base scenario in which no PFC controls are used illustrated with the dotted
(black) line,
' This range derives from the feasibility tests with the lowest measured NMHC emissions.
2-18
-------
Final Regulatory Impact Analysis
a scenario in which only those 15 states plus DC have implemented PFC controls
illustrated with the solid (blue) line, and
a scenario in which the PFC controls are implemented nationwide illustrated with
the dashed (red) line
Figure 2.1-3. Comparison of PFC Control Scenarios
Annual Nationwide VOC Emissions (Tons) from PFCs by Calendar Year
500,000
400,000
300,000 »
200,000
100,000
o
n
o
o
n
— * —
- - • Bas
_ O*~
r - - - * *
v
\
e
:e Programs
^™ ~ Nationwide
•
— «
* *
_^^^
\
\
\- — —
^ * *
- *
^^
— — "" ""
1990 2000 2010 2020 2030
Calendar Year
As noted in Section 2.1.1.2, the estimates of the VOC inventory in the basic scenario are
increasing (annually) at a rate of about 1.21 percent. The scenario containing just the
state programs has the estimated VOC inventory increasing at an annual rate of about
1.33 percent once all of the programs are phased in. Similarly, the scenario in which
nationwide requirements (of this RIA) are phased in exhibit an annual increase in the
VOC inventory of about 1.44 percent after phase-in.
Table 2.1.-11 compares the estimated national (annual) inventory of PFC-related VOC
with the control program to a reference case scenario that includes only State level
controls.
2-19
-------
Final Regulatory Impact Analysis
Table 2.1.-11. Nationwide Annual PFC VOC Emissions (tons)
Calendar
Year
1999
2007
2010
2015
2020
2030
With NO EPA
PFC Controls
325,030
327,320
316,756
329,504
353,470
402,916
With EPA
PFC Controls
NA
NA
256,175
127,157
137,175
157,661
Reduction
NA
NA
60,580
202,347
216,294
245,255
2.1.3 Strengths and Limitations of Criteria Pollutant Inventories
As previously discussed, the MSAT final rule inventories were estimated using
fuel properties developed for the RFS proposed rule. Because the RFS and MSAT
inventories were developed in relatively close proximity, we highlight in this section
some minor differences in methodologies, as well as uncertainties related to the RFS fuel.
Though these methodologies contribute to different baseline RFS and MSAT inventories,
they have little impact on our estimates of emission reduction benefits associated with
this MSAT final rule.
Future Volume of Renewable Fuel - Under the RFS mandate, a minimum volume
of ethanol must be blended in gasoline. However, the Energy Information Agency (EIA)
has forecasted that market forces alone will push ethanol use well beyond the minimum
volume required by the RFS mandate13. The volume of renewable fuel forecasted by
EIA, and not the RFS program mandate, was used as the baseline for developing RFS
fuel properties used in MSAT inventories. Though there are uncertainties related to the
future volume of renewable fuel use (and regional allocation), the effects on the emission
reduction benefits achieved by the MSAT final rule are likely minimal. Furthermore, as
presented in the RFS Draft RIA (DRIA)2, inventories for criteria pollutants never differ
by more than a few percent between the RFS mandate volume scenario (7.2 billion
gallons of national ethanol use) and the EIA-predicted scenario (9.6 billion gallons of
national ethanol use).
Ethanol Effects on Gasoline Properties - The MSAT rule inventories are based
on fuel properties estimated under the RFS program. In the RFS draft regulatory impact
analysis, we based our assessment of the effects of ethanol on gasoline fuel properties on
annual fuel survey data provided by the Alliance of Automobile Manufacturers. We
limited the analysis to cities for which data from both ethanol-blend and non-ethanol
gasoline samples were available. These criteria reduced the data to samples from four
cities, limiting the national geographic representation of the fuel effects. In addition,
there was no distinction to indicate whether the fuel came from multiple refineries within
any given city, which eliminates refinery-specific effects. However, we checked the
results against the AAM data from all U.S. cities, comparing all conventional gasoline
2-20
-------
Final Regulatory Impact Analysis
non-ethanol blends to all conventional gasoline ethanol blends. The results were very
similar to those from the four cities.
Seasonal and Permeation Effects- For MSAT inventories, we interpolated
summer and winter fuel properties to all 12 months and ran each month in NMEVI
individually. Due to time constraints during development of RFS proposal inventories,
we ran NMEVI for only January and July, using fuel survey data collected in summer and
winter, and assumed that emissions for those two months could be extrapolated to
represent the entire year. We estimated RFS annual emissions inventories by summing
the two monthly inventories and multiplying by six. For RFS, we also added the effect of
ethanol on permeation from onroad non-exhaust emissions. Again, these different
methodologies have minimal effect on the emissions benefits associated with this final
rule.
Light-Duty Gasoline Vehicles - Emission factors for hydrocarbons in the
MOBILE model are based on tens of thousands of tests under a wide variety of
conditions, and account for leaking fuel systems, aggressive driving, air conditioner use
and a variety of other parameters. These data are supported by over 50 technical reports,
and many of them received extensive scientific peer review. The strengths and limitations
of the MOBILE model have been evaluated by the Coordinating Research Council and
the National Research Council.14'15
There are significant uncertainties in emission inventories resulting from the use
of national default data rather than local inputs, as well as "top-down" allocation schemes
in estimating toxic emissions. Examples include use of national default vehicle
registration distributions, default average speed distributions, and use of county level
population data to allocate State or urban level VMT. Furthermore, emission rates were
modeled for only a subset of the total number of counties. Therefore, we do not fully
capture all local conditions, introducing additional uncertainty into the inventories.
Also, it should be noted that there are greater uncertainties in projection year
estimates. Estimates of emissions from advanced technology vehicles and engines that
will comply with planned future emission standards include assumptions regarding levels
of emission deterioration and performance under various conditions. Also, vehicle miles
traveled are estimated using economic projections with similar inherent limitations.
The revised estimates of cold start VOC emissions are based on a robust dataset at
temperatures of 20°F and above. At lower temperatures, however, data are more limited
and the magnitude of cold temperature effects is not as certain. Similarly, the estimate of
PM reductions from NMHC cold temperature controls are based on limited data,
although PM shows a very strong correlation with NMHC. Future control strategies may
also employ mechanisms that result in different PM/NMHC ratios than found in existing
vehicles.
Finally, the MSAT inventories used the fuel effects contained in MOBILE6.2. In
the RFS proposal, we accounted for uncertainties in MOBILE6.2 fuel effects by adjusting
2-21
-------
Final Regulatory Impact Analysis
the model output for exhaust VOC and NOx emissions by applying EPA Predictive
Model fuel effects instead. The MSAT inventories do not use the Predictive Model
effects since the use of the Predictive Model would have little impact on estimates of
emission benefits of this rule.
Portable Fuel Containers - To estimate PFC inventories we were able to build on
survey and test data collected by the California Air Resources Board. We also developed
inventories using a "bottom-up" approach which provides flexibility and permits very
detailed fine-tuning of the various scenarios. However, the inventory involved many
assumptions, including refueling activity and temperature effects. Spillage occurring
when non-road equipment is refueled is a significant source of VOC emissions. We are
assuming (from EPA's NONROAD model) that spillage is a constant 17 grams for each
refueling event. We are also assuming that each refueling event occurs when the fuel
tank on that piece of equipment is empty. However, if the user "tops off" the fuel tank
prior to each use, then we are underestimating the total VOC emissions.
Another assumption relates to whether inactive PFCs are stored with fuel. For
example, we assumed that a residence that uses a PFC to only fuel a lawn mower
(perhaps six months of the year) will have that PFC empty the remainder of the year (i.e.,
no permeation or evaporative emissions). However, if that PFC were to contain a small
amount of gasoline for those non-mowing months, then we are underestimating the total
inventory.
Uncertainty in the characterization of the population of PFCs (i.e., commercial
versus residential usage, open versus closed, metal versus plastic) is the major source of
uncertainty in our estimates of the inventory of VOCs from PFCs. Our characterization
of the population of PFCs is based on surveys performed by the Air Resources Board
(ARE) of California. We used the same distribution of open versus closed PFCs
determined by ARE. Since the PFC population in the rest of the country might not be
exactly like California's, we performed a sensitivity analysis to determine the effects of
varying that distribution. We found that even relatively large changes in that distribution
produced changes in estimated total VOC of less than 13 percent.16 Other source of
uncertainty include estimates of the frequency of refilling of containers, estimates of
effects of ambient temperature on vapor displacement and spillage, estimates of effects of
RVP on vapor displacement, impacts of temperature of the fuel itself on emissions, and
estimates of the amount of spillage during refilling.
2.2 Air Toxics
2.2.1 Emission Inventories Used in Air Quality Modeling
The data and methods employed to develop the county-level air toxics inventories
used for air quality, exposure and risk modeling to support this final rule are discussed in
detail in the EPA Technical Report, "National Scale Modeling of Air Toxics for the Final
Mobile Source Air Toxics Rule; Technical Support Document," Report Number EPA-
454/R-07-002. All underlying data and summary statistics are included in the docket for
2-22
-------
Final Regulatory Impact Analysis
this rule. Major revisions have been made to the inventories used for air quality,
exposure, and risk modeling since proposal. These revisions include:
• Revisions to cold temperature start emissions for gaseous air toxics in Tier
1 and later highway gasoline vehicles
• Estimation of air toxic emissions for nonroad equipment using
NMEVI2005 rather than NMIM2004
• Inclusion of air toxic emissions from portable fuel containers
• Revision of the benzene and naphthalene inventories for gasoline
distribution based on recent analysis of benzene in gasoline vapor emitted
during the distribution process
While cold temperature emissions and portable fuel container emissions were included in
analyses of emission benefits and cost-effectiveness for the proposal, the proposal did not
use NMTM2005 for nonroad equipment or include changes to the gasoline distribution
emissions estimates based on recent analyses. While the air quality modeling inventories
for the final rule included the above improvements, it did not include impacts of the
renewable fuel standard, as the inventories were developed well in advance of the
proposal for that standard. Furthermore, the modeling accounted only for the 0.62
percent standard, but not the 1.3 vol% maximum average. Thus, the emission reductions
from highway vehicles and other sources attributable to the fuel benzene standard are
underestimated in many areas of the country, particularly in areas where fuel benzene
levels were highest without control, such as the Northwest. It should be noted that the
inventories used in the proposal were presented in a peer reviewed journal article.17
The following sections summarize the methods used to develop the air quality
modeling inventories, including details of the major revisions listed above, and also
present inventory results. While air quality, exposure, and risk modeling was done for
years 1999, 2015, 2020, and 2030 (with modeling for 1999 done as the National Scale
Air Toxics Assessment), reference case inventories were also developed for 2010 in order
to better assess emission trends over time. Control case modeling which included
cumulative impacts of the controls being finalized in this rule was done for 2015, 2020
and 2030. For the reference case, we modeled all air toxic compounds listed in section
112 of the Clean Air Act for which we had adequate data to estimate emissions. Table
2.2.-1 lists the pollutants included in these inventories which were used in subsequent
modeling of air quality, exposure, and risk. For the control case, we modeled a smaller
subset of pollutants as discussed below. Emission inventories included stationary
sources, highway vehicles, and nonroad equipment.
2.2.1.1 Methods Used to Develop Air Toxics Inventories for Air Quality Modeling
2.2.1.1.1 Highway Vehicles
Highway gasoline vehicle inventories for all emission types except refueling were
developed using a modified version of NMIM2005.18'19'20 NMIM develops inventories
using EPA's MOBILE6.2 emission factor model for highway vehicles, EPA's
2-23
-------
Final Regulatory Impact Analysis
NONROAD emissions inventory model for nonroad equipment, and model inputs stored
in data files. Model inputs include data such as temperatures, fuel properties, vehicle
registration distributions, inspection and maintenance programs, vehicle miles traveled,
and toxics inputs in the form of toxic to volatile organic compound (VOC) ratios, toxic to
particulate matter (PM) ratios, or toxic emission factors. The toxics inputs were
developed from a variety of emissions testing programs conducted by EPA, States, and
industry over many years (see Section 2.2.1.1.6 for more information). Details on data
sources can be found in the documentation for the National Emissions Inventory.
Refueling emission estimates for 2015 and later years were carried over from the
inventories used for air quality modeling in the proposal. For 1999 and 2010, benzene
refueling emission estimates were not available, so benzene refueling emissions were
backcast from 2015, using ratios of VOC refueling emissions in 1999 or 2010 to 2015
VOC. The approach used to do this is discussed in detail in the technical support
document.
NMEVI was modified to include the hydrocarbon start emission adjustment factors
discussed in Section 2.1.1.1. Since the algorithms used to calculate toxic to hydrocarbon
emission ratios in MOBILE6.2 do not vary with temperature, reductions in hydrocarbon
emissions result in proportional reductions in air toxic emissions.
The assumption in MOBILE6.2 that reductions in air toxic emissions are
proportional to hydrocarbon emission reductions was based on testing done at
temperatures ranging from -20 to 75 °F in EPA's Office of Research and Development in
the late 1980's.21'22 These studies found that, overall, the composition of hydrocarbon
emissions did not vary appreciably with temperature, although fractions of formaldehyde
increased somewhat with lower temperature in port fuel injected vehicles. The validity of
the assumption was re-evaluated for later model vehicles.
EPA's Office of Research and Development recently tested several late model
vehicles at the same temperature ranges cited above.23'24'25 These results can be used to
reevaluate the validity of the assumption discussed above. The results of the test program
are unpublished, but are included in the docket for the rule. Vehicles included in the test
program were a 1993 Chevrolet Cavalier, a 1987 and 1993 Ford Taurus, a 1996 Chrysler
Concord, a 2001 Ford Focus, a 1993 Buick Regal, and a 2001 Dodge Intrepid. This test
program found increasing emissions of individual air toxics at lower temperatures.
Benzene and 1,3-butadiene emissions increased proportionally with hydrocarbon
emissions, with a very strong correlation. However, correlations were not as strong with
aldehydes. Results from the 1993 Cavalier and 1993 Taurus found a statistically
significant correlation for acetaldehyde but not for formaldehyde, whereas analysis of
data from the other vehicles found a correlation for formaldehyde but not acetaldehyde.
A major vehicle manufacturer also recently tested two Tier 2 compliant vehicles
at 75 and 20 °F. Although the data are confidential, they show emission of air toxics
increase at the same rate as hydrocarbons, with a very high correlation.
2-24
-------
Final Regulatory Impact Analysis
A third source of data is testing done by Southwest Research Institute for EPA on
four model year 2005 vehicles - a Ford F-150, a Mazda 3, a Honda Odyssey and a
Chevrolet Equinox.26 The four vehicles were tested at 0, 20 and 75 °F. Benzene and 1,3-
butadiene correlated very strongly with hydrocarbon emissions, with r-square values
above 0.9. Benzene accounted for about 3.6 percent of exhaust non-methane
hydrocarbon emissions at all temperatures, while 1,3-butadiene accounted for about
0.3%. However, formaldehyde and acetaldehyde fractions appeared to decrease with
decreasing temperature. When data for the largest vehicle, the Ford F-150, were
removed, there seemed to be a stronger correlation between aldehyde emissions and non-
methane hydrocarbons. This could be because this larger engine is running richer during
cold starts than the other vehicles, and not enough oxygen is available for aldehyde
formation.
Recent EPA testing of a Chevrolet Trailblazer, with its engine recalibrated to
meet the cold temperature standard, showed reductions in acetaldehyde and acrolein
proportional to the reduction in VOC. Formaldehyde was also reduced, but was not
reduced as much as acetaldehyde and acrolein. Other air toxic compounds, including
benzene, were not included in this testing. Figure 2.2.-1 depicts the relationship between
carbonyl compounds and NMHC.
Figure 2.2.-1. Regressions of Carbonyl Emissions Versus NMHC for Chevrolet
Trailblazer Recalibrated to Meet Cold Temperature Standard.
Acetaldehyde vs. NMHC
1000 2000
3000
NMHC
4000 5000 6000
2-25
-------
Final Regulatory Impact Analysis
Acrolein vs NMHC
1000 2000 3000 4000
NMHC (mg/mi)
5000 6000
Formaldehyde vs. NMHC
10 -, -
E °
| 6
O 4
O
I 2
o
•
+ v- 7F OCIY+ fi P17 a. '
f » R^= 0.0082
* I
0 1000 2000 3000 4000 5000 6000
NMHC (mg/mi)
Given available data, we have concluded it is reasonable to retain the
assumption that ratios of toxic emissions to hydrocarbon emission do not vary with
temperature. However, as more data become available, this assumption should be
reevaluated, particularly for aldehydes.
Within the MOBILE6.2 model, six MSATs (benzene, formaldehyde,
acetaldehyde, 1,3 butadiene, acrolein, and methyl tertiary butyl ether [MTBE]) can be
calculated directly by including detailed fuel parameters. These parameters are stored in
the NMIM database. The fuel parameters are: sulfur content, olefins content, aromatics
content, benzene content, E200 value, E300 value, oxygenate content by type, and
oxygenate sales fraction by type.8 Since these fuel parameters are area-specific, EPA
developed county-level inputs for each of these parameters. Fuel parameters were
collected for winter and summer seasons using a number of different data sources. These
sources include the Alliance of Automobile Manufacturers, Northrop Grumman Mission
Systems (formerly TRW Petroleum Technologies), and EPA reformulated gasoline
! E200 and E300, represent the percentage of vapor that gasoline produces at 200 and 300 °F, respectively.
2-26
-------
Final Regulatory Impact Analysis
surveys. Documentation for the National Emissions Inventory (NEI)27 describes the
development of the fuel parameter database used with MOBILE6.2 in detail. The fuel
parameter data through 1999 are posted at the following website:
ftp://ftp.epa.gov/EmisInventory/fmalnei99ver3/haps/datafiles/onroad/auxiliary/
Although fuel parameter data were prepared for only two seasons (summer and
winter), NMIM uses monthly rather than seasonal fuel parameters, and parameters for
spring and fall months are estimated by interpolating from summer and winter data. In
addition, documentation of the fuel parameters used in NMIM was compiled in 2003
(Eastern Research Group, 2003), and subsequently, a number of changes were made,
based on comments from States. These changes are documented in the change log for
NMIM, dated May 14, 2004. This change log is included in the docket for this rule,
along with the original documentation. In general, multiplicative adjustment factors were
used to calculate future year gasoline parameters (i.e., future year parameter = base year
parameter x adjustment factor). However, additive adjustment factors were used to
calculate future year parameters for E200, E300, and oxygenate market shares (i.e., future
year parameter = base year parameter + adjustment factor). These adjustment factors
were developed using results of several refinery modeling analyses conducted to assess
oŁ OQ ^n
impacts of fuel control programs on fuel properties. ' ' The database used for this
assessment assumes no Federal ban on MTBE, but does include State bans. Also, it did
not account for recent increases in the use of ethanol oxygenated gasoline, the renewable
fuels mandate in the recent Energy Policy Act, or the 1.3 vol% maximum average fuel
benzene level.
MOBILE6.2 also has a command (ADDITIONAL HAPS) which allows the user
to enter emission factors or air toxic ratios for additional air toxic pollutants. Emission
factors for the other HAPs in Table 2.2.-1 were calculated by MOBILE6.2 through the
use of external data stored in the NMIM database, specifying emission factors for these
pollutants in one of three ways: as fractions of volatile organic compounds (VOC),
fractions of PM, or by supplying the basic emission factors. The ratios used with this
command must be expressed as milligrams of HAP per gram of VOC or PM. Gaseous
hydrocarbons were estimated as fractions of VOC. Polycyclic aromatics hydrocarbons
(PAHs) were calculated as fractions of PM, although the data used to calculate mass
ratios included both gas and particle phase PAH emissions. Metals were estimated using
basic emission factors. Evaporative emissions (e.g., toluene, xylenes) can only be
estimated as fractions of VOC. Because toxic to VOC ratios for several gaseous HAPs
vary between baseline (i.e., non-oxygenated) gasoline and gasoline oxygenated with
MTBE or ethanol, separate ADDITIONAL HAPS input data were developed for: 1)
baseline gasoline; 2) gasoline oxygenated with 2% MTBE by weight (e.g., Federal
reformulated gasoline); 3) gasoline oxygenated with 2.7% MTBE by weight (e.g., winter
oxygenated gasoline); and 4) gasoline oxygenated with 3.5% ethanol by weight
(gasohol).
2-27
-------
Final Regulatory Impact Analysis
Table 2.2.-1. Air Toxics Included in Emission Inventories and Used for Air Quality,
Exposure, and Risk Modeling.
1,3 -Butadiene
2,2,4-Trimethylpentane
Acenaphthene
Acenaphthylene
Acetaldehyde
Acrolein
Anthracene
Benzene
Benz(a)anthracene
Benzo(a)pyrene
Benzo(b)fluoranthene
Benzo(g,h,i)perylene
Benzo(k)fluoranthene
Chromium
Chrysene
Dib enzo(a, h)anthracene
Ethyl Benzene
Fluoranthene
Fluorene
Formaldehyde
n-Hexane
Indeno(l,2,3,c,d)-pyrene
Manganese
Methyl tert-butyl ether (MTBE)
Naphthalene
Nickel
Phenanthrene
Propionaldehyde
Pyrene
Styrene
Toluene
Xylenes
Vehicle miles traveled used in this assessment were those developed for the Clean Air
Interstate Air Quality Rule (CAIR).
31
For years 2015, 2020, and 2030, inventories were developed that reflected the
cumulative impacts of the controls being finalized in this rule. These control case
inventories included all the pollutants in Table 2.2-1.
To develop these inventories, NMIM was rerun with revised gasoline fuel
parameter inputs for fuel benzene and aromatics levels, as well as estimated emission
reductions from new start emission standards for hydrocarbons. The fuel parameter
inputs were revised based on refinery modeling done for the proposed rule, rather than
the final rule refinery modeling discussed in Chapter 6 of the this document. As part of
the refinery modeling, average fuel properties under a 0.62% fuel benzene standard, with
no maximum average level, were estimated for each Petroleum Administration for
Defense District (PADD). Average fuel benzene levels for conventional gasoline and
reformulated gasoline in each PADD before and after implementation of the standards
were used to develop multiplicative factors which were applied to the reference case fuel
benzene levels for each county in the NMTM database. These multiplicative factors are
summarized in Table 2.2.-2. Although California is part of PADD5, it was treated
separately, since California has its own reformulated gasoline program. Table 2.2.-3
compares average fuel benzene levels for each PADD used in the air quality modeling
2-28
-------
Final Regulatory Impact Analysis
inventories, compared to levels predicted by refinery modeling for the final rule, which
assumes a 1.3 vol% maximum average. If the refinery modeling data had been available
to be used in the air quality modeling inventories, benzene emission reductions from the
fuel standard would have been significantly greater in PADDs 2 and 5, but slightly lower
inPADDs 1 and 3.
The refinery modeling also indicated that the reduction in fuel benzene levels
would result in small proportional decreases in aromatics levels as well.32 Thus
aromatics levels were adjusted using the additive factors calculated as follows:
Additive Factor = 0.77 (BZ(control) - BZ(ref)) (1)
Where BZ = benzene
An Excel workbook, designated "fuel changes.xls", summarizes the control and reference
case fuel benzene and aromatics levels used for 2015, 2020, and 2030. This file is
included in the docket for the rule. We also checked the control case fuel benzene levels
to make sure the nationwide average level was close to the standard. We did this by
weighting county fuel benzene level by VMT as a surrogate for fuel sales. The resulting
nationwide average level was a little under 0.63%, very close to the standard. The
refinery modeling methodology is discussed in Chapter 9 of the Regulatory Impact
Analysis. Since the reduction in fuel benzene changes well below one percent of the
gasoline, the level of uncertainty in the impacts on other fuel parameters and emissions is
quite small.
Once fuel parameters were developed for the control case, NMTM was rerun with
data files that included new start emission standards for hydrocarbons. Output included
exhaust emissions, non-refueling evaporative emissions, and refueling evaporative
emissions.
It should be noted that the inventory used for air quality modeling included an
error in contractor-supplied input files for 13 Northeastern states. This error had a small
impact on reference case inventories, but the impact on estimates of emission reductions
with controls was insignificant. In addition, the control case inventory for 2015 assumes
that the fuel program is fully phased in, which is a simplification of the actual phase-in.
For more information about fuel program phase-in, refer to Chapter 6 of the Regulatory
Impact Analysis.
2-29
-------
Final Regulatory Impact Analysis
Table 2.2.-2. Average Fuel Benzene Level (Volume Percent) by PADD with
Implementation of Fuel Benzene Standard (CG - Conventional Gasoline; RFG -
Reformulated Gasoline)
Reference
Case
Control Case
Multiplicative
Factor
CG
RFG
CG
RFG
CG
RFG
PADD
1
0.91 %
0.59%
0.55%
0.54%
0.60
0.92
PADD
2
1.26%
0.80%
0.68%
0.71%
0.54
0.89
PADD
3
0.95%
0.57%
0.54%
0.55%
0.57
0.96
PADD
4
1.47%
1.05%
0.93%
0.62%
0.63
0.59
PADD
5
1.42%
0.65%
0.85%
0.60%
0.60
0.92
Calif.
0.62%
0.62%
0.61%
0.61%
0.98
0.98
Table 2.2-3. Comparison of Average Fuel Benzene Level (Volume Percent) by
PADD In Inventories Versus Final Rule Refinery Modeling..
Average Fuel
Benzene Level
Assumed in
Inventories
(0.62% standard)
Average Fuel
Benzene Level,
Final Rule
Refinery
Modeling, with
0.62% Standard
and 1.3 vol%
Maximum
Average
CG
RFG
CG
RFG
PADD1
0.55 %
0.54%
0.61%
0.54%
PADD 2
0.68%
0.71%
0.62%
0.60%
PADD 3
0.54%
0.55%
0.63%
0.55%
PADD 4
0.93%
0.62%
0.85%
0.62%
PADD 5
0.85%
0.60%
0.65%
0.60%
Calif.
0.61%
0.61%
0.61%
0.61%
For highway diesel vehicles, we used a different approach than we used for
gasoline vehicles. NMIM2004 outputs for 1999, 2007, 2010, 2015 and 2020 were used
to develop ratios of future year to 1999 air toxic inventories. These were then applied to
1999 NEI inventory estimates by SCC, county and HAP:
2-30
-------
Final Regulatory Impact Analysis
f ± 20XX
where PF20xx is the projection factor for 2007, 2010, 2015, 2020, or 2030, E20xx is the
emissions for the corresponding year and Ł1999 is the 1999 emissions. Highway diesel
vehicle air toxic emission estimates remained unchanged from the proposal.
2.2. 1 . 1 .2 Nonroad Equipment in the Nonroad Model
Nonroad equipment in the NONROAD model includes such sources as
recreational, construction, industrial, lawn and garden, farm, light commercial, logging,
airport service, railway maintenance, recreational marine vessels. For final rule
modeling, we used 1999 and future year inventories developed using NMIM2005, which
includes NONROAD2005. NONROAD2005 includes a number of improvements over
NONROAD 2004, which was used in the proposed rule. These improvements include
new evaporative categories for tank permeation, hose permeation, hot soak, and running
loss emissions, a revised methodology for calculating diurnal emissions, and
improvements to allocation of emissions from recreational marine and construction
equipment.
As with highway vehicles, exhaust gaseous hydrocarbons were estimated as
fractions of VOC, PAHs were calculated as fractions of PM, and metals were estimated
using basic emission factors. Evaporative emissions were estimated as fractions of VOC.
Changes in fuel benzene and aromatics levels are expected to result in similar
emission changes for nonroad gasoline equipment as for gasoline highway vehicles.
However, NMIM does not have the capability to model impacts of these fuel changes on
nonroad equipment emissions. Thus, we assumed that changes in county-level exhaust
and evaporative emissions of nonroad gasoline equipment were proportional to changes
in highway light-duty gasoline vehicle emissions.
PF nonroad exhaust20XX = — (3)
h,LDGVExhaustNMIMReference20XX
EjLDGVevap ^nfr^fr t 120XX
PF nonroad evap20XX = (4)
The nonroad refueling associated with PFCs was subtracted from the nonroad
inventory prior to air quality modeling, and the inventory summaries presented in Section
2.2.1.2.1 include this subtraction.
2-31
-------
Final Regulatory Impact Analysis
2.2.1.1.3 Commercial Marine Vessels, Locomotives and Aircraft
These source sectors will not be impacted by the fuel benzene standards being
finalized in this rule. Final rule inventories are unchanged from those used to model the
proposal.
Emissions for these source sectors in 1999 were obtained from the 1999 National
Emissions Inventory, Final Version 3. Gaseous air toxic and PAH emissions for turbine
engine aircraft were estimated by applying toxic to VOC ratios obtained from detailed
characterization of turbine engine emissions. Since no emissions data were available for
piston engine aircraft, a speciation profile from a non-catalyst light-duty gasoline vehicle
was used as a surrogate. Metal emissions were not estimated for aircraft. No speciated
emissions data were available for commercial marine vessels. For diesel marine vessels,
profiles from heavy-duty diesel highway vehicles were used; for steamships, a profile for
stationary and industrial boilers was used. Locomotive air toxic emissions were
estimated using speciation data from a year 2000 study done by the California Air
Resources Board.33 More detailed information on methods used to develop air toxic
inventories for these sectors can be found in the documentation for the 1999 NEI.34 This
documentation also describes methods used to develop inventories for 1990 and 1996.
The following approaches were used to project emissions for these source
categories:
Locomotives and commercial marine vessels - For gaseous HAPs, inventories
were developed by applying ratios of future year to 1999 national level 50 state VOC
inventory estimates (from the recent Clean Air Nonroad Diesel rule) by SCC code. For
polycyclic aromatic hydrocarbons, PM ratios were used. Metal inventory estimates were
projected to future years based on activity. Locomotive activity was projected using fuel
consumption data from the Energy Information Administration, as discussed in the
Regulatory Impact Analysis for Clean Air Nonroad Diesel Rule. For commercial marine
vessels, projected equipment populations from 1998 Power Systems Research (PSR) data
were used to develop factors. The future year inventories do not account for potential
reductions of additional locomotive or commercial marine vessel emission controls
currently under consideration.
Aircraft - To project emissions from aircraft and from aviation gas distribution
emissions, we developed and applied growth factors (in EMS-HAP) to 1999 emissions
based on landing and take off data. The Federal Aviation Administration's Terminal
Area Forecast System provided landing and take off data for future years up to 2020,
associated with commercial aircraft, general aviation, air taxi and military aircraft.35
These four categories map directly to the inventory categories for aircraft emissions. The
landing and take off data were summed across airports to create growth factors at the
national level. The general aviation growth factors were used for aviation gas
distribution emissions. After 2020, activity was assumed to increase at the same rate as
the increase from 2015 to 2020.
2-32
-------
Final Regulatory Impact Analysis
2.2.1.1.4. Portable Fuel Containers
Any MSATs contained in the liquid gasoline will be present as a component of
the VOCs. Specifically, the VOC emissions (estimated in Sections 2.1.1.2 and 2.1.3) will
contain the following eight MSATs:
benzene,
MTBE,
n-hexane,
toluene,
xylenes,
ethylbenzene,
naphthalene, and
2,2,4-trimethylpentane.
While MSAT inventories for portable fuel containers (PFCs) were developed at
the State level (benzene) or national level (other MSATs above) to estimate emission
benefits and cost-effectiveness of the proposed rule, county-level inventories for portable
fuel containers were not developed for use in air quality modeling for the proposal. In
this section, we describe the methodology used to develop such inventories for 1999,
2010, 2015, 2020, and 2030, for the final rule.
As discussed in Section 2.1.1., VOC inventories were developed at the State level for the
following years - 1990, 2005, 2010, 2015, 2020, and 2030. Thus an inventory had to be
developed for 1999. This was done by linear interpolation of the 1990 and 2005
inventories. Permeation and evaporative emissions had to be separated as well, since
they were combined in the State-level VOC inventories. Based on analyses done by the
California Air Resources Board, 33.87 percent of combined permeation plus evaporative
emissions was assumed to be permeation (see Section 2.1.1.2). This percentage was
applied to all the State inventory estimates.
Statewide total annual VOC inventories were allocated to counties using county
level fuel consumption ratios for calendar year 2002, obtained from the public release
version of NONROAD2005:
PFC VOC Emissions Emsswn Type ^ scc YYY^ County z =
PFC VOC Emissions x County Fml ConsumPtion^nroad^ (5)
.r.rU VV^. ^ml^lon^ X
State Fuel ConsumptionNonroad
2005
For all compounds except benzene and naphthalene, the fraction of total PFC emissions
that is composed of each of those HAPs was assumed to be directly proportional to the ratio of
each of those HAPs at the county level in total evaporative emissions from light-duty gasoline
vehicles (Equation 8).
2-33
-------
Final Regulatory Impact Analysis
MSAT Emissions LDGveVap7riYY
PFCMSATevap7nYY = - -^^ x PFCVOC Emissions (6)
riOXX VOC EmiSSiOnS LDGVEvap20xx
These ratios were obtained from the database of toxic to VOC ratios in the NMIM model,
discussed in previous sections. NMIM has ratios that vary by fuel type (conventional or baseline
gasoline, ethanol oxygenated gasoline, and MTBE oxygenated gasoline).
Another approach was used to estimate emissions of benzene with and without
PFC control, and also with and without the fuel benzene standard. First, we divided
county-level benzene refueling emissions by county-level VOC refueling emissions
estimated by NMIM, for both reference and control case scenarios. The resultant ratios
were multiplied by VOC emissions from evaporation, vapor displacement, and spillage.
These ratios were then adjusted based on a recent analysis of average nationwide
percentage of benzene in gasoline vapor from gasoline distribution with an RVP of 10 psi
at 60 degrees Fahrenheit.36 That analysis shows that the percentage of benzene in
gasoline vapor from gasoline distribution is 0.27%, in contrast to 0.74% benzene on
average nationwide in vehicle refueling emissions from highway vehicles. The reason
for this difference is that the refueling algorithm in MOBILE6.2 is based on a
temperature of 90 degrees, whereas temperatures for gasoline marketing emissions will
typically be lower. Thus a ratio of 0.36 was applied to the gasoline vehicle refueling
ratios. For all emission types except permeation, the equation used was:
PFC Benzene Emissions Emission Type ^ scc YYY^ Comfy z =
PFC VOC Emissions Emisswn Type x^ scc 777> Comty z x (7)
Re fueling Benzene LDGVi Comty z
V.Q fueling VOCLDG^Comty
xO.36
ity Z
A separate ratio was used for permeation emissions since recent research suggests
that the ratio of benzene from permeation is higher than for evaporation, vapor
displacement or spillage. A recent study37 suggests that the ratio of benzene from
permeation to total VOC from permeation is about 1.7727 times higher than the ratio
associated with evaporation. Thus, we multiplied the benzene refueling ratios for each
state by 1.7727 to obtain the benzene to VOC ratios for permeation:
PFC Benzene Emissions Emjssjon Type x, scc YYY , county z =
PFC VOC Emissions Emjsslon Type x, scc YYY , county z x (8)
(Re fueling Benzene WG^County z]
1 —: — x 0.36 x 1.77
Re fueling VOC
LDOV , County Z
A similar adjustment was applied to naphthalene emissions with and without fuel
benzene control, based on a recent analysis of average nationwide percentage of
naphthalene in gasoline vapor from gasoline distribution with an RVP of 10 psi at 60
2-34
-------
Final Regulatory Impact Analysis
degrees Fahrenheit.38'39 The percentage is 0.00027, in contrast to 0.05% naphthalene on
average nationwide in vehicle refueling emissions from highway vehicles. Thus a ratio
of 0.0054 was applied to the gasoline vehicle refueling ratios:
PFC Naphthalen e Emissions Emisslon Type x^ scc YYY^ County z =
PFC VOC Emissions Emssion Type x> scc YYY^ County z x (9)
(Refueling Naphthalen e LDGV ^ County z^\ ^ ^^
{ Re fueling VOCLDGVtCountyZ }
2.2.1.1.5. Gasoline Distribution
EPA's estimates of gasoline distribution emissions reflect improvements in its
methodology developed for the 2002 National Emissions Inventory (NEI). The key changes are:
1) Vehicle refueling emissions are estimated as part of the highway vehicle inventory using
NMTM2004, as discussed previously, and included in the highway vehicle inventory. Details
of how the modeling was done can be found in the documentation for the mobile source 2002
NEI.40 The previous methodology is described in the nonpoint 1999 NEI documentation.41
In this older method, national VOC emissions were calculated using fuel sales data and
estimates of emissions per fuel volume in areas with and without Stage 2 vapor recovery
systems. Air toxic emissions were estimated from VOC by applying speciation profiles for
different fuel types, such as baseline gasoline, MTBE oxygenated gasoline, and ethanol
oxygenated gasoline. Total emissions for each combination of vapor recovery system and
fuel type were allocated to individual counties using vehicle miles traveled.
2) For all other source categories in the gasoline distribution sector, EPA is using an improved
set of methods. These improvements include: (a) for source categories where activity-based
emission factors were available (all except bulk terminals and pipelines), EPA established
methods that maintain mass balance for storage and transfer activities, such that there is
agreement with the activity estimates used for each of the different distribution sectors; (b)
EPA developed criteria pollutant and air toxic emission estimates using the same activity
data and a consistent set of speciation profiles; and (c) EPA accounted for local differences in
fuel properties for downstream emissions (e.g. bulk plants, transit, unloading, storage, Stage
1 evaporative losses). More details on these improvements can be found in a technical
memorandum on the website for the 2002 NEI.42
The results of these changes were a significant increase in the air toxic inventory
estimates for vehicle refueling and a small increase nationwide for other sources of gasoline
distribution emissions. County-level estimates for some gasoline distribution sources changed
considerably since local differences in fuel properties were accounted for. Table 2.2.-4 compares
benzene estimates in the 1999 NEI, final version 3, and the final 2002 NEI.
2-35
-------
Final Regulatory Impact Analysis
Table 2.2.-4. Vehicle Refueling and Gasoline Distribution Benzene Emissions (Tons), 1999
and 2002 NEI.
Vehicle Refueling
Gasoline Distribution
1999 NEI
1558
4978
2002 NEI
2129
5119
% Difference
+36
+3
In order to develop better estimates of the emission benefits of the fuel benzene control being
finalized in this rule, EPA developed updated air toxic inventories for vehicle refueling and
gasoline distribution to reflect the changes made in the 2002 NEI. In addition, the same
adjustment factors for benzene and naphthalene described above for PFC emissions were also
applied to gasoline distribution emissions.
Inventories were developed as follows:
1) Vehicle refueling emissions were estimated using NMIM2004. Refueling emissions were
estimated for reference case inventories in 1999, 2010, 2015, 2020 and 2030. Control case
inventories were estimated for 2015, 2020 and 2030.
2) For other gasoline distribution emissions, for each air toxic pollutant, EPA estimated a
national-scale adjustment factor as follows:
Adjustment factor = 2002 NEI national emissions/2002 national emissions estimated from
interpolation of the 1999 NEI and a 2007 projection for the proposed rule.
3) EPA developed new county-level reference case inventories for these pollutants by applying
these adjustment factors to county-level gasoline distribution emissions for 1999 and future
years. The gasoline distribution projections were based on projection information (growth
factors, closures, reductions, etc.) from the 1999 NEI.43 Revised inventories were
developed for years 1999, 2015, and 2020. 2030 was assumed to be the same as 2020.
4) Additional nationwide adjustments of 0.36 and 0.0054, respectively, were applied to
emissions of benzene and naphthalene. The basis for these adjustments is discussed in the
Section 2.2.1.1.4.
5) EPA developed new control case inventories for gasoline distribution, for benzene, for years
2015, 2020, and 2030. These revised county-level inventories were estimated by applying
the following ratios:
emissions proposed rule control scenario/emissions proposed rule reference case
These ratios reflect reductions estimated based on the assumption that reductions
are proportional to reductions in vehicle refueling emissions.
2.2.1.1.6. Other Stationary Sources
2-36
-------
Final Regulatory Impact Analysis
Stationary source estimates for 1999, for all source categories except gasoline
distribution, were obtained from the National Emissions Inventory.44' 45
For nearly all stationary sources (point and non-point source inventories), we used
the Emissions Modeling System for Hazardous Air Pollutants (EMS-HAP), Version 3.0
to apply growth and control factors to the 1999 NEI, source type by source type.46 EMS-
HAP has the capability of projecting emissions to 2020. After 2020, stationary source
emissions were assumed to remain constant.
The general methodology for projecting stationary source emissions using EMS-
HAP is as follows:
Future Year Emissions = Base Year Emissions * Growth Factor * (100% - % Reduction)/100 (10)
The actual equations used by EMS-HAP also allow the application of a "new
source" reduction to a fraction of the emissions to allow for a different level of emission
reduction to be applied to a portion of the emissions. In addition, if the source is already
controlled, and the value of the overall control efficiency is provided in the emission
inventory, EMS-HAP adjusts the percent reduction (% Reduction) based on the overall
control efficiency value provided in the inventory. The actual projection equations are
provided in Chapter 6 (PtGrowCntl) of the EMS-HAP User's Guide (U. S. EPA, 2004b,
pp. 6-15-6-17).
Stationary source growth — EMS-HAP allows growth factors to be applied to the
inventory on either a national, state or county level basis, based on one of the following
inventory codes that describe the source: (1) MACT, which identifies an emission source
as a belonging to a particular regulatory category or subcategory; (2) Standard Industrial
Classification (SIC), which classifies establishments by their primary type of activity, as
defined by the U.S. Census Bureau; (3) Source Category Code (SCC), which defines the
source using EPA's coding system for the NEI. The MACT and SCC code definitions
are contained in the code tables supplied with the NEI. Note that even though the code is
called "MACT", it is also used for other regulations besides MACT such as section 129
rules. The hierarchy built into EMS-HAP is to use a MACT-based growth factor first,
followed by an SIC-based and lastly, an SCC-based growth factor. The most detailed
geographic level is used first (e.g., a state-specific growth factor replaces a national
growth factor). EMS-HAP does not have the capability to apply growth factors to
specific point source facilities, nor can they be applied differently for the different
pollutants for a particular source category.
For stationary sources, growth factors were developed using three primary sources of
information:
Regional Economic Models, Inc. (REMI) Policy Insight® model, version 5.5;47'48
Regional and National fuel-use forecast data from the Energy Information
Administration, U.S. Department of Energy, Annual Energy Outlook (AEO)49
2-37
-------
Final Regulatory Impact Analysis
Rule development leads or economists who had obtained economic information in
the process of rule development.
The first two sources of information were also used in projecting criteria pollutant
emissions for EPA's 2005 Clean Air Interstate Rule.50
More details on how these sources were used can be found in the EPA technical
report, "National Scale Modeling of Mobile Source Air Toxic Emissions, Air Quality,
Exposure and Risk for the Mobile Source Air Toxics Rule," cited previously.
Stationary source reductions -- Emission reductions were applied to the grown
emissions to account for regulatory efforts which are expected to reduce HAPs from 1999
levels. The percent reductions we determined were primarily based on estimates of
national average reductions for specific HAPs or for groups of HAPs from a source
category or subcategory as a result of regulatory efforts. These efforts are primarily the
MACT and section 129 standards, mandated in Title III of the 1990 Clean Air Act
Amendments. We determined percent reductions, and whether they apply to major only
or both major and area sources, for the various rules from rule preambles, fact sheets and
through the project leads (questionnaire and phone calls). A major source is defined as
any stationary source or group of stationary sources located within a contiguous area and
under common control that has the potential to emit considering controls, in the
aggregate, 10 tons per year or more of any hazardous air pollutant or 25 tons per year or
more of any combination of hazardous air pollutants. For some rules, percent reductions
were provided for specific HAPs or groups of HAPs (e.g., all metals, or all volatiles)
rather than a single number for all HAPs in the categories. After 2010, stationary source
emissions are based only on economic growth. They do not account for reductions from
ongoing toxics programs such as the urban air toxics program, residual risk standards and
area source program, which are expected to further reduce toxics.
2.2.1.1.7 Precursor Emissions
In addition to the air toxics in Table 2.2.-1, emissions of a number of other
compounds were estimated because they are precursor emissions which are
atmospherically transformed into air toxics. These pollutants are listed in Table 2.2.-5,
along with air toxic pollutants included in the inventory which can be transformed into
other air toxics. Precursor emissions in 1999 were estimated by applying speciation
profiles from SPECIATE to VOC estimates from version 2 of the 1999 NEI.51 Stationary
source precursor emissions were assumed to remain at 1999 levels in future year
modeling since the impact of growth and control is unknown. However, mobile source
precursor emissions are expected to increase along with VOCs. To account for this in
modeling done to support the final rule, we estimated secondary concentrations from
mobile sources in future years by assuming they increased proportionally with primary
concentrations. For the proposed rule, we had projected precursor emissions for 1999 to
future years using ratios of VOCs for future years versus 1999, then used these projected
emissions to model secondary concentrations. A comparison of the two approaches,
2-38
-------
Final Regulatory Impact Analysis
using modeling data from the proposal, yielded very similar results. A more detailed
discussion of the comparison can be found in EPA Technical Report Number EPA-
454/R-07-002
2.2.1.1.8 Strengths and Limitations
Highway Vehicles - Limitations in the VOC and PM emission estimates which
are the basis for calculating air toxic emissions are discussed in Section 2.1.3.
MOBILE6.2 toxic to VOC ratios for key toxics from gasoline vehicles, such as benzene,
1,3-butadiene, formaldehyde and acetaldehyde, are based on almost 900 vehicle tests on a
wide variety of fuels. These data account for impacts of emissions control technology,
normal vs. high emitters, and impacts of a variety of fuel properties, including benzene.
level, aromatics levels, olefin level, sulfur level, RVP, E200, E300, and oxygenate
content.
However, there are a number of significant uncertainties in our highway vehicle
air toxic inventories for air quality modeling. Among the uncertainties are:
• The Agency has limited emissions data for advanced technology highway
vehicles, including hybrid and alternative technology vehicles. The toxic to VOC
ratios in MOBILE6.2 are all based on Tier 0 and earlier vehicles. EPA has
recently evaluated data on more recent technology vehicles and what might be the
potential impacts of these data on inventories. The result of this analysis is
discussed in Section 2.3.1.
• MOBILE6.2 uses the same toxic to VOC ratios for cold starts and hot running
operation even though these ratios for benzene and 1,3-butadiene are higher
during cold starts than hot running. We have a limited understanding of the
impact of off-cycle operation on highway vehicle air toxic emissions.
• Data are limited for certain sources and pollutants not significant to this rule. For
heavy-duty highway vehicles (both gasoline and diesel engines) the toxic to VOC
ratios used in MOBILE6.2 to develop inventory estimates are based on very
limited data. Moreover, we lack data on how diesel fuel properties impact air
toxic emissions, and we have very little data on mobile source metal emissions.
There are also significant uncertainties resulting from the use of national default
data rather than local inputs, as well as "top-down" allocation schemes in estimating toxic
emissions. Examples include use of national default vehicle registration distributions,
default average speed distributions, and use of county level population data to allocate
State or urban level VMT. A recent paper evaluated the impacts of these default inputs
and allocation schemes on local level inventories.52
Finally, as discussed in Section 2.1.3, there are greater uncertainties in projection
year estimates.
2-39
-------
Final Regulatory Impact Analysis
Table 2.2.-S. Precursor Pollutants.
Pollutant
Acetaldehyde
1,3-Butadiene
1-Butene
1-2,3-Dimethyl butene
1-2-Ethylbutene
1-2-Methyl butene
1-3-Methyl butene
2-Butene
2-2-Methyl butene
1-Decene
Ethanol
Ethene
1-Heptene
2-Heptene
1-Hexene
2-Hexene
3-Hexene
Precursor for
Formaldehyde (reactive and
inert)
Formaldehyde (reactive and
inert), Acrolein (reactive
and inert)
Formaldehyde (reactive and
inert), Propionaldehyde
(reactive and inert)
Formaldehyde (reactive and
inert)
Formaldehyde (reactive and
inert)
Formaldehyde (reactive and
inert)
Formaldehyde (reactive and
inert)
Acetaldehyde (reactive and
inert)
Acetaldehyde (reactive and
inert)
Formaldehyde (reactive and
inert)
Acetaldehyde (reactive and
inert)
Formaldehyde (reactive and
inert)
Formaldehyde (reactive and
inert)
Acetaldehyde (reactive and
inert)
Formaldehyde (reactive and
inert)
Acetaldehyde (reactive and
inert)
Propionaldehyde (reactive
and inert)
Pollutant
Isoprene
MTBE
Methanol
1-Nonene
2-Nonene
1-Octene
2-Octene
1-Pentene
1-2,4,4-Trimethyl pentene
1-2-Methyl pentene
1-3-Methyl pentene
1-4-Methyl pentene
2-Pentene
2-3-Methyl pentene
2-4-Methyl pentene
Propene
2-Methylpropene
Precursor for
Formaldehyde (reactive and
inert)
Formaldehyde (reactive and
inert)
Formaldehyde (reactive and
inert)
Formaldehyde (reactive and
inert)
Acetaldehyde (reactive and
inert)
Formaldehyde (reactive and
inert)
Acetaldehyde (reactive and
inert)
Formaldehyde (reactive and
inert)
Formaldehyde (reactive and
inert)
Formaldehyde (reactive and
inert)
Formaldehyde (reactive and
inert)
Formaldehyde (reactive and
inert)
Acetaldehyde (reactive and
inert),
Propionaldehyde (reactive
and inert)
Acetaldehyde (reactive and
inert)
Acetaldehyde (reactive and
inert)
Acetaldehyde (reactive),
Acetaldehyde (inert),
Formaldehyde (reactive and
inert)
Formaldehyde (reactive and
inert)
2-40
-------
Final Regulatory Impact Analysis
NonroadEquipment - The toxic to VOC ratios in NMIM for lawn and garden
equipment, which makes the single largest contribution of any nonroad sector to the air
toxics inventory, is supported by a large amount of test data. The VOC estimates for
uncontrolled engines in the NONROAD model are based on a large amount of in-use test
data and peer reviewed methodologies. Estimates for controlled engines are based on
certification test data and emission standards. However, for a number of source
categories—in particular heavy-duty diesel engines and aircraft engines—the toxic to
VOC ratios used to develop inventory estimates are based on very limited data. In
addition, the Agency has limited emissions data for nonroad equipment with emission
controls. The Agency has been doing test data to address some of the limitations. This
work is discussed in Sections 2.3.3 and 2.3.4. There are also significant uncertainties
associated with allocating nonroad equipment emissions from the national to the local
level. As with highway sources, future year inventories are more uncertain. Finally, the
relationship between fuel parameters and emission rates for gasoline nonroad equipment
is much more poorly understood than the relationship for highway gasoline vehicles. In
our modeling, we assumed that the impacts of fuel control on emissions from nonroad
equipment would be proportional to the impact on highway vehicle emissions, as
discussed above.
Portable Fuel Containers -- Since no direct measurements of air toxic emissions
from evaporation of gasoline in portable fuel containers were available, they were
estimated based on toxic to VOC ratios obtained from evaporative emissions
measurements taken from light-duty gasoline vehicles. However, since evaporation of
fuel occurs at higher temperatures in vehicles than in PFCs, speciation profiles are
different. An effort to account for these differences was made for benzene and
naphthalene based on recent analyses done for the gasoline distribution sector.
Stationary Sources — For the 1999 NEI, there are a number of known or
suspected issues for stationary source emissions listed on the emission inventory website
(U. S. EPA, 2004a). The issues listed are generally limited to specific geographic areas
and are not expected to influence national-level results. Of these, it is expected that
issues related to acrolein are most likely to affect the results for assessment of noncancer
effects. Another uncertainty concerning the base year inventory is the proper
identification of sources using the inventory codes. These codes are utilized for applying
growth and reduction factors.
There are several uncertainties associated with the growth and reduction
information. The growth information is uncertain for a number of reasons. For most
sources, activity growth is used as a surrogate for emissions growth, which may not be
appropriate for some industry sectors. In addition the growth information available is
from economic models, is typically specific to broad industry categories, and is not
resolved geographically for all categories. The stationary source reductions are uncertain
because they are generally based on national-average reductions (although we have used
facility-specific reductions where available). We do not expect this uncertainty to have
an impact on national-level results.
2-41
-------
Final Regulatory Impact Analysis
As previously mentioned, after 2010, stationary source emissions are based only
on economic growth. They do not account for reductions from ongoing toxics programs
such as the urban air toxics program, residual risk standards and area source program,
which are expected to further reduce toxics. Furthermore, the 2030 stationary source
inventory estimates are equal to the 2020 estimates, because of additional uncertainties in
the available growth data past 2020 and the lack of knowledge of the effect of stationary
source control programs that far into the future.
2.2.1.2 Trends in Air Toxic Emissions
2.2.1.2.1 Emission Trends Without Controls
In 1999, based on the National Emissions Inventory (NEI), mobile sources
accounted for 44% of total emissions of 188 hazardous air pollutants (see Figure 2.2.-2).
Diesel particulate matter is not included in this list of 188 pollutants. Sixty-five percent
of the mobile source tons in this inventory were attributable to highway mobile sources,
and the remainder to nonroad sources. Furthermore, over 90% of mobile source air toxic
emissions are attributable to gasoline vehicles and equipment
Overall, emissions from all air toxics are projected to decrease from 5,030,000
tons in 1999 to 4,010,000 tons in 2020, as a result of existing and planned emission
controls on major, area, and mobile sources. In the absence of Clean Air Act emission
controls currently in place, EPA estimates air toxic emissions would total 11,590,000
tons in 2020 (Figure 2.2-2). It should be noted that these estimates do not account for
higher estimates of cold temperature hydrocarbon emissions in vehicles, PFC emissions,
or categories of nonroad gasoline evaporative emissions included in NONROAD2005
and discussed in Section 2.2.1.1.2.
If higher estimates of cold temperature hydrocarbon emissions and vehicles and
evaporative emissions from nonroad gasoline equipment are accounted for, air toxic
emissions emitted from mobile sources will be reduced 46% between 1999 and 2020
without the controls in this proposal, from 2.38 million to 1.29 million tons (Figure 2.2-
3). This reduction will occur despite a projected 57% increase in vehicle miles traveled,
and a 47% projected increase in nonroad activity (See Figures 2.2.-4 and 2.2.-5). It
should be noted, however, that EPA anticipates mobile source air toxic emissions will
begin to increase after 2020, from about 1.29 million tons in 2020 to 1.42 million tons in
2030. Benzene emissions from all sources decrease from about 366,000 tons in 1999 to
279,000 tons in 2020, and as is the case with total air toxic emissions, begin to increase
between 2020 and 2030 (Figure 2.2.-5).
2-42
-------
Final Regulatory Impact Analysis
Figure 2.2.-2. Contribution of Source Categories to Air Toxic Emissions, 1990 to
2020 (not Including Diesel Particulate Matter). Dashed Line Represents Projected
Emissions without Clean Air Act Controls. Does not Account for Higher Estimates
of Cold Temperature Hydrocarbon Emissions in Vehicles, PFC Emissions, or
Categories of Nonroad Gasoline Evaporative Emissions Included in
NONROAD2005.
t~
o
tf>
c
o
1
w
0
w
E
LU
12
11
10
9
8
7
6
5
4
3
2
1
U.S. Contributions of Source Categories to Total Emissions for all HAPs
I U Major
I | Area and Other
HI Fires - Prescribed and Wild
I I Non-Road Mobile
| | On-Road Mobile
^
r^
f^r
*v^
7.X "!
"
2.69
0.91
0.34
0.75
2.55
5.03
1.26
0.29
0.76
1.44
^
^9V92 ' ^-\
,,-
4.09
0.95
1.45
0.29
0.65
0.77
,--'
-^
3.9
0.9
1 51
0.29
058
0.63
s
ci
«
^'
CO
*"
CO
CO
^,J
3.88
0.97
1.64
0.29
049
0.49
4.01
1.06
1.78
0.29
0.44
0.44
'-""'
T-
00
^
s
-
9^1
-
-^
5
^
1990
1999
2007
2010
2015
2020
2-43
-------
Final Regulatory Impact Analysis
Figure 2.2.-S. Contribution of Source Categories to Mobile Source Air Toxic
Emissions, 1999 to 2030 (Not Including Diesel Particulate Matter). Includes Higher
Estimates of Cold Temperature Hydrocarbon Emissions and Vehicles, Evaporative
Emisions from Nonroad Gasoline Equipment, and PFC Emissions as Part of Area
Source Inventory.
4,000,000 i
3,500,000
3,000,000
2,500,000
2,000,000
1,500,000
1,000,000
500,000
• Nonroad Equipment
D Highway Vehicles
D Fires — Prescribed and Wild
• Area and Other
• Major
1999
2030
2-44
-------
Final Regulatory Impact Analysis
Figure 2.2.-4. Trend in Highway Vehicle Air Toxic Emissions Versus VMT, 1999 to
2030.
Figure 2.2.-S. Trend in Emissions of Nonroad Equipment Air Toxic Emissions
(Excluding Commercial Marine Vessels, Locomotives and Aircraft) versus Activity,
1999 to 2030.
§ 500000 - -
2010 2015
Year
3E+11 Ł ^TonsMSATs
.Ł• —•—Nonroad Equipment Activity (hp-hrs)
2020 2030
2-45
-------
Final Regulatory Impact Analysis
Figure 2.2.-6. Trend in Benzene Emissions, 1999 to 2030.
400,000
350,000
300,000
250,000
200,000
150,000
100,000
50,000
• Nonroad Equipment
D Highway Vehicles
D Fires — Prescribed and Wild
• Area and Other
D Major
1999
2030
Highway Vehicle Trends - Table 2.2.-6 summarizes nationwide emissions of
individual air toxics from highway vehicles from 1999 to 2030. Fifteen POM compounds
listed in Table 2.2.-1 (except for naphthalene) are grouped together as POM. For mobile
sources, forty percent of the chromium from highway vehicles and eighteen percent of
the chromium from nonroad sources was assumed to be the highly toxic hexavalent form.
The estimate for highway vehicles is based on data from utility boilers,53 and the estimate
for nonroad equipment is based on combustion data from stationary combustion turbines
that burn diesel fuel.54
2-46
-------
Final Regulatory Impact Analysis
Table 2.2.-6. Nationwide Emissions (Tons) of Individual Air Toxic Pollutants from
Highway Vehicles.
Pollutant
1,3 -Butadiene
2,2,4-Trimethylpentane
Acetaldehyde
Acrolein
Benzene
Chromium III
Chromium VI
Ethyl Benzene
Formaldehyde
Hexane
MTBE
Manganese
Naphthalene
Nickel
POM
Propionaldehyde
Styrene
Toluene
Xylenes
1999
23,876
182,120
29,821
3,845
183,661
8
5
73,439
80,458
66,267
57,801
5
4,056
10
497
4,288
14,284
489,873
277,285
2010
11,473
101,880
17,169
1,824
110,526
10
7
40,732
38,885
39,801
29,886
6
2,261
13
255
2,327
7,652
268,871
152,046
2015
10,763
94,469
16,149
1,650
105,956
11
8
37,528
35,857
33,481
23,089
6
2,022
14
234
2,154
7,368
250,646
141,710
2020
11,355
96,315
16,893
1,704
110,317
12
8
38,080
37,153
30,727
18,372
7
1,986
16
239
2,222
7,814
257,367
145,473
2030
13,378
111,783
19,879
1,981
129,290
15
10
44,055
43,404
33,468
17,957
9
2,259
19
278
2,574
9,253
299,677
169,369
Table 2.2.-7 summarizes total tons of air toxic emissions from highway vehicles
by vehicle class in 1999, 2007, 2010, 2015, 2020, and 2030. Table 2.2.-S provides the
percentage of total highway vehicle emissions associated with each vehicle class. In
1999, 55% of air toxic emissions from highway vehicles were emitted by light-duty
gasoline vehicles (LDGVs) and 37% by light-duty trucks (LDGTs). EPA projects that in
2020, only 34% of highway vehicle HAP emissions will be from LDGVs and 60% will
be from LDGTs. More detailed summaries of emissions by individual pollutant, by State,
and for urban versus rural area can be found in Excel workbooks included in the docket
for this rule.
2-47
-------
Final Regulatory Impact Analysis
Table 2.2.-7. Tons of Air Toxic Emissions from Highway Vehicle Classes, 1999 to
2030 (Not Including Diesel Particulate Matter).
Vehicle Type
HDDV
HDGV
LDDT
LDDV
LDGT1
LDGT2
LDGV
MC
Total Highway
Emissions (tons/yr)
1999
36,958
66,672
1,215
688
353,671
188,134
836,995
7,267
1,491,600
2010
22,622
21,323
589
41
279,674
144,254
349,220
7,899
825,624
2015
19,605
14,812
528
23
287,644
141,165
290,746
8,595
763,117
2020
19,469
11,638
470
16
319,974
144,247
270,956
9,291
776,062
2030
22,172
10,188
389
16
375,603
159,682
319,395
11,213
898,659
HDDV: Heavy Duty Diesel Vehicles
HDGV: Heavy Duty Gasoline Vehicles
LDDT: Light Duty Diesel Trucks
LDDV: Light Duty Diesel Vehicles
LDGT1 : Light Duty Gasoline Trucks 1
LDGT2: Light Duty Gasoline Trucks 2
LDGV: Light Duty Gasoline Vehicles
MC: Motorcycles
Table 2.2.-S. Percent Contribution of Vehicle Classes to Highway Vehicle Air Toxic
Emissions, 1999 to 2030 (Not Including Diesel Particulate Matter).
Vehicle
LDGV
LDGT1 and 2
HDGV
HDDV
Other (motorcycles and
light-duty diesel
vehicles and trucks)
1999
56%
36%
5%
2%
1%
2010
42%
51%
3%
3%
1%
2015
38%
56%
2%
2%
2%
2020
35%
60%
1%
2%
2%
2030
35%
60%
1%
2%
2%
Tables 2.2.-9 through 2.2.-14 summarize total tons of emissions nationwide for
benzene, 1,3-butadiene, formaldehyde, acetaldehyde, naphthalene, and acrolein from
highway vehicles. About 90% of benzene emissions from gasoline vehicles were in
exhaust, with the remainder in evaporative and refueling emissions. Benzene emissions
from diesel vehicles were all exhaust. There are no evaporative emissions of 1,3-
butadiene, formaldehyde, acetaldehyde, and acrolein.
2-48
-------
Final Regulatory Impact Analysis
Table 2.2.-9. Tons of Benzene Emissions from Highway Vehicle Classes, 1999 to
2030.
Vehicle Type
HDDV
HDGV
LDDT
LDDV
LDGT1
LDGT2
LDGV
MC
Total Highway
Emissions (tons/yr)
1999
2,564
6,665
200
112
46,358
21,392
105,724
646
183,661
2010
1,574
2,383
97
7
39,456
19,742
46,598
669
110,526
2015
1,366
1,715
87
4
41,796
20,074
40,186
728
105,956
2020
1,358
1,399
78
3
47,352
21,083
38,257
787
110,317
2030
1,547
1,213
64
3
56,290
23,737
45,489
947
129,290
Table 2.2.-10. Tons of 1,3-Butadiene Emissions from Highway Vehicle Classes, 1999
to 2030.
Vehicle Type
HDDV
HDGV
LDDT
LDDV
LDGT1
LDGT2
LDGV
MC
Total Highway
Emissions (tons/yr)
1999
1,489
1,177
90
50
5,307
3,526
12,034
202
23,876
2010
915
197
44
3
3,820
1,991
4,280
224
11,473
2015
794
99
39
2
3,929
1,913
3,743
243
10,763
2020
789
78
35
1
4,520
2,064
3,605
263
11,355
2030
899
63
29
1
5,411
2,344
4,312
318
13,378
2-49
-------
Final Regulatory Impact Analysis
Table 2.2.-11. Tons of Formaldehyde Emissions from Highway Vehicle Classes,
1999 to 2030.
Vehicle Type
HDDV
HDGV
LDDT
LDDV
LDGT1
LDGT2
LDGV
MC
Total Highway
Emissions (tons/yr)
1999
19,094
6,142
386
217
15,666
9,916
28,522
516
80,458
2010
1 1 ,724
1,213
188
13
9,702
4,851
10,627
567
38,885
2015
10,176
688
168
7
10,030
4,656
9,515
617
35,857
2020
10,114
556
150
5
11,487
4,961
9,213
667
37,153
2030
1 1 ,522
460
124
5
13,790
5,652
1 1 ,044
806
43,404
Table 2.2.-12. Tons of Acetaldehyde Emissions from Highway Vehicle Classes, 1999
to 2030.
Vehicle Type
HDDV
HDGV
LDDT
LDDV
LDGT1
LDGT2
LDGV
MC
Total Highway
Emissions (tons/yr)
1999
7,032
1,411
123
69
6,050
3,429
11,555
152
29,821
2010
4,318
390
60
4
4,808
2,367
5,043
180
17,169
2015
3,748
248
54
2
5,068
2,329
4,504
196
16,149
2020
3,725
204
48
2
5,836
2,502
4,364
213
16,893
2030
4,243
173
40
2
7,039
2,880
5,246
258
19,879
2-50
-------
Final Regulatory Impact Analysis
Table 2.2.-13. Tons of Acrolein Emissions from Highway Vehicle Classes, 1999 to
2030.
Vehicle Type
HDDV
HDGV
LDDT
LDDV
LDGT1
LDGT2
LDGV
MC
Total
Highway
Emissions (tons/yr)
1999
855
689
35
20
623
326
1,286
13
3,845
2010
525
76
17
1
457
231
503
14
1,824
2015
455
24
15
1
472
226
442
15
1,650
2020
453
17
14
0
538
240
425
16
1,704
2030
516
12
11
0
644
271
508
20
1,981
Table 2.2.-14. Tons of Naphthalene Emissions from Highway Vehicle Classes, 1999
to 2030.
Vehicle Type
HDDV
HDGV
LDDT
LDDV
LDGT1
LDGT2
LDGV
MC
Total
Highway
Emissions (tons/yr)
1999
167
773
7
7
760
489
1,830
23
4,056
2010
65
400
2
0
640
267
861
25
2,261
2015
32
248
1
0
697
273
743
27
2,022
2020
19
195
1
0
769
281
693
29
1,986
2030
16
176
1
0
900
315
817
35
2,259
NonroadEquipment Trends — Table 2.2.-15 summarizes nationwide emissions of
individual air toxics from nonroad equipment, from 1999 to 2030. The lead emissions in
the table are from piston engine aircraft, which use leaded gasoline. Table 2.2.-16
summarizes total tons of air toxic emissions from categories of nonroad equipment by
equipment type in 1999, 2010, 2015, 2020, and 2030. Table 2.2.-17 provides the
percentage of total nonroad equipment emissions associated with each equipment type.
Air toxic emissions from nonroad equipment are dominated by lawn and garden
equipment, recreational equipment, and pleasure craft, which collectively account for
about 80% of nonroad HAP emissions in all years. More detailed summaries of
emissions by individual pollutant, by State, and for urban versus rural area can be found
in Excel workbooks included in the docket for this rule.
2-51
-------
Final Regulatory Impact Analysis
Table 2.2.-15. Nationwide Emissions of Individual Air Toxics from Nonroad
Equipment, from 1999 to 2030.
Pollutant
1,3-Butadiene
2,2,4-Trimethylpentane
Acetaldehyde
Acrolein
Benzene
Chromium III
Chromium VI
Ethyl Benzene
Formaldehyde
Hexane
Manganese
MTBE
Naphthalene
Nickel
POM
Propionaldehyde
Styrene
Toluene
Xylenes
Annual Total Nonroad Emissions (Tons)
1999
10,333
109,793
21,952
2,754
74,902
15
3
46,072
52,083
36,925
2
78,585
1,212
31
347
4,968
3,055
234,558
208,728
2010
7,136
83,546
16,208
2,264
54,763
15
4
33,435
38,213
29,758
2
28,464
1,182
34
305
3,462
2,297
189,605
147,242
2015
6,586
71,362
14,459
2,179
49,985
16
4
29,489
34,406
27,430
2
27,238
1,228
36
287
3,036
2,003
164,871
126,825
2020
6,518
62,991
13,663
2,168
48,453
16
4
27,057
32,678
26,083
2
27,245
1,291
37
275
2,824
1,807
146,220
114,252
2030
7,004
62,250
14,153
2,340
51,647
16
4
28,033
33,994
27,439
2
29,865
1,440
41
287
2,865
1,835
145,330
116,764
2-52
-------
Final Regulatory Impact Analysis
Table 2.2.-16. Tons of Air Toxic Emissions from Nonroad Equipment Types, 1999
to 2030 (Not Including Diesel Particulate Matter).
Equipment
Type
Agriculture
Aircraft
Airport Support
Commercial
Commercial Marine
Vessel
Construction
Industrial
Lawn/Garden
Logging
Pleasure Craft
Railroad
Recreational
Underground Mining
Total Nonroad
Annual Total Nonroad Emissions (Tons)
1999
21,397
14,276
325
59,302
8,736
42,496
1 1 ,422
261,635
3,578
332,631
4,412
125,933
177
886,318
2010
12,512
14,965
198
33,977
9,742
22,280
4,247
129,932
2,094
202,760
3,972
201,118
138
637,934
2015
9,686
16,081
157
35,994
10,213
18,688
2,793
130,157
2,228
163,953
3,886
167,488
114
561 ,439
2020
7,875
17,256
141
39,207
10,973
16,439
2,239
139,762
2,452
148,746
3,752
124,640
101
513,583
2030
6,567
19,603
152
46,503
13,354
15,207
2,093
160,669
2,960
147,720
3,533
106,845
102
525,309
Table 2.2.-17. Contribution of Equipment Types to Nonroad Air Toxic Emissions,
1999 to 2030 (not Including Diesel Particulate Matter).
Equipment
Type
Lawn and
Garden
Pleasure Craft
Recreational
All Others
1999
30%
38%
14%
18%
2010
20%
32%
32%
16%
2015
23%
29%
30%
18%
2020
27%
29%
24%
19%
2030
31%
28%
20%
21%
Over 90% of nonroad toxic emissions are from 2-stroke and 4-stroke gasoline
engines, with the remainder from diesel engines and turbine engine aircraft. Similarly,
over 90% of benzene emissions from nonroad equipment are from gasoline engines, and
these emissions would be reduced by a fuel benzene standard.
Tables 2.2.-18 through 2.2.-23 summarize total tons of emissions nationwide for
benzene, 1,3-butadiene, formaldehyde, acetaldehyde, naphthalene, and acrolein from
nonroad equipment types.
2-53
-------
Final Regulatory Impact Analysis
Table 2.2.-18. Tons of Benzene Emissions from Nonroad Equipment Types, 1999 to
2030.
Equipment Type
Agriculture
Aircraft
Airport Support
Commercial
Commercial Marine
Vessel
Construction
Industrial
Lawn/Garden
Logging
Pleasure Craft
Railroad
Recreational
Underground Mining
Total Nonroad
Annual Total Nonroad Emissions (Tons)
1999
2,105
1,102
33
7,931
644
3,945
1,498
25,753
202
24,963
162
6,548
15
74,902
2010
1,283
1,163
19
5,140
719
2,111
524
15,996
131
16,698
143
10,825
12
54,763
2015
1,020
1,247
15
5,478
753
1,786
335
15,540
130
14,101
139
9,430
10
49,985
2020
855
1,335
14
6,010
809
1,595
263
16,644
140
13,145
134
7,502
9
48,453
2030
736
1,511
15
7,178
982
1,494
233
19,133
168
13,264
126
6,798
9
51,647
Table 2.2.-19. Tons of 1,3-Butadiene Emissions from Nonroad Equipment Types,
1999 to 2030.
Equipment Type
Agriculture
Aircraft
Airport Support
Commercial
Commercial Marine Vessel
Construction
Industrial
Lawn/Garden
Logging
Pleasure Craft
Railroad
Recreational
Underground Mining
Total Nonroad
Annual Total Nonroad Emissions (Tons)
1999
236
824
4
1,324
6
455
242
4,034
35
2,069
114
990
1
10,333
2010
145
859
2
774
6
231
76
2,240
21
1,291
104
1,385
1
7,136
2015
116
924
2
820
e7
198
47
2,085
21
1,034
102
1,230
11
6,586
2020
98
993
2
901
6
180
37
2,225
23
928
99
1,025
1
6,518
2030
85
1,131
1,080
171
31
2,558
28
909
94
907
7,004
2-54
-------
Final Regulatory Impact Analysis
Table 2.2.-20. Tons of Formaldehyde Emissions from Nonroad Equipment Types,
1999 to 2030.
Equipment Type
Agriculture
Aircraft
Airport Support
Commercial
Commercial Marine Vessel
Construction
Industrial
Lawn/Garden
Logging
Pleasure Craft
Railroad
Recreational
Underground Mining
Total Nonroad
Annual Total Nonroad Emissions (Tons)
1999
8,890
6,549
123
3,516
4,715
12,103
2,487
7,050
334
2,345
1,895
1,990
87
52,083
2010
5,051
6,809
83
2,331
5,252
7,352
1,212
3,902
153
1,548
1,721
2,731
68
38,213
2015
3,759
7,333
66
2,122
5,499
5,662
837
3,633
117
1,274
1,683
2,365
56
34,406
2020
2,915
7,885
58
2,019
5,899
4,541
697
3,816
109
1,160
1,624
1,904
50
32,678
2030
2,296
8,990
63
2,080
7,152
3,858
718
4,328
116
1,147
1,527
1,669
50
33,994
Table 2.2.-21. Tons of Acetaldehyde Emissions from Nonroad Equipment Types,
1999 to 2030.
Equipment Type
Agriculture
Aircraft
Airport Support
Commercial
Commercial Marine Vessel
Construction
Industrial
Lawn/Garden
Logging
Pleasure Craft
Railroad
Recreational
Underground Mining
Total Nonroad
Annual Total Nonroad Emissions (Tons)
1999
3,986
2,019
55
1,390
2,364
5,433
1,087
2,381
133
1,615
850
599
39
21,952
2010
2,265
2,098
37
999
2,639
3,308
539
1,522
59
1,098
772
843
30
16,208
2015
1,685
2,259
30
902
2,768
2,550
372
1,410
41
920
755
743
25
14,459
2020
1,306
2,430
26
850
2,974
2,046
310
1,476
37
844
728
613
22
13,663
2030
1,028
2,770
28
866
3,619
1,739
320
1,670
37
834
685
533
23
14,153
2-55
-------
Final Regulatory Impact Analysis
Table 2.2.-22. Tons of Acrolein Emissions from Nonroad Equipment Types, 1999 to
2030.
Equipment Type
Agriculture
Aircraft
Airport Support
Commercial
Commercial Marine Vessel
Construction
Industrial
Lawn/Garden
Logging
Pleasure Craft
Railroad
Recreational
Underground Mining
Total Nonroad
Annual Total Nonroad Emissions i
1999
232
968
3
143
98
326
72
398
11
218
130
151
2
2,754
2010
132
1,005
2
89
112
195
33
206
5
134
119
231
2
2,264
2015
99
1,083
2
85
118
151
23
195
5
106
117
194
1
2,179
2020
77
1,165
2
86
129
123
19
207
5
95
113
148
1
2,168
Tons)
2030
61
1,329
2
94
161
105
19
236
5
93
107
128
1
2,340
Table 2.2.-2S. Tons of Naphthalene Emissions from Nonroad Equipment Types,
1999 to 2030.
Equipment Type
Agriculture
Aircraft
Airport Support
Commercial
Commercial Marine Vessel
Construction
Industrial
Lawn/Garden
Logging
Pleasure Craft
Railroad
Recreational
Underground Mining
Total Nonroad
Annual Total Nonroad Emissions i
1999
42
456
1
104
65
56
26
305
2
34
61
59
0
1,212
2010
26
496
1
103
68
37
13
245
2
36
44
112
0
1,182
2015
21
530
0
113
72
30
9
246
2
37
42
127
0
1,228
2020
17
566
0
125
79
22
6
264
1
39
40
132
0
1,291
Tons)
2030
12
638
0
149
102
17
4
303
2
42
35
136
0
1,440
Portable Fuel Containers - Table 2.2.-24 summarizes nationwide emissions of
individual air toxics from gasoline in portable fuel containers (PFCs), from 1999 to 2030.
2-56
-------
Final Regulatory Impact Analysis
Table 2.2.-24. Tons of Air Toxic Emissions from Portable Fuel Containers, 1999 to
2030.
Pollutant
2,2,4-Trimethylpentane
Benzene
Ethyl Benzene
Hexane
MTBE
Naphthalene
Toluene
Xylenes
Total
1999
4,870
853
2,135
5,417
6,969
1
10,733
6,189
37,166
2010
4,461
833
1,900
5,176
4,763
1
9,668
5,432
32,232
2015
4,741
889
2,027
5,532
4,987
1
10,329
5,800
34,306
2020
5,088
953
2,175
5,935
5,007
1
11,082
6,223
36,464
2030
5,805
1,086
2,480
6,766
5,503
1
12,636
7,096
41,374
About 75% of all HAP emissions and benzene emissions from PFCs are associated with
residential use, and the rest are from commercial use. As can be seen in Figure 2.2.-7,
most commercial PFC air toxic emissions are associated with equipment refueling, and
most residential emissions are associated with evaporation and permeation.
Figure 2.1.-7. Distribution of air toxic emissions (tons) among emission types for
commercial versus residential PFCs, 1999.
Air Toxic Emissions from Commercial PFCs, 1999
(Tons)
Evaporation, Permeation, 240,
Spillage During61'5%
Transport, 2124,^
21%
Refueling
Equipment:
Vapor
Displacement,
957,10%
Refilling at Pump:
Spillage, 84,1%
Refilling at Pump:
^_ Vapor
Displacement,
957,10%
Refueling
Equipment:
Spillage, 5172,
51%
2-57
-------
Final Regulatory Impact Analysis
Air Toxic Emissions from Residential PFCs, 1999,
Refueling
Equipment:
Spillage, 2644,
10%
Refilling at
Pump: Vapor
Displacement, ^x\
497, 2%
Refilling at
Pump: Spillage,^
44, 0%
Refueli ng
Equipment:
Vapor
Displacement,
\ 497,2%
Tons
Spillage During
^Transport, 1557,
6%
Permeation
7505, 28%
Evaporation,
14429,52%
Diesel Paniculate Matter - The inventory estimates presented above for mobile
source air toxics do not include diesel particulate matter. Table 2.2.-25 summarizes the
trend in diesel particulate matter between 1999 and 2030, by source category. These
inventory estimates were obtained from EPA's recently proposed national ambient air
quality standard for particulate matter.55 Diesel particulate matter emissions will be
reduced by 75% between 2001 and 2030. As controls on highway diesel engines and
nonroad diesel engines phase in, diesel-powered locomotives and commercial marine
vessels increase from 13% of the inventory in 2001 to 55% in 2030.
Table 2.2.-2S. Percent Contribution of Mobile Source Categories to Diesel
Particulate Matter (PMio) Emissions, 2001 to 2030 in Tons Per Year (Percent of
Total).
Source
Highway Vehicles
Commercial
Marine Vessels
Locomotives
Other Nonroad
Equipment
2001
125,162
(36.7%)
20,541
(6%)
25,173
(7.4%)
170,357
(49.9%)
2015
37,463
(24.8%)
17,085
(11.3%)
17,521
(11.6%)
78,930
(52.3%)
2020
26,471
(24.4%)
16,984
(15.7%)
16,535
(15.3%)
48,284
(44.6%)
2030
18,135
(21.6%)
21,388
(25.5%)
25,086
(29.9%)
19,285
(23.0%)
2.2.1.2.2 Impact on Inventory of Controls
The controls being finalized in this rule would reduce air toxic emissions from
highway gasoline vehicles, nonroad gasoline equipment, gasoline distribution and
2-58
-------
Final Regulatory Impact Analysis
portable fuel containers. The total air toxic emissions reduced in the 2030 inventories
used for air quality modeling for these sectors are 335,000 tons, and the total benzene
emissions reduced are 65,000 tons. It should be emphasized that the air quality, exposure
and risk modeling inventory does not account for recent increases in the use of ethanol
oxygenated gasoline or implementation of the renewable fuels standard. For inventories
which include these emissions, see Section 2.2.2.2.
Table 2.2.-26 summarizes the nationwide impact of the controls on emissions of
key air toxics from highway vehicles in 2015, 2020, and 2030. The reductions in
highway vehicle air toxic emissions by 2030 are dramatic, about 35%. Benzene
reductions are over 40%. Nonroad equipment emissions are impacted by fuel benzene
control, which result in reductions of about 14% for that pollutant (Table 2.2.-27).
Emissions from PFCs will be impacted by both controls on the containers themselves as
well as the fuel benzene standard (Table 2.2-28), with reductions in total air toxic
emissions of over 60% in 2030, and reductions in benzene of about 80%. In addition,
fuel benzene controls would reduce emissions within the gasoline distribution sector.
Table 2.2.-29 presents estimated reductions for this source in 2015 and 2020, which total
over 30%, due to the fuel benzene standard. Figures 2.2.-8 and 2.2.-9 depict the trend in
total MSAT and benzene emissions for all sources with the controls being finalized in
this rule. More detailed summaries of emissions by individual pollutant, by State, and for
urban versus rural areas can be found in Excel workbooks included in the docket for this
rule.
2-59
-------
Final Regulatory Impact Analysis
Table 2.2.-26. Nationwide Impact of Controls on Emissions of Gaseous Air Toxics from Highway Vehicles in 2015, 2020, and
2030.
Pollutant
1,3 -Butadiene
2,2,4-
Trimethylpentane
Acetaldehyde
Acrolein
Benzene
Ethyl Benzene
Formaldehyde
Hexane
MTBE
Propionaldehyde
Styrene
Toluene
Xylenes
Total
Annual Emissions (tons) by Vehicle Type
2015
Reference
Case
10,763
94,469
16,149
1,650
105,956
37,528
35,857
33,481
23,089
2,154
7,368
250,646
141,710
760,821
2015
Control
Case
9,160
80,630
13,970
1,458
79,034
32,189
31,475
30,802
22,363
1,925
6,134
212,901
120,444
642,486
2015
Reduction
1,602
13,840
2,180
192
26,922
5,339
4,382
2,679
725
230
1,234
37,745
21 ,266
118,336
2020
Reference
Case
11,355
96,315
16,893
1,704
110,317
38,080
37,153
30,727
18,372
2,222
7,814
257,367
145,473
773,793
2020
Control
Case
8,655
73,103
13,222
1,382
73,141
29,117
29,877
26,241
17,226
1,837
5,743
194,002
109,772
583,319
2020
Reduction
2,700
23,212
3,671
322
37,176
8,962
7,276
4,486
1,146
385
2,071
63,365
35,701
190,474
2030
Reference
Case
13,378
110,895
19,879
1,981
129,290
43,676
43,404
32,435
17,109
2,574
9,253
297,748
168,285
889,908
2030
Control
Case
8,707
72,262
13,677
1,434
72,673
28,770
31,196
25,832
16,080
1,919
5,720
191,607
108,480
578,358
2030
Reduction
4,670
38,634
6,202
548
56,617
14,906
12,207
6,602
1,029
655
3,533
106,141
59,805
311,549
2-60
-------
Final Regulatory Impact Analysis
Table 2.2.-21. Nationwide Impact of Controls on Emissions of Key Air Toxics from all Nonroad Equipment in 2015, 2020, and
2030.
Pollutant
1,3 -Butadiene
Acetaldehyde
Acrolein
Benzene
Formaldehyde
5 MSAT Total
Annual Emissions (tons)
2015
Reference
Case
6586
14459
2179
49985
34406
107615
2015
Control
Case
6599
14468
2179
43220
34433
101418
2015
Reduction
-13
-9
0
6265
-27
6197
2020
Reference
Case
6518
13663
2168
48453
32678
103480
2020
Control
Case
6530
13671
2168
41736
32703
97339
2020
Reduction
-12
-8
0
6717
-25
6141
2030
Reference
Case
7004
14153
2340
51647
33994
109138
2030 Control
Case
7017
14162
2340
44427
34020
102528
2030
Reduction
-13
-9
0
7220
-26
6610
2-61
-------
Final Regulatory Impact Analysis
Table 2.2.-2S. Nationwide Impact of Controls on Emissions of Air Toxics from Portable Fuel Containers in 2010, 2015, 2020,
and 2030.
Pollutant
2,2,4-
Trimethylpentane
Benzene
Ethyl Benzene
Hexane
MTBE
Naphthalene
Toluene
Xylenes
Total
2010
Reference
Case
4,461
833
1,900
5,176
4,763
1
9,668
5,432
32,232
2010
Control
Case
4,003
752
1,700
4,622
4,295
1
8,646
4,862
28,880
Annual Emissions (tons)
2015
Reference
Case
4,741
889
2,027
5,532
4,987
1
10,329
5,800
34,306
2015
Control
Case
1,864
179
756
1,932
2,360
0
3,752
2,157
13,000
2020
Reference
Case
5,088
953
2,175
5,935
5,007
1
11,082
6,223
36,464
2020
Control
Case
2,012
193
816
2,085
2,382
0
4,050
2,328
13,867
2030
Reference
Case
5,805
1,086
2,480
6,766
5,503
1
12,636
7,096
41,374
2030
Control
Case
2,315
222
939
2,399
2,638
0
4,658
2,678
15,849
Table 2.2.-29. Nationwide Impact of Controls Emissions of Benzene from Gasoline Distribution in 2015 and 2020 (2030
Assumed to be the Same as 2020).
Annual Emissions (tons)
2015
Reference
Case
2,160
2015
Control
Case
1,460
2015
Reduction
700
2020
Reference
Case
2,234
2020
Control
Case
1,516
2020
Reduction
719
2-62
-------
Final Regulatory Impact Analysis
Figure 2.2.-S. Contribution of Source Categories to Mobile Source Air Toxic
Emissions, 1999 to 2030, with Final Rule Standards in Place (Not Including Diesel
Particulate Matter).
• Nonroad Equipment
DHighway Vehicles
D Fires -- Prescribed and Wild
• Area and Other
D Major
Figure 2.2.-9. Contribution of Source Categories to Mobile Source Benzene
Emissions, 1999 to 2030, with Final Rule Standards in Place.
• Nonroad Equipment
D Highway Vehicles
D Fires - Prescribed and Wild
•Area and Other
D Major
2-63
-------
Final Regulatory Impact Analysis
2.2.2 Emission Reductions from Controls
Section 2.2.2 describes revisions made to emission inventories after we developed
MSAT inventories for air quality modeling ("air quality inventories"). The primary
revision is accounting for the impacts of implementing the federal Renewable Fuel
Standard (RFS). The revised inventories were used to estimate emission benefits of the
rule and the cost-effectiveness of the control strategies. We refer to the revised
inventories as "cost-effectiveness inventories" in this section to distinguish them from the
air quality inventories discussed in Section 2.2.1.
2.2.2.1 Methodology Changes from Air Quality Inventories
2.2.2.1.1 Highway Vehicles
The fundamental difference between the air quality and cost-effectiveness
inventories is the use of fuel parameters that reflect implementation of the Renewable
Fuel Standard (RFS fuel), as described in Section 2.1.1. We also corrected a minor error
which addresses how MOBILE6.2 calculates benzene evaporative emissions with ethanol
oxygenated fuel. In addition, for the control case, aromatics levels were adjusted using a
different algorithm to calculate additive adjustment factors:
Additive Factor = 1.0 (BZ(control) - BZ(ref)) (11)
Where BZ = benzene
We assume that with increased ethanol use, when fuel benzene is reduced there will be no
increase in other aromatic levels to help compensate for octane loss. An Excel workbook
with all the fuel parameters used, "MSAT Fuels Cost Effectiveness.xls," is included in
the docket for this rule. Also, we estimated vehicle refueling emissions using NMIM
2005, instead of projecting them from the 1999 NEI, as discussed in Section 2.2.1.
Finally, it should be noted that inventories do not account increased permeation due to
ethanol use, nor do they account for the 1.3 vol% maximum average fuel benzene level.
2.2.2.1.2 Nonroad Equipment
Unlike the air quality inventories, the cost-effectiveness inventories for nonroad
equipment used the RFS fuel as described in Section 2.1. As with the air quality
inventories, we assumed that changes in county-level exhaust and evaporative emissions
of nonroad gasoline equipment were proportional to changes in highway light-duty
gasoline vehicle emissions. It should be noted that our inventories did not account for
increased hose and tank permeation associated with increased ethanol use. As a result,
our estimates of emission reductions from fuel benzene control may be slightly
underestimated.
2-64
-------
Final Regulatory Impact Analysis
2.2.2.1.3 Portable Fuel Containers
The RFS fuel was used to develop cost-effectiveness inventories for PFCs, as
described in Section 2.1. Air toxic inventories for PFCs for the reference and control
cases were developed by speciating VOC, following the same approach used for the air
quality modeling inventories (See Section 2.2.1.1.4). However, since the air quality
modeling inventories did not account for RFS fuel, we used revised highway gasoline
vehicle inventories for benzene and VOC from refueling that did account for RFS fuel to
develop benzene to VOC ratios, and total evaporative emission ratios for other air toxics.
2.2.2.1.4 Gasoline Distribution
Gasoline distribution inventories were also revised to account for the RFS fuel.
The reference case (RC) inventory was estimated for each source category (SCC code) at
the county level as follows:
GasDistr. Benzene Emissions RCSCC ^ CounlyZ: RFSMMal =
^ r,- D D/-Z7 • • ( Nonrefueling evap Benzene RC1DGrcz
GasDistr.BenzeneRCEmisswnsSCCYYY — - — — - - - —
SCCYYYCa z FinalRtlleAQIn . — - — — - - - —
{Nonrefuelmg evap Benzene RC
LDGVCountyZAQImentca
Where,
Final Rule AQ Inventory = the inventory for SCC code YYY in county Z from the air
quality inventory, as discussed in Section 2.2.1
RFS Max 9.6 = the inventory for SCC code YYY in county Z assuming 9.6 billion
gallons of national ethanol consumption nationwide, attributing as much as possible for
use as an oxygenate in reformulated gasoline.
The air quality inventory was adjusted using ratios of non-refueling evaporative
emissions, because the methodology for estimating refueling emissions differed for the
air quality inventory versus the final rule inventory, as discussed above.
The control case (CC) inventory was estimated using the following equation:
Gas Distr. Benzene Emissions CC
Gas Distr. Benzene RC Emissions
JSCC YYY, County Z, RFS 9.6Max ~
"Re fueling BenzeneCCWQVtCountyZfRFS96Max
SCC YYY, County Z,RFS9.6Max ~
'(13)
Re fueling Benzene RCLDGr^ County z ^ 96Max
2.2.2.2 Estimated Reductions for Air Toxic Pollutants of Greatest Concern
The following sections present control case inventories and reductions for each
individual control, and then cumulative reductions for all controls combined.
2-65
-------
Final Regulatory Impact Analysis
2.2.2.2.1 Fuel Benzene Standard
Highway Gasoline Vehicles - The fuel benzene standard will reduce emissions
from light-duty gasoline vehicles and trucks, motorcycles, and heavy-duty gasoline
trucks. Tables 2.2.-30, 2.2.-31, and 2.2.-32 present nationwide benzene emissions for
these vehicle classes with and without the fuel standard in 2015, 2020, and 2030. Total
benzene emissions from these vehicle classes were 178,000 tons in 1999. Since impacts
of fuel benzene control on emissions of other MSATs are negligible (see Section 2.2.1.2),
they are not presented here, although they are available in the docket for the rule.
Table 2.2.-30. Impact of Fuel Benzene Control on Benzene Emissions from
Highway Vehicles, by Class, 2015.
Vehicle Class
LDGV
LDGT1
LDGT2
MC
HDGV
TOTAL
Reference Case
Tons
37,881
39,657
17,696
773
1,782
97,789
Control Case Tons
33,766
35,279
15,658
663
1,509
86875
Reduction
4,115
4,378
2,037
110
273
10914
Table 2.2.-31. Impact of Fuel Benzene Control on Benzene Emissions from
Highway Vehicles, by Class, 2020.
Vehicle Class
LDGV
LDGT1
LDGT2
MC
HDGV
TOTAL
Reference Case
Tons
35,987
44,611
18,627
833
1,456
101514
Control Case Tons
32,213
39,849
16,572
714
1,240
90,588
Reduction
3,774
4,762
2,056
118
215
10926
2-66
-------
Final Regulatory Impact Analysis
Table 2.2.-S2. Impact of Fuel Benzene Control on Benzene Emissions from
Highway Vehicles, by Class, 2030.
Vehicle Class
LDGV
LDGT1
LDGT2
MC
HDGV
TOTAL
Reference Case
Tons
42,752
52,993
20,996
1,002
1,273
119016
Control Case Tons
38,345
47,477
18,738
861
1,081
106502
Reduction
4,407
5,516
2,259
141
192
12514
Reductions from the fuel benzene control vary significantly across the U.S.,
depending on the average level of benzene in gasoline sold, as discussed in Section
2.2.1.2 on air quality modeling inventories. Table 2.2.-33 summarizes impacts of fuel
benzene control on the benzene emission inventory for gasoline vehicles in each State in
2030.
2-67
-------
Final Regulatory Impact Analysis
Table 2.2.-3S. Impacts of Fuel Control on Gasoline Vehicle Benzene by State in
2030.
State
ALABAMA
ALASKA
ARIZONA
ARKANSAS
CALIFORNIA
COLORADO
CONNECTICUT
DELAWARE
DISTRICT OF COLUMBIA
FLORIDA
GEORGIA
HAWAII
IDAHO
ILLINOIS
INDIANA
IOWA
KANSAS
KENTUCKY
LOUISIANA
MAINE
MARYLAND
MASSACHUSETTS
MICHIGAN
MINNESOTA
MISSISSIPPI
MISSOURI
MONTANA
NEBRASKA
NEVADA
NEW HAMPSHIRE
NEW JERSEY
NEW MEXICO
NEW YORK
NORTH CAROLINA
NORTH DAKOTA
OHIO
OKLAHOMA
OREGON
PENNSYLVANIA
RHODE ISLAND
SOUTH CAROLINA
SOUTH DAKOTA
TENNESSEE
TEXAS
UTAH
VERMONT
VIRGINIA
WASHINGTON
WEST VIRGINIA
WISCONSIN
WYOMING
2030 Reference
Case Tons
2,260.4
1,304.4
1,788.9
1,349.2
9,422.4
2,728.3
1,033.1
269.6
112.0
4,175.1
4,176.9
189.7
1,149.1
4,075.3
3,392.9
1,580.3
1,385.9
1,988.4
1,540.8
668.5
1,716.1
1,690.7
5,642.0
4,086.7
967.8
2,839.5
795.0
985.5
1,021.5
641.0
1,858.8
1,739.4
4,519.9
3,922.7
515.0
4,619.9
1,808.7
3,724.9
4,102.3
324.4
2,038.9
523.5
2,545.4
6,294.5
1,731.2
463.7
3,312.0
5,856.9
862.5
2,693.8
581.2
2030 Control
Case Tons
2,013.9
895.8
1,631.6
1,197.4
9,387.8
2,359.2
1,019.1
265.7
110.5
3,687.3
3,781.8
188.9
969.7
3,740.6
2,956.3
1,354.9
1,154.5
1,749.9
1,356.0
639.8
1,667.0
1,667.2
4,827.3
3,349.5
856.5
2,471.7
675.0
831.7
969.3
614.6
1,833.1
1,448.3
4,278.1
3,521.1
430.1
4,005.6
1,609.1
3,108.3
3,821.3
320.0
1,833.7
447.6
2,239.0
5,651.4
1,488.9
428.1
3,109.4
4,888.8
759.9
2,397.2
491.5
2030 Tons
Reduced
246.4
408.7
157.4
151.8
34.6
369.2
14.0
3.9
1.5
487.8
395.1
0.7
179.5
334.8
436.6
225.4
231.4
238.5
184.8
28.7
49.2
23.4
814.8
737.2
111.3
367.8
120.0
153.8
52.2
26.4
25.6
291.0
241.8
401.6
84.9
614.3
199.6
616.6
281.0
4.4
205.2
75.8
306.3
643.1
242.3
35.6
202.6
968.1
102.6
296.6
89.7
% Change
10.9
31.3
8.8
11.3
0.4
13.5
1.4
1.4
1.3
11.7
9.5
0.4
15.6
8.2
12.9
14.3
16.7
12.0
12.0
4.3
2.9
1.4
14.4
18.0
11.5
13.0
15.1
15.6
5.1
4.1
1.4
16.7
5.3
10.2
16.5
13.3
11.0
16.6
6.8
1.4
10.1
14.5
12.0
10.2
14.0
7.7
6.1
16.5
11.9
11.0
15.4
2-68
-------
Final Regulatory Impact Analysis
Gasoline Nonroad Equipment - Table 2.2.-34 summarizes the nationwide impact of the
fuel benzene control on benzene emissions from gasoline nonroad equipment. As with
highway gasoline vehicles, emission benefits vary across the U. S. As can be seen in
Table 2.2.-35, these benefits vary from 1 to 31% by State in 2030.
Table 2.2-34. Nationwide Impact of Fuel Benzene Control on Benzene Emissions
from Nonroad Gasoline Equipment.
2015 Reference Case
2015 Control Case
20 15 Reduction
2020 Reference Case
2020 Control Case
2020 Reduction
2030 Reference Case
2030 Control Case
2030 Reduction
Tons
41,343
35,825
5,518
40,161
34,717
5,444
42,994
37168
5,826
Portable Fuel Containers -Table 2.2.-36 summarizes MSAT emissions from
PFCs with no fuel benzene or federal PFC control (but including State control programs).
The fuel benzene control will reduce benzene emissions from PFCs, as summarized in
Table 2.2.-37. Again, emission benefits vary across the U. S., as seen in Table 2.2.-38.
Gasoline Distribution -Table 2.2.-39 presents the benzene inventory from
gasoline distribution (not including refueling) in 2015 and 2020 with and without the fuel
benzene control. Table 2.2.-40 presents the inventory for 2020 at the State level with and
without fuel benzene control. More detailed inventory estimates by county are available
in the docket for the rule.
2-69
-------
Final Regulatory Impact Analysis
Table 2.2.-3S. Gasoline Nonroad Equipment Benzene Emission Reductions (Tons)
from Fuel Control by State, 2030.
State
ALABAMA
ALASKA
ARIZONA
ARKANSAS
CALIFORNIA
COLORADO
CONNECTICUT
DELAWARE
DISTRICT OF COLUMBIA
FLORIDA
GEORGIA
HAWAII
IDAHO
ILLINOIS
INDIANA
IOWA
KANSAS
KENTUCKY
LOUISIANA
MAINE
MARYLAND
MASSACHUSETTS
MICHIGAN
MINNESOTA
MISSISSIPPI
MISSOURI
MONTANA
NEBRASKA
NEVADA
NEW HAMPSHIRE
NEW JERSEY
NEW MEXICO
NEW YORK
NORTH CAROLINA
NORTH DAKOTA
OHIO
OKLAHOMA
OREGON
PENNSYLVANIA
RHODE ISLAND
SOUTH CAROLINA
SOUTH DAKOTA
TENNESSEE
TEXAS
UTAH
VERMONT
VIRGINIA
WASHINGTON
WEST VIRGINIA
WISCONSIN
WYOMING
2030 Reference
Case Tons
1,024.8
188.3
715.3
637.6
4,055.2
623.4
412.9
98.2
28.1
3,752.5
1,576.1
127.6
285.1
1,298.3
722.9
489.1
309.2
483.1
1,133.7
257.3
789.2
678.8
1,585.6
900.5
534.4
778.8
133.6
209.1
310.3
270.3
1,053.7
266.5
2,366.1
1,654.7
107.2
1,329.4
596.0
639.2
1,516.0
110.5
907.6
109.5
739.7
3,156.9
356.5
134.2
1,105.1
1,039.3
326.6
962.0
107.4
2030 Control
Case Tons
830.6
129.4
615.1
514.1
4,032.6
525.5
403.5
95.8
27.4
3,070.6
1,324.2
126.7
232.3
1,192.6
605.6
404.2
245.4
403.6
896.6
238.2
737.3
663.1
1,288.9
708.8
428.8
654.7
110.3
168.4
279.1
245.4
1,029.4
208.0
2,154.0
1,382.2
86.0
1,100.3
483.6
514.6
1,353.4
108.0
749.8
89.0
606.4
2,688.5
294.5
116.4
990.5
841.3
269.3
815.1
87.8
2030 Tons
Reduced
194.2
58.9
100.2
123.5
22.6
97.9
9.4
2.4
0.8
682.0
251.9
0.9
52.8
105.7
117.4
84.9
63.7
79.5
237.1
19.0
52.0
15.7
296.7
191.7
105.6
124.1
23.3
40.8
31.2
24.9
24.3
58.5
212.1
272.5
21.2
229.1
112.4
124.7
162.6
2.5
157.8
20.5
133.3
468.5
62.0
17.8
114.7
198.0
57.3
146.9
19.6
% Change
19.0
31.3
14.0
19.4
0.6
15.7
2.3
2.5
2.8
18.2
16.0
0.7
18.5
8.1
16.2
17.4
20.6
16.4
20.9
7.4
6.6
2.3
18.7
21.3
19.8
15.9
17.5
19.5
10.0
9.2
2.3
21.9
9.0
16.5
19.8
17.2
18.9
19.5
10.7
2.2
17.4
18.7
18.0
14.8
17.4
13.3
10.4
19.0
17.5
15.3
18.2
2-70
-------
Final Regulatory Impact Analysis
Table 2.2.-S6. MSAT Emissions (Tons) from Uncontrolled PFCs (No Fuel Benzene
Control, No Federal PFC Control, But Including State Programs)
Pollutant
2,2,4-Trimethylpentane
Benzene
Ethylbenzene
n-Hexane
MTBE
Naphthalene
Toluene
Xylenes
TOTAL
1999
4,870
853
2,135
5,417
6,969
1
10,733
6,189
37,167
2010
4,994
943
1,805
4,679
0
1
8,764
5,004
21,186
2015
5,195
992
1,884
4,895
0
1
9,161
5,226
27,354
2020
5,573
1,063
2,021
5,250
0
1
9,825
5,605
29,338
2030
6,353
1,210
2,303
5,981
0
1
11,195
6,387
33,430
Table 2.2.-S7. Reduction in Benzene PFC Emissions (Tons) with Fuel Control (No
Control on PFC Emissions).
Year
1999
2015
2020
2030
Reference Case
853
992
1,063
1,210
Control Case
N.A.
619
664
756
Reduction
N.A.
373
399
454
2-71
-------
Final Regulatory Impact Analysis
Table 2.2.-3S. Reduction in Benzene PFC Emissions (Tons) with Fuel Control in
2030 by State (No Additional Control on PFC Emissions).
State
ALABAMA
ALASKA
ARIZONA
ARKANSAS
CALIFORNIA
COLORADO
CONNECTICUT
DELAWARE
DISTRICT OF COLUMBIA
FLORIDA
GEORGIA
HAWAII
IDAHO
ILLINOIS
INDIANA
IOWA
KANSAS
KENTUCKY
LOUISIANA
MAINE
MARYLAND
MASSACHUSETTS
MICHIGAN
MINNESOTA
MISSISSIPPI
MISSOURI
MONTANA
NEBRASKA
NEVADA
NEW HAMPSHIRE
NEW JERSEY
NEW MEXICO
NEW YORK
NORTH CAROLINA
NORTH DAKOTA
OHIO
OKLAHOMA
OREGON
PENNSYLVANIA
RHODE ISLAND
SOUTH CAROLINA
SOUTH DAKOTA
TENNESSEE
TEXAS
UTAH
VERMONT
VIRGINIA
WASHINGTON
WEST VIRGINIA
WISCONSIN
WYOMING
Reference Case
Tons
33.1
19.3
25.8
23.7
36.4
31.5
3.5
1.1
0.4
138.2
42.5
5.1
12.8
42.6
38.0
20.8
20.8
25.3
39.2
1.7
7.0
5.9
64.1
40.9
24.6
31.7
8.0
12.7
11.9
1.7
8.7
14.0
17.9
54.9
4.5
33.3
20.8
29.9
16.3
0.9
31.5
4.8
41.8
42.5
16.7
0.8
11.5
44.5
14.3
26.0
4.2
Control Case
Tons
19.1
11.6
15.5
13.5
35.7
19.8
3.2
1.0
0.4
82.9
25.5
5.0
8.1
32.9
21.4
11.2
11.2
16.1
22.4
1.3
6.2
5.4
34.6
22.1
14.0
19.9
5.1
6.8
7.8
1.4
8.0
8.0
14.4
32.9
2.4
18.0
11.3
17.9
11.5
0.9
18.9
2.6
22.6
30.6
10.5
0.5
8.9
26.7
8.6
16.9
2.7
Reduction
14.0
7.7
10.3
10.2
0.7
11.7
0.3
0.1
0.0
55.3
17.0
0.1
4.7
9.6
16.6
9.6
9.6
9.2
16.9
0.3
0.9
0.5
29.5
18.8
10.6
11.8
3.0
5.8
4.1
0.3
0.7
6.0
3.5
21.9
2.1
15.3
9.6
12.0
4.8
0.1
12.6
2.2
19.2
12.0
6.2
0.3
2.6
17.8
5.7
9.1
1.6
% Change
42.3
40.0
40.0
43.0
2.0
37.0
8.0
8.0
8.0
40.0
40.0
2.0
37.0
22.6
43.7
46.0
46.0
36.4
43.0
20.0
12.5
8.0
46.0
46.0
43.0
37.2
37.0
46.0
34.5
16.5
8.0
43.0
19.6
40.0
46.0
46.0
46.0
40.0
29.5
8.0
40.0
46.0
46.0
28.1
37.0
40.0
22.5
40.0
40.0
34.9
37.0
2-72
-------
Final Regulatory Impact Analysis
Table 2.2.-S9. Nationwide Impact of Controls on Emissions of Benzene from
Gasoline Distribution in 2015 and 2020.
Tons of
Benzene
2015
Reference
Case
2,445
2015
Control
Case
1,635
2015
Reduction
810
2020
Reference
Case
2,621
2020
Control
Case
1,772
2020
Reduction
849
2-73
-------
Final Regulatory Impact Analysis
Table 2.2.-40. Reduction in Gasoline Distribution Emissions of Benzene (Tons) with
Fuel Benzene Control by State, 2020.
State
ALABAMA
ALASKA
ARIZONA
ARKANSAS
CALIFORNIA
COLORADO
CONNECTICUT
DELAWARE
DISTRICT OF COLUMBIA
FLORIDA
GEORGIA
HAWAII
IDAHO
ILLINOIS
INDIANA
IOWA
KANSAS
KENTUCKY
LOUISIANA
MAINE
MARYLAND
MASSACHUSETTS
MICHIGAN
MINNESOTA
MISSISSIPPI
MISSOURI
MONTANA
NEBRASKA
NEVADA
NEW HAMPSHIRE
NEW JERSEY
NEW MEXICO
NEW YORK
NORTH CAROLINA
NORTH DAKOTA
OHIO
OKLAHOMA
OREGON
PENNSYLVANIA
RHODE ISLAND
SOUTH CAROLINA
SOUTH DAKOTA
TENNESSEE
TEXAS
UTAH
VERMONT
VIRGINIA
WASHINGTON
WEST VIRGINIA
WISCONSIN
WYOMING
Reference Case
Tons
34.0
3.0
40.9
16.5
98.0
27.5
18.6
3.4
3.4
88.4
36.6
3.1
20.2
98.6
41.2
53.6
70.0
48.4
108.9
21.4
34.1
25.0
82.3
74.3
42.6
26.6
12.0
11.1
7.8
3.9
41.7
26.3
359.7
36.2
9.3
89.7
61.0
118.6
72.7
4.7
19.9
6.5
57.9
344.2
25.5
1.2
42.1
57.1
55.2
23.7
11.8
Control Case
Tons
19.6
1.8
24.6
9.4
96.1
17.3
17.1
3.1
3.1
53.1
22.0
3.1
12.7
69.0
23.7
29.0
37.8
28.4
62.1
16.3
28.6
23.0
44.5
40.1
24.3
16.6
7.5
6.0
5.7
3.2
38.3
15.0
313.4
21.7
5.0
48.4
32.9
71.2
46.1
4.4
11.9
3.5
31.3
243.6
16.0
0.7
29.8
34.3
33.1
15.1
7.4
Reduction
14.4
1.2
16.4
7.1
2.0
10.2
1.5
0.3
0.3
35.4
14.6
0.1
7.5
29.6
17.5
24.7
32.2
20.0
46.8
5.1
5.5
2.0
37.9
34.2
18.3
9.9
4.4
5.1
2.1
0.6
3.3
11.3
46.3
14.5
4.3
41.3
28.1
47.4
26.6
0.4
8.0
3.0
26.7
100.6
9.4
0.5
12.3
22.8
22.1
8.6
4.4
% Change
42.4
40.0
40.0
43.0
2.0
37.0
8.0
8.0
8.0
40.0
40.0
2.0
37.0
30.0
42.5
46.0
46.0
41.3
43.0
23.9
16.1
8.0
46.0
46.0
43.0
37.4
37.0
46.0
27.1
16.7
8.0
43.0
12.9
40.0
46.0
46.0
46.0
40.0
36.6
8.0
40.0
46.0
46.0
29.2
37.0
40.0
29.2
40.0
40.0
36.4
37.0
2-74
-------
Final Regulatory Impact Analysis
2.2.2.2.2 Cold Temperature VOC Emission Control
Reductions in MSATs are proportional to reduced NMHC start emissions from
vehicles subject to this rule. The magnitude of the reductions from these vehicles
operating on a given gasoline is based entirely on the number and duration of events
between engine off and engine on (vehicle soak) and the ambient conditions. The
emissions reduced are those created by the engine start following the vehicle soak. These
parameters are currently modeled by vehicle class and vehicle age in MOBILE6.2.56'57'
58'59 MOBILE6.2 also provides the necessary information to adjust MSAT emission
factors to account for geographic and seasonal effects on in-use fuels.
When all the affected vehicle classes meet the new emission standard we expect a
60% reduction of benzene and 1,3-butadiene from gasoline-fueled highway vehicles with
GVWR < 6000 Ibs and 30% from gasoline-fueled highway vehicles with GVWR > 6000
Ibs. This estimate does not include the effects of fuel benzene control. Effects on the
trends in the inventories for the affected MSATs are shown in Table 2.2.-41 through
Table 2.2.-4S.
Table 2.2.-41. Reference Case, Light-Duty Gasoline Vehicles and Trucks, 1999
MSAT Inventory.
Pollutant
1,3 -Butadiene
2,2,4-Trimethylpentane
Acetaldehyde
Acrolein
Benzene
Ethyl Benzene
Formaldehyde
n-Hexane
MTBE
Propionaldehyde
Styrene
Toluene
Xylenes
Total MSATs
Emissions in Tons
20,868
175,241
21,035
2,234
173,474
69,299
54,104
61,664
54,990
2440
13,070
464,646
262,298
1,376,002
2-75
-------
Final Regulatory Impact Analysis
Table 2.2.-42. Reference and Vehicle Control Case, Light-Duty Gasoline Vehicles
and Trucks, 2010 MSAT Inventories.
Pollutant
1,3 -Butadiene
2,2,4-Trimethylpentane
Acetaldehyde
Acrolein
Benzene
Ethyl Benzene
Formaldehyde
n-Hexane
MTBE
Propionaldehyde
Styrene
Toluene
Xylenes
Total MSATs
Reference Case
Tons in Calendar
Year 2010
9,159
95,194
16,680
1,132
99,559
36,001
23,466
32,850
0
1254
6,688
238,683
134,742
695,408
Vehicle Control
Case Tons in
Calendar Year 2010
8,417
88,628
15,220
1,041
91,621
33,489
21,371
31,590
0
1144
6,107
220,939
124,744
644,312
Reduction
in Tons
742
6,566
1,460
91
7,939
2,512
2,095
1,260
0
110
581
17,744
9,998
51,987
Percent
Reduction
8
7
9
8
8
7
9
4
0
9
9
7
7
7
Table 2.2.-4S. Reference and Vehicle Control Case, Light-Duty Vehicles, 2015
MSAT Inventories.
Pollutant
1,3 -Butadiene
2,2,4-Trimethylpentane
Acetaldehyde
Acrolein
Benzene
Ethyl Benzene
Formaldehyde
n-Hexane
MTBE
Propionaldehyde
Styrene
Toluene
Xylenes
Total MSATs
Reference Case
Tons in Calendar
Year 2015
8,635
87,857
16,253
1,080
95,234
33,276
22,657
27,699
0
1216
6,481
223,510
126,114
650,012
Vehicle Control
Case Tons in
Calendar Year 2015
7,083
73,956
13,123
887
78,664
27,970
18,298
25,034
0
985
5,254
186,031
104,997
542,281
Reduction
in Tons
1,552
13,901
3,131
193
16,570
5,305
4,359
2,665
0
231
1,227
37,480
21,117
107,731
Percent
Reduction
18
16
19
18
17
16
19
10
0
19
19
17
17
17
2-76
-------
Final Regulatory Impact Analysis
Table 2.2.-44. Reference and Vehicle Control Case, Light-Duty Vehicles, 2020
MSAT Inventories.
Pollutant
1,3 -Butadiene
2,2,4-Trimethylpentane
Acetaldehyde
Acrolein
Benzene
Ethyl Benzene
Formaldehyde
n-Hexane
MTBE
Propionaldehyde
Styrene
Toluene
Xylenes
Total MSATs
Reference Case
Tons in Calendar
Year 2020
9,131
89,711
17,345
1,139
99,225
33,992
24,007
25,765
0
1293
6,898
230,933
130,267
669,707
Vehicle Control
Case Tons in
Calendar Year 2020
6,592
66,807
12,143
822
72,128
25,268
16,922
21,380
0
914
4,880
169,303
95,543
492,700
Reduction
in Tons
2,539
22,904
5,203
317
27,097
8,724
7,086
4,385
0
379
2,018
61,630
34,725
177,007
Percent
Reduction
28
26
30
28
27
26
30
17
0
29
29
27
27
26
Table 2.2.-4S. Reference and Vehicle Control Case, Light-Duty Vehicles, 2030
MSAT Inventories.
Pollutant
1,3 -Butadiene
2,2,4-Trimethylpentane
Acetaldehyde
Acrolein
Benzene
Ethyl Benzene
Formaldehyde
n-Hexane
MTBE
Propionaldehyde
Styrene
Toluene
Xylenes
Total MSATs
Reference Case
Tons in Calendar
Year 2030
10,798
104,511
20,663
1,347
116,742
39,603
28,529
28,437
0
1540
8,207
270,625
152,647
783,648
Vehicle Control
Case Tons in
Calendar Year 2030
6,540
66,317
12,064
818
71,704
25,053
16,897
21,125
0
907
4,841
167,829
94,728
488,824
Reduction
in Tons
4,257
38,194
8,599
529
45,037
14,551
11,632
7,312
0
633
3,366
102,796
57,919
294,824
Percent
Reduction
39
37
42
39
39
37
41
26
0
41
41
38
38
38
2-77
-------
Final Regulatory Impact Analysis
State-level reductions in calendar year 2030 benzene inventories are reported in
Table 2.2.-46.
Table 2.2.-46. Impacts of Vehicle Control on Light-Duty Gasoline Vehicle Benzene
Emissions, by State in 2030
State
ALABAMA
ALASKA
ARIZONA
ARKANSAS
CALIFORNIA
COLORADO
CONNECTICUT
DELAWARE
DISTRICT OF COLUMBIA
FLORIDA
GEORGIA
HAWAII
IDAHO
ILLINOIS
INDIANA
IOWA
KANSAS
KENTUCKY
LOUISIANA
MAINE
MARYLAND
MASSACHUSETTS
MICHIGAN
MINNESOTA
MISSISSIPPI
MISSOURI
MONTANA
NEBRASKA
NEVADA
NEW HAMPSHIRE
NEW JERSEY
NEW MEXICO
NEW YORK
NORTH CAROLINA
NORTH DAKOTA
OHIO
OKLAHOMA
OREGON
PENNSYLVANIA
RHODE ISLAND
SOUTH CAROLINA
SOUTH DAKOTA
TENNESSEE
TEXAS
UTAH
VERMONT
VIRGINIA
WASHINGTON
WEST VIRGINIA
WISCONSIN
WYOMING
2030 Benzene Totals
2030 Reference
Case Benzene Tons
in Calendar 2030
2,199.4
1,293.1
1,734.6
1,323.0
9,286.2
2,674.7
1,019.9
264.8
109.9
4,032.2
4,076.8
183.7
1,132.6
4,004.2
3,335.4
1,564.5
1,362.3
1,952.6
1,502.6
657.6
1,689.2
1,649.1
5,560.1
4,038.3
946.8
2,787.8
785.0
970.8
989.8
632.5
1,815.8
1,698.0
4,421.5
3,836.4
509.0
4,536.7
1,771.7
3,631.9
4,026.6
320.3
1,998.9
516.9
2,495.0
6,111.9
1,705.9
457.1
3,271.3
5,778.8
852.8
2,651.4
573.9
116,741.6
2030 Control Case
Benzene Tons in
Calendar 2030
1,562.7
652.7
1,190.5
900.4
5,634.4
1,560.6
540.4
148.9
62.1
3,374.8
2,698.4
174.7
653.4
2,255.1
2,035.6
927.2
859.7
1,197.3
1,122.2
358.9
943.6
816.5
3,279.8
2,190.2
652.0
1,722.2
443.4
582.1
627.4
347.0
976.9
1,079.3
2,201.2
2,474.0
282.7
2,595.6
1,206.9
2,270.3
2,235.7
175.6
1,358.6
296.1
1,623.0
4,356.2
1,041.8
246.7
1,977.9
3,564.6
489.6
1,407.8
329.5
71,704.3
Reduction (Tons)
636.6
640.4
544.1
422.5
3,651.7
1,114.1
479.6
115.9
47.7
657.4
1,378.4
9.0
479.3
1,749.0
1,299.8
637.3
502.6
755.3
380.4
298.7
745.6
832.6
2,280.3
1,848.0
294.8
1,065.6
341.6
388.7
362.4
285.5
838.9
618.7
2,220.3
1,362.4
226.3
1,941.1
564.8
1,361.5
1,790.9
144.6
640.3
220.7
872.1
1,755.7
664.2
210.4
1,293.4
2,214.2
363.2
1,243.6
244.4
45,037.3
Percent Reduction
28.9
49.5
31.4
31.9
39.3
41.7
47.0
43.8
43.5
16.3
33.8
4.9
42.3
43.7
39.0
40.7
36.9
38.7
25.3
45.4
44.1
50.5
41.0
45.8
31.1
38.2
43.5
40.0
36.6
45.1
46.2
36.4
50.2
35.5
44.5
42.8
31.9
37.5
44.5
45.2
32.0
42.7
35.0
28.7
38.9
46.0
39.5
38.3
42.6
46.9
42.6
38.6
2-78
-------
Final Regulatory Impact Analysis
2.2.2.2.3 Portable Fuel Container Control
The effect of PFC control on nationwide MS AT emissions are reported in Tables
2.2.-47 through 2.2.-50. Table 2.2.-51 reports benzene reductions by State in 2030 as a
result of federal PFC control.
Table 2.2-41. Estimated Reductions in MSAT Emissions from PFC Control, 2010
(No Fuel Benzene Control).
Pollutant
2,2,4-Trimethylpentane
Benzene
Ethyl Benzene
n-Hexane
MTBE
Naphthalene
Toluene
Xylenes
Total
Reference Case
4,994
943
1,805
4,679
0
1
8,764
5,004
26,189
Control Case
4,039
743
1,450
3,742
0
1
7,021
4,015
21,010
Reduction in
Tons
955
201
355
937
0
0
1,743
989
5,179
Percent
Reduction
19
21
20
20
0
19
20
20
20
Table 2.2.-4S. Estimated Reductions in MSAT Emissions from PFC Control, 2015
(No Fuel Benzene Control).
Pollutant
2,2,4-Trimethylpentane
Benzene
Ethyl Benzene
n-Hexane
MTBE
Naphthalene
Toluene
Xylenes
Total
Reference Case
5,195
992
1,884
4,895
0
1
9,161
5,226
27,355
Control Case
2,005
320
695
1,750
0
0
3,316
1,912
9,998
Reduction in
Tons
3,190
672
1,189
3,145
0
0
5,846
3,314
17,357
Percent
Reduction
61
68
63
64
0
61
64
63
63
2-79
-------
Final Regulatory Impact Analysis
Table 2.2.-49. Estimated Reductions in MSAT Emissions from PFC Control, 2020
(No Fuel Benzene Control).
Pollutant
2,2,4-Trimethylpentane
Benzene
Ethyl Benzene
n-Hexane
MTBE
Naphthalene
Toluene
Xylenes
Total
Reference Case
5,573
1,063
2,021
5,250
0
1
9,825
5,605
29,338
Control Case
2,163
345
750
1,888
0
0
3,577
2,063
10,785
Reduction in
Tons
3,410
718
1,271
3,362
0
0
6,248
3,543
18,553
Percent
Reduction
61
68
63
64
0
61
64
63
63
Table 2.2.-50. Estimated Reductions in MSAT Emissions from PFC Control, 2030
(No Fuel Benzene Control).
Pollutant
2,2,4-Trimethylpentane
Benzene
Ethyl Benzene
n-Hexane
MTBE
Naphthalene
Toluene
Xylenes
Total
Reference Case
6,353
1,210
2,303
5,981
0
1
11,195
6,387
33,430
Control Case
2,486
396
862
2,169
0
0
4,110
2,370
12,394
Reduction in
Tons
3,867
814
1,442
3,812
0
1
7,085
4,017
21,036
Percent
Reduction
61
67
63
64
0
61
63
63
63
2-80
-------
Final Regulatory Impact Analysis
Table 2.2.-51. Reductions in Benzene Emissions (Tons) with PFC Control by State,
2030 (No Fuel Benzene Control).
State
ALABAMA
ALASKA
ARIZONA
ARKANSAS
CALIFORNIA
COLORADO
CONNECTICUT
DELAWARE
DISTRICT OF COLUMBIA
FLORIDA
GEORGIA
HAWAII
IDAHO
ILLINOIS
INDIANA
IOWA
KANSAS
KENTUCKY
LOUISIANA
MAINE
MARYLAND
MASSACHUSETTS
MICHIGAN
MINNESOTA
MISSISSIPPI
MISSOURI
MONTANA
NEBRASKA
NEVADA
NEW HAMPSHIRE
NEW JERSEY
NEW MEXICO
NEW YORK
NORTH CAROLINA
NORTH DAKOTA
OHIO
OKLAHOMA
OREGON
PENNSYLVANIA
RHODE ISLAND
SOUTH CAROLINA
SOUTH DAKOTA
TENNESSEE
TEXAS
UTAH
VERMONT
VIRGINIA
WASHINGTON
WEST VIRGINIA
WISCONSIN
WYOMING
Reference Case
33.1
19.3
25.8
23.7
36.4
31.5
3.5
1.1
0.4
138.2
42.5
5.1
12.8
42.6
38.0
20.8
20.8
25.3
39.2
1.7
7.0
5.9
64.1
40.9
24.6
31.7
8.0
12.7
11.9
1.7
8.7
14.0
17.9
54.9
4.5
33.3
20.8
29.9
16.3
0.9
31.5
4.8
41.8
42.5
16.7
0.8
11.5
44.5
14.3
26.0
4.2
Control Case
6.8
3.7
7.3
4.3
36.4
11.1
3.0
0.9
0.3
30.0
11.7
0.9
3.8
14.3
9.7
4.7
5.3
5.9
6.3
1.2
5.5
4.8
16.9
11.0
3.7
8.4
2.1
3.1
3.2
1.3
7.0
3.8
13.7
13.1
1.2
20.0
5.0
9.2
11.6
0.8
6.5
1.2
8.2
26.2
4.5
0.6
8.8
14.7
3.2
8.1
1.2
Reduction
26.3
15.6
18.6
19.4
0.0
20.4
0.4
0.3
0.1
108.3
30.8
4.2
9.0
28.3
28.3
16.1
15.5
19.3
32.9
0.4
1.6
1.0
47.1
29.9
20.9
23.3
6.0
9.5
8.7
0.4
1.7
10.2
4.2
41.7
3.4
13.3
15.8
20.6
4.7
0.2
25.0
3.6
33.6
16.3
12.1
0.2
2.7
29.8
11.1
17.9
3.0
% Change
79.5
80.9
71.8
82.0
0.0
64.9
12.8
22.5
32.3
78.3
72.4
82.4
70.7
66.5
74.4
77.5
74.5
76.6
83.9
26.4
22.1
17.6
73.6
73.1
85.1
73.4
74.2
75.2
73.1
24.4
19.5
73.0
23.5
76.1
73.9
39.9
75.9
69.1
28.7
18.9
79.4
75.1
80.3
38.4
72.7
26.0
23.6
66.9
77.5
69.0
70.5
2-81
-------
Final Regulatory Impact Analysis
2.2.2.2.4 Cumulative Reductions of Controls
Air toxic emissions from light-duty vehicles depend on both fuel benzene content
and vehicle hydrocarbon emission controls. Similarly, the air toxic emissions from PFCs
depend on both fuel benzene content and the PFC emission controls. Tables 2.2.-52 and
2.2.-53 summarize the expected reductions in benzene and MSAT emissions,
respectively, from the combined effects of our vehicle, fuel, and PFC controls.
Table 2.2.-54 summarizes the cumulative benzene emission reductions from these
controls on highway gasoline vehicles, nonroad gasoline vehicles, PFCs, and gasoline
distribution at the State level in 2030.
Table 2.2.-55 presents the impact of controls on total benzene emissions from
mobile sources and PFCs, and the impacts on total benzene emissions from all sources.
Table 2.2.-56 presents the cumulative impact of controls on total emissions of MSATs
from mobile sources and PFCs, as well as the impact on total emissions of MSATs from
both mobile and stationary sources. As discussed previously, the fuel benzene control
reduces stationary source emissions of benzene associated with gasoline distribution.
2-82
-------
Final Regulatory Impact Analysis
Table 2.2-52. Estimated Reductions in Benzene Emissions from All Control Measures by Sector, 2015 to 2030.
Benzene
Gasoline
Onroad Mobile
Sources
Gasoline
Nonroad
Mobile Sources
PFCs
Gasoline
Distribution
Total
1999
183,660
68,589
853
1,984
255,086
2015
Without
Rule (tons)
97,789
41,343
992
2,445
142,569
With Rule
(tons)
71,688
35,825
215
1,635
109,363
Reduction
(tons)
26,101
5,518
111
810
33,206
2020
Without
Rule (tons)
101,514
40,161
1,063
2,621
145,359
With Rule
(tons)
65,878
34,717
232
1,772
102,599
Reduction
(tons)
35,636
5,444
831
849
42,760
2030
Without
Rule (tons)
119,016
42,994
1,210
2,621
165,841
With Rule
(tons)
65,601
37,167
267
1,772
104,807
Reduction
(tons)
53,415
5,827
944
849
61,035
2-83
-------
Final Regulatory Impact Analysis
Table 2.2.-5S. Estimated Reductions in MSAT Emissions from All Control Measures by Sector, 2015 to 2030
MSAT
Gasoline
Onroad Mobile
Sources
Gasoline
Nonroad
Mobile Sources
PFCs
Gasoline
Distribution
Total
1999
1,452,739
806,725
37,166
57,765
2,354,395
2015
Without
Rule (tons)
675,781
449,422
27,355
62,870
1,215,428
With Rule
(tons)
558,666
443,973
9,893
62,059
1,074,591
Reduction
(tons)
117,115
5,449
17,462
811
140,837
2020
Without
Rule (tons)
693,189
406,196
29,338
64,942
1,193,665
With Rule
(tons)
507,782
400,816
10,672
64,092
983,362
Reduction
(tons)
185,408
5,380
18,666
850
210,303
2030
Without
Rule (tons)
808,141
412,617
33,430
64,942
1,319,130
With Rule
(tons)
505,074
406,856
12,264
64,092
988,286
Reduction
(tons)
303,067
5,761
21,166
850
330,844
2-84
-------
Final Regulatory Impact Analysis
Table 2.2.-S4. Cumulative Benzene Emission Reductions From All Controls at the State Level in 2030.
State
ALABAMA
ALASKA
ARIZONA
ARKANSAS
CALIFORNIA
COLORADO
CONNECTICUT
DELAWARE
DISTRICT OF COLUMBIA
FLORIDA
GEORGIA
HAWAII
IDAHO
ILLINOIS
INDIANA
IOWA
KANSAS
KENTUCKY
LOUISIANA
MAINE
MARYLAND
MASSACHUSETTS
MICHIGAN
MINNESOTA
MISSISSIPPI
MISSOURI
MONTANA
NEBRASKA
NEVADA
Gasoline Highway
Vehicles
Tons Reduced
826.3
849.8
665.3
534.7
3,675.9
1,348.7
488.0
118.3
48.7
1,090.6
1,667.6
9.8
588.6
1,966.1
1,591.8
780.9
658.6
915.2
528.2
315.5
776.5
845.7
2,799.1
2,270.1
378.9
1,316.8
413.0
487.8
402.7
%
36.6
65.1
37.2
39.6
39.0
49.4
47.2
43.9
43.5
26.1
39.9
5.1
51.2
48.2
46.9
49.4
47.5
46.0
34.3
47.2
45.2
50.0
49.6
55.5
39.2
46.4
52.0
49.5
39.4
Nonroad Gasoline Engines
Tons
Reduced
194.2
58.9
100.2
123.5
22.6
97.9
9.4
2.4
0.8
682.0
251.9
0.9
52.8
105.7
117.4
84.9
63.7
79.5
237.1
19.0
52.0
15.7
296.7
191.7
105.6
124.1
23.3
40.8
31.2
%
19.0
31.3
14.0
19.4
0.6
15.7
2.3
2.5
2.8
18.2
16.0
0.7
18.5
8.1
16.2
17.4
20.6
16.4
20.9
7.4
6.6
2.3
18.7
21.3
19.8
15.9
17.5
19.5
10.0
PFCs
Tons
Reduced
29.1
17.1
21.5
21.3
0.7
24.5
0.7
0.3
0.2
120.2
35.4
4.2
10.4
31.5
32.5
18.3
18.0
21.5
35.6
0.7
2.2
1.4
54.9
35.0
22.5
26.4
6.7
11.0
9.8
%
88.2
88.5
83.1
89.7
2.0
77.9
19.8
28.7
37.7
87.0
83.4
82.7
81.5
74.0
85.6
87.9
86.3
85.1
90.8
41.1
31.8
24.2
85.7
85.5
91.5
83.3
83.7
86.6
82.4
Gasoline Distribution
Tons
Reduced
14.4
1.2
16.4
7.1
2.0
10.2
1.5
0.3
0.3
35.4
14.6
0.1
7.5
29.6
17.5
24.7
32.2
20.0
46.8
5.1
5.5
2.0
37.9
34.2
18.3
9.9
4.4
5.1
2.1
%
42.4
40.0
40.0
43.0
2.0
37.0
8.0
8.0
8.0
40.0
40.0
2.0
37.0
30.0
42.5
46.0
46.0
41.3
43.0
23.9
16.1
8.0
46.0
46.0
43.0
37.4
37.0
46.0
27.1
Total
Tons
Reduced
1,064.0
926.9
803.4
686.5
3,701.2
1,481.2
499.6
121.4
49.9
1,928.2
1,969.5
15.0
659.3
2,132.9
1,759.2
908.8
772.5
1,036.1
847.8
340.3
836.2
864.8
3,188.6
2,530.9
525.4
1,477.3
447.5
544.6
445.8
%
31.7
61.2
31.2
33.9
27.2
43.4
34.0
32.6
34.7
23.6
33.8
4.6
44.9
38.7
41.9
42.4
43.3
40.7
30.0
35.9
32.8
36.0
43.2
49.6
33.5
40.2
47.2
44.7
33.0
2-85
-------
Final Regulatory Impact Analysis
State
NEW HAMPSHIRE
NEW JERSEY
NEW MEXICO
NEW YORK
NORTH CAROLINA
NORTH DAKOTA
OHIO
OKLAHOMA
OREGON
PENNSYLVANIA
RHODE ISLAND
SOUTH CAROLINA
SOUTH DAKOTA
TENNESSEE
TEXAS
UTAH
VERMONT
VIRGINIA
WASHINGTON
WEST VIRGINIA
WISCONSIN
WYOMING
Gasoline Highway
Vehicles
Tons Reduced
301.2
855.1
814.4
2,354.8
1,648.3
276.2
2,326.7
717.4
1,774.1
1,963.3
147.3
791.2
267.2
1,093.5
2,255.3
821.5
230.9
1,427.6
2,848.3
425.9
1,417.3
298.3
%
47.0
46.0
46.8
52.1
42.0
53.6
50.4
39.7
47.6
47.9
45.4
38.8
51.0
43.0
35.8
47.5
49.8
43.1
48.6
49.4
52.6
51.3
Nonroad Gasoline Engines
Tons
Reduced
24.9
24.3
58.5
212.1
272.5
21.2
229.1
112.4
124.7
162.6
2.5
157.8
20.5
133.3
468.5
62.0
17.8
114.7
198.0
57.3
146.9
19.6
%
9.2
2.3
21.9
9.0
16.5
19.8
17.2
18.9
19.5
10.7
2.2
17.4
18.7
18.0
14.8
17.4
13.3
10.4
19.0
17.5
15.3
18.2
PFCs
Tons
Reduced
0.6
2.3
11.8
6.9
47.0
3.9
22.5
18.1
24.3
8.1
0.2
27.6
4.2
37.4
23.7
13.8
0.5
4.7
35.7
12.3
20.7
3.4
%
36.9
26.0
84.6
38.5
85.6
85.9
67.5
87.0
81.5
49.7
25.4
87.6
86.6
89.4
55.7
82.8
55.6
40.8
80.2
86.5
79.8
81.4
Gasoline Distribution
Tons
Reduced
0.6
3.3
11.3
46.3
14.5
4.3
41.3
28.1
47.4
26.6
0.4
8.0
3.0
26.7
100.6
9.4
0.5
12.3
22.8
22.1
8.6
4.4
%
16.7
8.0
43.0
12.9
40.0
46.0
46.0
46.0
40.0
36.6
8.0
40.0
46.0
46.0
29.2
37.0
40.0
29.2
40.0
40.0
36.4
37.0
Total
Tons
Reduced
327.4
885.1
896.0
2,620.1
1,982.3
305.6
2,619.6
876.0
1,970.6
2,160.6
150.4
984.6
294.9
1,290.9
2,848.0
906.7
249.7
1,559.2
3,104.8
517.7
1,593.5
325.7
%
35.7
29.9
43.8
36.1
35.0
48.0
43.1
35.2
43.7
37.9
34.1
32.8
45.8
38.1
28.9
42.6
41.6
34.9
44.4
41.1
43.0
46.2
2-86
-------
Final Regulatory Impact Analysis
Table 2.2-55. Impact of Controls on Total Benzene Emissions From Mobile Sources and All Sources.
2015
Fuel Benzene Control
Vehicle Control
Fuel, Vehicle, and PFC Control
2020
Fuel Benzene Control
Vehicle Control
Fuel, Vehicle, and PFC Control
2030
Fuel Benzene Control
Vehicle Control
Fuel, Vehicle, and PFC Control
Mobile Source
and PFC Tons
Reduced
16,804
16,570
32,396
16,768
27,097
41 ,91 1
18,796
45,037
60,186
Mobile Source
and PFC Tons,
Reference Case
140,124
140,124
140,124
142,738
142,738
142,738
163,220
163,220
163,220
% of Mobile
Source and
PFC Tons
Reduced
12
12
23
12
19
29
12
28
37
Total Tons
Reduced From
All Sources
17,614
16,570
33,206
17,617
27,097
42,760
19,645
45,037
61,035
Total Mobile
and Stationary
Tons, Reference
Case
256,755
256,755
256.755
262,828
262,828
262,828
283,310
283,310
283,310
% of Total
Benzene
Reduced
7
7
13
7
10
16
7
16
22
2-87
-------
Final Regulatory Impact Analysis
Table 2.2.-56. Cumulative Impact of Controls on Total MSAT Emissions From Mobile Sources and PFCs, and From All Sources.
2015
2020
2030
Mobile
Source and
RFC Tons
Reduced
140,726
210,173
330,713
Mobile Source
and RFC Tons,
Reference
Case
1,215,146
1,193,281
1,318,746
% of Mobile
and RFC
Tons
Reduced
12
18
25
Total Tons
Reduced
141,536
211,022
331,562
Total Mobile
and
Stationary
Tons
2,636,063
2,733,020
2,858,485
% of Mobile and
Stationary Tons
Reduced
5
8
12
2-88
-------
Final Regulatory Impact Analysis
2.3 Potential Implications of New Emissions Data for Inventories
2.3.1 Newer Technology Light Duty Vehicles
MOBILE6.2 explicitly estimates emissions for the following air toxic compounds:
benzene, 1,3-butadiene, formaldehyde, acetaldehyde, MTBE, and acrolein.60'61
MOBILE6.2 estimates air toxics emission factors by multiplying an air toxic to VOC
(volatile organic compound) ratio by MOBILE6.2 VOC. For light-duty gasoline vehicles
and trucks, the product for exhaust emissions is then multiplied by an off-cycle
adjustment factor, which accounts for the difference in toxic fractions between Federal
Test Procedure (FTP) and Unified Cycle (UC) operation.
Toxic to VOC ratios vary by technology group, vehicle type, whether a vehicle is
a normal or high emitter (same definition as MOBILE6.2), and fuel characteristics.
Evaporative toxic/VOC ratios do not vary among gasoline vehicle classes. Since toxic
emission rates are a product of toxic/VOC emission ratios and VOC emission rates,
anything that reduces VOC will also result in toxic emission reductions. Toxic/VOC
ratios for individual technology group/vehicle type/emitter class combinations are
determined using a series of algorithms which calculate the ratios based on fuel
parameter inputs. These algorithms were derived from tests on 1990 model year
technology vehicles and form the basis of the Complex Model for Reformulated
Gasoline. MOBILE6.2 assumes that the same ratios are applicable to all post-1990
technology vehicles, including advanced technology low emission vehicles (LEVs)
meeting Tier 2 standards.62
Eastern Research Group, under contract to EPA, recently compared exhaust
emissions data from newer technology vehicles to see if the toxics to VOC fractions
estimated from these data were statistically different from ratios predicted by
MOBILE6.2. To make these comparisons, we used data collected by EPA Office of
Research and Development/National Exposure Research Laboratory on 23 1998-2003
vehicles, the California Air Resources Board (46 vehicles) and Southwest Research
Institute (3 vehicles). The contractor report and the data used are available in the docket
for this rule.63 The data from EPA's Office of Research and Development have been
published.64
The conclusions from t-test comparisons were as follows:
1) When the off-cycle adjustment for benzene is factored out of the model results,
MOBILE6.2 predicts statistically higher toxic fractions than one gets from the
California Air Resources Board and Southwest Research Institute data, although
for the large California dataset, the difference is only 10%. The fractions from the
EPA Office of Research and Development data are higher than predicted by
MOBILE6.2, but the difference is not statistically significant.
2) MOBILE6.2 is over-predicting toxic fractions for 1,3-butadiene.
2-89
-------
Final Regulatory Impact Analysis
3) The available data do not support a conclusion that MOBILE6.2 underestimates
or overestimates fractions for MTBE, formaldehyde, acetaldehyde or acrolein.
There is a significant amount of scatter in the available test data, which makes it
difficult to draw strong conclusions from the statistical comparisons. Also data are very
limited for high emitters and off-cycle operation, which make a large contribution to total
emissions. Nonetheless, at this point it appears that MOBILE6.2 toxic to VOC fractions
for benzene, MTBE, formaldehyde, acetaldehyde, and acrolein are reasonably accurate
for newer technology vehicles, but that fractions used for 1,3-butadiene are
overestimating emissions for this pollutant.
The recent Energy Policy Act passed by Congress requires EPA to develop a new
fuel effects model that reflects a 2007 fleet. The collection of a large amount of data and
substantial analytical work is needed to meet this requirement, and to update the
algorithms used in the current Complex Model and MOBILE6.2. As part of this ongoing
effort, EPA is reviewing engine exhaust data, which includes air toxic emissions, from
the CRC (Coordinating Research Council) E-67 study on engine emissions from light-
duty vehicles using ethanol fuels.65 Likewise, work is underway in a collaborative test
program between EPA and members of the Alliance of Automobile Manufacturers to
examine emissions of both regulated pollutants and air toxics from Tier 2 compliant
vehicles. The current program focuses on changes in fuel sulfur, vapor pressure, and
benzene levels, and will provide data for the air toxics rulemaking process as well as
inform the design of a more comprehensive program covering a wider range of fuel
properties and vehicle certification levels.
2.3.2 Heavy-Duty Vehicles (CRC E-55/E-59)
The primary objective of the E-55/59 research program was to quantify gaseous
and PM emissions from primarily in-use heavy-duty diesel trucks in California's South
Coast Air Basin, in support of emissions inventory development.66 A second program
objective was to quantify the influence of tampering and mal-maintenance on emissions
from these vehicles. The program was conducted in four Phases (denoted as 1, 1.5, 2 and
3). The Phase 1 test fleet consisted of 25 heavy heavy-duty diesel trucks (HHDDT),
selected to match a distribution of model years (MY) and to reflect engines in common
use in California. In Phase 1.5 an additional twelve HHDDT were studied, with a
thirteenth truck tested at idle alone. The Phase 2 test fleet consisted often HHDDT and
nine medium heavy-duty trucks (MHDT), which included seven diesel-fueled medium
heavy-duty trucks (MHDDT) and two gasoline-fueled medium heavy-duty trucks
(MHDGT). Phase 3 gathered data from nine HHDDT, eight MHDDT, and two MHDGT.
The Phase 2 and 3 data added post-2002 MY HHDDT (at 2.5 g/bhp-hr NOX standard) to
the program.
Sampling for chemical speciation was performed on thirteen HHDDT in Phase 1
and on five HHDDT and one MHDDT in Phase 2. However, only three of the thirteen
Phase 1 trucks had their exhaust samples analyzed for air toxic emissions, and the
remaining samples were being archived. Toxics species were measured from five
2-90
-------
Final Regulatory Impact Analysis
HHDDT and one MHDDT (medium HDDTs) in the Phase 2 test fleet. PM data were
acquired in Phases 1.5, 2 and 3. Exhaust data were acquired for methane and VOC.
Semi-volatile organic compounds and PM soluble fractions were captured and analyzed,
along with carbonyls and nitrosamines. Ions and elemental/organic carbon (EC/OC) split
were determined from quartz filters. The ion and metal analyses varied widely between
trucks.
These data will be incorporated into EPA's MS AT inventories, and will help
address limitations discussed in Sections 2.1.4 and 2.2.1.1.5.
2.3.3 Small Spark Ignition Engines
The National Mobile Inventory Model (NMIM) calculates air toxic emissions for
small Spark Ignition (SI) engines by multiplying compound-specific fractions with
volatile organic carbon (VOC) or particulate matter (PM) emission outputs from EPA's
NONROAD model. These fractions were used in the 1999 National Air Toxics
Assessment (NATA). These data were all obtained from a small number of uncontrolled
engines.67'68'69'70'71 In fiscal year 2004 EPA tested a mixture of in-use and new pre-
control and Phase 1 small hand held SI trimmers, chain saws and a leaf blower.72 In the
same time period EPA performed engine tests on Phase I residential four-stroke lawn
mowers. The emission data from both programs may impact future versions of NMIM
and the inventories it calculates.
EPA tested four pre-control, nine Phase 1, two California-certified, and eight
Phase 2 handheld engines. Five of the Phase 2 engines were new. All tests were fueled
by either of two summer grades of gasoline. One was a gasoline ethanol blend meant to
represent a reformulated gasoline and the other a conventional gasoline. All but one of
the engines were two-cycle designs. However, the four-cycle engine was designed to
operate on a typical two-cycle fuel lubricating oil mixture. All the test engines require
that lubricating oil be mixed and consumed with the fuel. The program therefore used
two different types of lubricating oil, one a mineral-based product and the other a "low
smoke" synthetic. Both oils were commercially available. The testing was done over the
Composite Two Mode (C2M) duty cycle. Table 2.3.-1 compares the emission factors
used in NONROAD and the fractions used in NMIM with those based on the testing.
NONROAD and NMIM have not been adjusted to use the new data, but some
increase in projected benzene inventories is likely once this occurs. In all but one engine
and fuel combination the benzene/VOC fraction is greater than that currently used in
NMEVI. It is significant that two-cycle engines have a large proportion of their fuel being
emitted in an unburned state. A reduction in fuel benzene content will have a significant
effect on benzene emissions from them.
The other MSAT fractions are found in Table 2.3.-2. Some of the measured
values are more consistent with NMIM values, but some are not (e.g., xylenes).
The second EPA test program involved six new Phase 2 four cycle lawn mower
engines. These data are unpublished. The engines were tested after 20 hours of
2-91
-------
Final Regulatory Impact Analysis
operation. The testing was done using the certification test procedure on certification
gasoline. Formaldehyde and acetaldehyde were the only MSATs measured in the test
program. A comparison of NMIM fractions and measured fractions are in Table 2.3.-3.
The measured values are similar to the values used in NMIM. Incorporation of
the new test data would not result in a dramatic change in inventories from these engines
and use types.
Table 2.3.-1. Comparison between NONROAD Outputs and NMIM MSAT
Fractions and Averaged Test Data for PM, VOC and Benzene from EPA Testing of
18 Handheld SI Engines Aggregated by Use, Engine Class, Emission Standard
(Phase), Catalyst, and Engine Cycle
Type
Class
Condition
Phase
Catalyst
Equipped
Engine
Cycle
NONROAD
PM10EF
(g/bhp)
Average
Tested
PM2.5
(g/bhp)
NONROAD
HCEF
(g/bhp)
Average
Tested
THC
(g/bhp)
NMIM
Benzene
Fraction
Average
Tested
Benzene
Fraction
BLOWER
CHAIN
SAW
CHAIN
SAW
CHAIN
SAW
CHAIN
SAW
CHAIN
SAW
STRING
TRIMMER
STRING
TRIMMER
STRING
TRIMMER
STRING
TRIMMER
STRING
TRIMMER
STRING
TRIMMER
V
IV
IV
IV
IV
V
III
III
IV
IV
IV
IV
New
New
Used
Used
Used
Used
Used
Used
New
New
Used
Used
2
2
0
1
2
1
0
1
2
2
0
1
YES
YES
NO
NO
NO
NO
NO
NO
YES
NO
NO
NO
2
2
2
2
2
2
2
2
2
4
2
2
7.70
7.70
9.24
9.93
9.93
9.75
9.24
9.55
7.70
0.06
9.24
9.93
0.028
0.228
3.072
2.051
1.483
1.330
4.915
7.519
0.641
0.231
3.093
3.856
40.15
26.87
313.20
231.84
42.66
152.00
313.20
272.79
26.87
25.83
313.20
231.84
24.842
30.254
185.976
110.567
98.066
80.026
265.205
243.167
31.581
12.791
221.354
154.140
0.024
0.080
0.080
0.080
0.080
0.080
0.011
0.011
0.011
0.011
0.011
0.011
0.038
0.022
0.016
0.014
0.014
0.016
0.019
0.013
0.028
N.A.
0.015
0.017
2-92
-------
Final Regulatory Impact Analysis
Table 2.S.-2. NMIM MSAT Fractions versus Fractions from EPA Testing of
18 Handheld SI Engines
Type Standard Fuel
BLOWER Ph2 CG
SAW CG
SAW Phi CG
SAW Phi RFG
SAW Ph2 CG
SAW Ph2 RFG
TPJMMER CG
TRIMMER RFG
TRIMMER Phi CG
TRIMMER Phi CG
TRIMMER Phi RFG
TRIMMER Phi RFG
TRIMMER Ph2 CG
TRIMMER Ph2 RFG
Formaldehyde
NMTM
0.0068
0.0068
0.0068
0.0068
0.0068
0.0029
0.0029
0.0029
0.0029
0.0029
0.0029
Tested
0.0050
0.0042
0.0053
0.0052
0.0056
0.0072
0.0077
0.0039
0.0045
0.0050
0.0080
Acetaldehyde
NMTM
0.0013
0.0013
0.0013
0.0013
0.0013
0.0006
0.0006
0.0006
0.0006
0.0006
0.0006
Tested
0.0011
0.0009
0.0046
0.0011
0.0055
0.0016
0.0066
0.0009
0.0046
0.0010
0.0073
Acrolein
NMTM
0.0004
0.0004
0.0004
0.0004
0.0004
0.0003
0.0003
0.0003
0.0003
0.0003
0.0003
Tested
0.0003
0.0003
0.0004
0.0004
0.0004
0.0006
0.0006
0.0003
0.0003
0.0003
0.0005
Propionaldehyde
NMTM
0.0001
0.0003
0.0004
0.0004
0.0004
0.0004
0.0004
0.0003
0.0004
0.0006
0.0009
Tested
0.0002
0.0002
0.0002
0.0002
0.0002
0.0002
0.0002
0.0002
0.0002
0.0002
0.0002
Toluene
NMTM
0.0978
0.0598
0.0598
0.0598
0.0598
0.0598
0.0978
0.0890
0.0978
0.0978
0.0890
0.0890
0.0978
0.0890
Tested
0.0979
0.0998
0.1064
0.1105
0.1065
0.0955
0.1049
0.0891
0.1093
0.1000
0.1096
0.0906
0.1303
0.1014
2,2,4-
Trimethypentane
NMTM
0.0372
0.0372
0.0372
0.0372
0.0372
0.0372
0.0372
0.0372
0.0372
0.0372
0.0372
0.0372
0.0372
0.0372
Tested
0.0122
0.0490
0.0487
0.0280
0.0409
0.0252
0.0437
0.0242
0.0432
0.0497
0.0249
0.0279
0.0559
0.0326
Xylene
NMTM
0.1075
0.0931
0.0931
0.0931
0.0931
0.0931
0.1075
0.0978
0.1075
0.1075
0.0978
0.0978
0.1075
0.0978
Tested
0.0224
0.0166
0.0151
0.0231
0.0177
0.0228
0.0174
0.0232
0.0204
0.0163
0.0299
0.0238
0.0205
0.0235
Table 2.3.-S. Comparison of NMIM Acetaldehyde and Formaldehyde to
VOC fractions with Measured Fractions from OTAQ Test Program
MSAT
Acetaldehyde
Formaldehyde
NMIM Fraction
0.00440
0.01256
Average Measured Fraction
0.00396
0.01541
2.3.4 Nonroad CI engines
The Agency conducted three separate emission test programs measuring exhaust
emissions from fifteen nonroad diesel engines and in-use pieces of nonroad diesel
equipment
73,74,75
The engines tested derived from construction, utility and agricultural
equipment applications for the most part and ranged from seven horsepower (hp) up
through 850 hp (425 hp, as tested). The test fuels used varied by sulfur concentration
from nonroad-grade diesel fuels at 2500 and 3300 ppm sulfur to a nominal "D-2" diesel
at 350 ppm sulfur and, lastly, to an ultra-low sulfur diesel, measured at less than 10 ppm
sulfur. Test engines were run over both steady-state and transient duty cycles. Several of
the transient cycles were application-specific, having been based on rubber-tire loader or
excavator operations, for example. Criteria pollutants in the exhaust emissions were
quantified for each test engine as well as sulfate, ammonia, N2O and a range of Ci - Cn
compounds (aldehydes, ketones, alcohols, etc.). Emissions of several additional air toxic
compounds were identified in two of the three programs. These emission species
included benzene, toluene, ethylbenzene, xylenes, polyaromatic hydrocarbons (PAH),
nitrated-PAHs and several metals. Emission results were summarized in both grams/hour
and grams/brake-horsepower/hour.
With this new emission data, the Agency has begun an effort to update the toxics
2-93
-------
Final Regulatory Impact Analysis
portion of its NMIM model. EPA will also address differences between Tier 1 and
unregulated NR diesel emissions, the impact of diesel fuel sulfur level on engine
emissions, whether any adjustments to default modeling TAFs (transient adjustment
factors) used in the NONROAD emissions model are warranted by the new data and the
necessity for creating category- and power-specific TAFs for NONROAD.
2.4 Description of Current Mobile Source Emissions Control Programs
that Reduce MSATs
As described above, existing mobile source control programs will reduce MSAT
emissions (not including diesel PM) by 60% between 1999 and 2020. Diesel PM from
mobile sources will be reduced by 75% between 2001 and 2030. The mobile source
programs include controls on fuels, highway vehicles, and nonroad equipment. These
programs are also reducing hydrocarbons and PM more generally, as well as oxides of
nitrogen. The sections immediately below provide general descriptions of these
programs, as well as voluntary programs to reduce mobile source emissions, such as the
National Clean Diesel Campaign and Best Workplaces for Commuters.SM
2.4.1 Fuels Programs
Several federal fuel programs reduce MSAT emissions. Some of these programs
directly control air toxics, such as the reformulated gasoline (RFG) program's benzene
content limit and required reduction in total toxics emissions, and the anti-backsliding
requirements of the anti-dumping and current MSAT programs, which require that
gasoline cannot get dirtier with respect to toxics emissions. Others, such as the gasoline
sulfur program, control toxics indirectly by reducing hydrocarbon and related toxics
emissions. Some fuel programs will have a mixed impact on the species and quantity of
MSAT emissions expected with the introduction of these new fuels into commerce.
2.4.1.1 RFG
The RFG program contains two direct toxics control requirements. The first is a
fuel benzene standard, requiring RFG to average no greater than 0.95 volume percent
benzene annually (on a refinery or importer basis). The RFG benzene requirement
includes a per-gallon cap on fuel benzene level of 1.3 volume percent. In 1990, when the
Clean Air Act was amended to require reformulated gasoline, fuel benzene averaged 1.60
volume percent. For a variety of reasons, including other regulations, chemical product
prices and refining efficiencies, most refiners and importers have achieved significantly
greater reductions in benzene than required by the program. In 2003, RFG benzene
content averaged 0.62 percent. The RFG benzene requirement includes a per-gallon cap
on fuel benzene level of 1.3 volume percent.
The second RFG toxics control requires that RFG achieve a specific level of
toxics emissions reduction. The requirement has increased in stringency since the RFG
program began in 1995, when the requirement was that RFG annually achieve a 16.5%
reduction in total (exhaust plus evaporative) air toxics emissions. Currently, a 21.5%
2-94
-------
Final Regulatory Impact Analysis
reduction is required. These reductions are determined using the Complex Model. As
mentioned above, for a variety of reasons most regulated parties have overcomplied with
the required toxics emissions reductions. During the 1998-2000 timeframe, RFG
achieved, on average, a 27.5% reduction in toxics emissions.
2.4.1.2 Anti-dumping
The anti-dumping regulations were intended to prevent the dumping of "dirty"
gasoline components, which were removed to produce RFG, into conventional gasoline
(CG). Since the dumping of "dirty" gasoline components, for example, benzene or
benzene-containing blending streams, would show up as increases in toxics emissions,
the anti-dumping regulations require that a refiner's or importer's CG be no more
polluting with respect to toxics emissions than the refiner's or importer's 1990 gasoline.
The anti-dumping program considers only exhaust toxics emissions and does not include
evaporative emissions.0 Refiners and importers have either a unique individual anti-
dumping baseline or they have the statutory anti-dumping baseline if they did not fulfill
the minimum requirements for developing a unique individual baseline. In 1990, average
exhaust toxics emissions (as estimated by EPA's Complex Model) were 104.5 mg/mileD;
in 2004, CG exhaust toxics emissions averaged 90.7 mg/mile. Although CG has no
benzene limit, benzene levels have declined significantly from the 1990 level of 1.6
volume percent to 1.1 volume percent for CG in 2004.
2.4.1.3 2001 Mobile Source Air Toxics Rule (MSAT1)
As discussed above, both RFG and CG have, on average, exceeded their
respective toxics control requirements. In 2001, EPA issued a mobile source air toxics
rule (MS ATI, for the purposes of this second proposal), as discussed in section ID. The
intent of MS ATI is to prevent refiners and importers from backsliding from the toxics
performance that was being achieved by RFG and CG. In order to lock in superior levels
of control, the rule requires that the annual average toxics performance of gasoline must
be at least as clean as the average performance of the gasoline produced or imported
during the three-year period 1998-2000. The period 1998-2000 is called the baseline
period. Toxics performance is determined separately for RFG and CG, in the same
manner as the toxics determinations required by the RFG76 and anti-dumping rules.
Like the anti-dumping provisions, MS ATI utilizes an individual baseline against
which compliance is determined. The average 1998-2000 toxics performance level, or
baseline, is determined separately for each refinery and importer.E To establish a unique
individual MS ATI baseline, EPA requires each refiner and importer to submit
documentation supporting the determination of the baseline. Most refiners and many
cSee RFG rule for why evaporative emissions are not included in the anti-dumping toxics
determination.
DPhase II
EExcept for those who comply with the anti-dumping requirements for conventional gasoline on an
aggregate basis, in which case the MS ATI requirements for conventional gasoline must be met on the same
aggregate basis (40 CFR Part 80, Subpart E).
2-95
-------
Final Regulatory Impact Analysis
importers in business during the baseline period had sufficient data to establish an
individual baseline. An MSAT1 baseline volume is associated with each unique
individual baseline value. The MS ATI baseline volume reflects the average annual
volume of such gasoline produced or imported during the baseline period. Refiners and
importers who did not have sufficient refinery production or imports during 1998-2000 to
establish a unique individual MSAT1 baseline must use the default baseline provided in
the rule.
The MS ATI program began with the annual averaging period beginning January
1, 2002. Since then, the toxics performance for RFG has improved from a baseline
period average of 27.5% reduction to 29.5% reduction in 2003. Likewise, CG toxics
emissions have decreased from an average of 95 mg/mile during 1998-2000 to 90.7
mg/milein2003.
2.4.1.4 Gasoline Sulfur
Beginning in 2006, EPA's gasoline sulfur program77 requires that sulfur levels in
gasoline can be no higher than 80 ppm as a per gallon cap and must average 30 ppm
annually. When the program is fully effective, gasoline will have 90 percent less sulfur
than before the program. Reduced sulfur levels are necessary to ensure that vehicle
emission control systems are not impaired. These systems effectively reduce non-
methane organic gas (NMOG) emissions, of which some are air toxics as well as
emissions of NOx. With lower sulfur levels, emission control technologies can work
longer and more efficiently. Both new and older vehicles benefit from reduced gasoline
sulfur levels.
2.4.1.5 Gasoline Volatility
A fuel's volatility defines its evaporation characteristics. A gasoline's volatility is
commonly referred to as its Reid vapor pressure, or RVP. Gasoline summertime RVP
ranges from about 6 to 9 psi and wintertime RVP, when additional volatility is required
for starting in cold temperatures, ranges from about 9 to 14 psi. Gasoline vapors contain a
subset of the liquid gasoline components and thus can contain toxics compounds, such as
benzene. Since 1989, EPA has controlled summertime gasoline RVP primarily as a VOC
and ozone precursor control, resulting in additional toxics pollutant reductions.
2.4.1.6 Diesel Fuel
In early 2001, EPA issued rules requiring that diesel fuel for use in highway
vehicles contain no more than 15 ppm sulfur beginning June 1, 2006. 78 This program
contains averaging, banking and trading provisions during the transition to the 15 ppm
level, as well as other compliance flexibilities. In June 2004, EPA issued rules governing
the sulfur content of diesel fuel used in nonroad diesel engines.79 In the nonroad rule,
sulfur levels are limited to a maximum of 500 ppm sulfur beginning in 2007 (current
levels are approximately 3000 ppm). In 2010, nonroad diesel sulfur levels must not
exceed 15 ppm.
2-96
-------
Final Regulatory Impact Analysis
EPA's diesel fuel requirements are part of a comprehensive program to combine
engine and fuel controls to achieve the greatest emission reductions. The diesel fuel
provisions will enable the use of advanced emission-control technologies on diesel
vehicles and engines. The diesel fuel requirements will also provide immediate public
health benefits by reducing PM emissions from current diesel vehicles and engines.
2.4.1.7 Phase-Out of Lead in Gasoline
One of the first programs to control toxic emissions from motor vehicles was the
removal of lead from gasoline. Beginning in the mid-1970s, unleaded gasoline was
phased in to replace leaded gasoline. The phase-out of leaded gasoline was completed
January 1, 1996, when lead was banned from motor vehicle gasoline.
2.4.2 Highway Vehicle and Engine Programs
The 1990 Clean Air Act Amendments set specific emission standards for
hydrocarbons and for PM. Air toxics are present in both of these pollutant categories. As
vehicle manufacturers develop technologies to comply with the hydrocarbon (HC) and
particulate standards (e.g., more efficient catalytic converters), air toxics are reduced as
well. Since 1990, we have developed a number of programs to address exhaust and
evaporative hydrocarbon emissions and PM emissions. Table 2.4-1 shows current mobile
source programs for highway vehicles.
Two of our recent initiatives to control emissions from motor vehicles and their
fuels are the Tier 2 control program for light-duty vehicles and the 2007 heavy-duty
engine rule. Together these two initiatives define a set of comprehensive standards for
light-duty and heavy-duty motor vehicles and their fuels. In both of these initiatives, we
treat vehicles and fuels as a system. The Tier 2 control program establishes stringent
tailpipe and evaporative emission standards for light-duty vehicles and a reduction in
sulfur levels in gasoline fuel beginning in 2004.80 The 2007 heavy-duty engine rule
establishes stringent exhaust emission standards for new heavy-duty engines and vehicles
for the 2007 model year as well as reductions in diesel fuel sulfur levels starting in
2006.81 Both of these programs will provide substantial emissions reductions through the
application of advanced technologies. We expect 90% reductions in PM from new diesel
engines compared to engines under current standards.
Some of the key earlier programs controlling highway vehicle and engine
emissions are the Tier 1 and NLEV standards for light-duty vehicles and trucks;
enhanced evaporative emissions standards; the supplemental federal test procedures
(SFTP); urban bus standards; and heavy-duty diesel and gasoline standards for the
2004/2005 time frame.
2-97
-------
Final Regulatory Impact Analysis
Table 2.4-1. Current On-Highway Engine and Vehicle Programs Providing
Significant Additional MSAT Reductions.
Category
Light-duty cars and
trucks
Heavy-duty trucks
Urban Buses
Highway
motorcycles
Rule & FRM Date
Tier 2 (including low sulfur fuel
and enhanced evaporative
emissions regulations)
February 10, 2000
NLEV (National Low-Emitting
Vehicle)
SFTP (Supplemental FTP)
Procedures
2004 Heavy-duty Rule
October 6, 2000
2007 Heavy-duty Rule (including
low sulfur fuel) January 18, 2001
HD Diesel Retrofit
December 2003
Implementation
Schedule
2004 - 2009
1999-2003
2001 (start)
2004 - 2007
2007-2010
1994- 1998
2006-2010
voc
Standards*
X
X
X
X
X
X
PM
Standards
X
X
X
X
X
* Standards in various forms including HC, NMHC, NMOG, and NOx+NMHC
2-98
-------
Final Regulatory Impact Analysis
Table 2.4-2 Current Nonroad Engine/Vehicle Programs.
Category
^and-based diesel
^ocomotives
Vlarine
Large spark-
ignition engines
Small spark-
ignition engines
Aircraft
(NOx Std in 2005;
Smoke Std in 1982)
Recreational
vehicles
Rule & FRM Date
Tier 2, October 23, 1998
Tier 3, October 23, 1998
Tier 4 (w/ low sulfur fuel)
June 29, 2004
Tier 0, Tier 1, Tier 2
April 16, 1998
Spark-ignition Gasoline
Engine Standards,
October 4, 1996
Diesel Engines, less than 50hp
Recreational Diesel,
November 8,2002
Commercial Diesel,
February 28, 2003
Tier 1 Standards
Tier 2 Standards
November 8,2002
Phase 1 Standards,
Handheld Phase 2 Standards,
April 25, 2000
Non-handheld Phase 2
Standards, March 30, 1999
November 8, 2002
Implementation
Schedule
2001-2006
2006-2008
2008-2014
2002 - 2005
1998-2006
1999-2005
Starting 2006/2009
Starting 2004/2007
2004 - 2006
2007, and later
1997-2007
2002 - 2007
2001 -2007
No current/recent
standards for VOC or
PM
2006-2012
VOC
Standards*
X
X
X
X
X
X
X
X
X
X
X
PM
Standards
X
X
X
X
X
X
X
* Standards in various forms including HC, NMHC, NMOG, and NOx+NMHC
2.4.3 Nonroad Engine Programs
There are various categories of nonroad engines, including land-based diesel
engines (e.g., farm and construction equipment), small land-based spark-ignition (SI)
engines (e.g., lawn and garden equipment, string trimmers), large land-based SI engines
(e.g., forklifts, airport ground service equipment), marine engines (including diesel and
SI, propulsion and auxiliary, commercial and recreational), locomotives, aircraft, and
recreational vehicles (off-road motorcycles, "all terrain" vehicles and snowmobiles).
Table 2.4-2 shows current mobile source programs for nonroad engines. Brief summaries
2-99
-------
Final Regulatory Impact Analysis
of our current and anticipated programs for these nonroad categories follow. As with
highway vehicles, the VOC standards we have established for nonroad engines will also
significantly reduce VOC-based toxics from nonroad engines. In addition, the standards
for diesel engines (in combination with the stringent sulfur controls on nonroad diesel
fuel) will significantly reduce diesel PM and exhaust organic gases, which are mobile
source air toxics.
In addition to the engine-based emission control programs described below, fuel
controls will also reduce emissions of air toxics from nonroad engines. For example,
restrictions on gasoline formulation (the removal of lead, limits on gasoline volatility and
RFG) are projected to reduce nonroad MSAT emissions because most gasoline-fueled
nonroad vehicles are fueled with the same gasoline used in on-highway vehicles. An
exception to this is lead in aviation gasoline. Aviation gasoline, used in general (as
opposed to commercial) aviation, is a high octane fuel used in a relatively small number
of aircraft (those with piston engines). Such aircraft are generally used for personal
transportation, sightseeing, crop dusting and similar activities.
2.4.3.1 Land-based Diesel Engines
We recently finalized stringent new emissions standards for land-based nonroad
diesel engines, used in agricultural and construction equipment as well as many other
applications (although the standards do not apply to locomotive engines, mining
equipment or marine engines).82 These standards are similar in stringency to the 2007
highway diesel engine standards, and are likewise enabled by stringent controls on sulfur
levels is diesel fuel, as explained earlier in section 2.4.1.6. The new engine standards,
starting in 2008, will reduce PM from new 2008 nonroad diesel engines by about 95
percent compared to engines under today's standards. The fuels controls are scheduled to
begin in mid-2007.
2.4.3.2 Land-Based SI Engines
The category of land-based nonroad SI engines is comprised of a broad mix of
service and recreational equipment with engines which range from less than 10
horsepower to several hundreds of horsepower. Most of these engines have been subject
to one or more tiers of engine emission controls for some time, while others in the
category, such as recreational vehicles, are just coming under engine emission control
regulations in 2006.
2.4.3.2.1 Large Land-Based SI Engines
Since the MS ATI rule was published, we have also finalized emissions standards
OT
for SI engines above 25 hp used in commercial applications. Such engines are used in a
variety of industrial equipment such as forklifts, airport ground service equipment,
generators and compressors. The Tier 1 standards went into effect in 2004 and the Tier 2
standards will start in 2007, providing additional emissions reductions. These standards
2-100
-------
Final Regulatory Impact Analysis
will provide about a 90 percent reduction in HC emissions on average for new engines
versus Tier 1 controlled engines.
2.4.3.2.2 Recreational Vehicles
In 2006, new recreational vehicles, which include snowmobiles, off-road
motorcycles and "all terrain vehicles", began a first tier of engine emission standards.
These standards require significant reductions in HC emissions from new engines,
ranging from 50 to 86 percent compared to pre-controlled engines.84
2.4.3.2.3 Small Land-B ased SI Engines
Small land-based spark-ignition (small SI) engines at or below 25 hp may be
either handheld or non-handheld and are used primarily in lawn and garden equipment
such as walk-behind and tractor mowers, string trimmers, chain saws and other similar
equipment. Our Phase 1 exhaust emission controls for this category of engines took
effect beginning in 1997 and are projected to result in a roughly 32 percent reduction in
VOC emissions for new engines, on average, versus pre-controlled engines.85 We also
have Phase 2 regulations for these engines which, when fully phased-in, are projected to
result in additional combined HC and NOx exhaust emission reductions beyond the Phase
1 levels of 60 percent for new non-handheld engines and of 70 percent for new handheld
engines.86 We are currently developing a proposal for new combined HC and NOx
exhaust standards for Phase 3 non-handheld small SI engines that should be
approximately 35 to 40 percent lower than present Phase 2 standards for this class of
engine. Further, we also expect to propose new evaporative emission standards for small
SI engines and equipment to control fuel hose permeation, fuel tank permeation, diurnal
and running loss emissions.
Phase 3 standards for Small SI engines are expected to achieve toxics benefits
through reduction of engine VOC emissions from three sources. The new standards
would result in fewer evaporative VOC and, therefore, air toxics emissions by lowering
hose and tank permeation losses for these types of equipment. Phase 3 engines will also
have lower exhaust VOC emissions under these new standards. Finally, Phase 3 Small SI
engines are expected to achieve a small fuel economy benefit during operation. While
small, VOC emission savings from increased fuel economy will feed back through a
reduced number of gallons of fuel kept onboard these engines during operation. This will
result in less VOC from tank/hose permeation, and less fuel burned overall will mean
fewer exhaust emissions
2.4.3.3 Marine Engines
Marine engines cover a very wide range of products, from 10-horsepower
outboard engines to 100,000-horsepower engines on oceangoing vessels. We have active
emission-control programs to address the need for emission controls for every kind of
marine engine.
2-101
-------
Final Regulatory Impact Analysis
2.4.3.3.1 Marine SI Engines
For gasoline-fueled outboard and personal-watercraft engines, we adopted an
initial phase of exhaust standards which became fully implemented with the 2006 model
year. These standards have led to a major technology shift in this category of engines to
four-stroke engines and advanced-technology two-stroke engines, for an estimated 75
percent reduction in HC emissions from uncontrolled levels.87 We are developing a
proposal to adopt new, more stringent exhaust standards for these engines that would
further reduce emissions from this class of engines by an additional 60 percent or more
from the initial phase of standards .
Another class of marine SI engine, referred to as stern drive and inboard engines,
uses automotive-type engines. These engines have uncontrolled emission rates that are
well below the current standards that apply to outboard and personal-watercraft engines.
These engines are not currently subject to emission standards, but we intend to include
00
new emission standards for these engines in an upcoming marine SI engine proposal.
These new standards would likely be based on the application of catalyst technology to
substantially reduce HC and NOx emissions from the operation of these engines.
The proposals mentioned above will also cover fuel evaporative emission
standards for fuel lines, fuel tanks and diurnal venting emissions for vessels powered by
gasoline-fueled engines in both of these engine classes.
2.4.3.3.2 Marine Diesel Engines
We have adopted emission standards for marine diesel engines in four separate
rulemakings. All of these standards are based on in-engine controls and do not require
aftertreatment. First, we adopted two tiers of standards for marine engines below 50
horsepower that apply equally to land-based and marine engines. These standards were
phased in from 1999 to 2005. Second, we adopted emission standards for commercial
marine diesel engines with per-cylinder engine displacement up to 30 liters. These
standards are comparable to the standards for land-based nonroad diesel engines that
apply in the same time frame, with several adjustments to test procedures and compliance
provisions appropriate for marine engines.89 The emission standards generally apply in
2007 for locomotive-size engines and in 2004 for smaller engines. Third, the emission
standards adopted for recreational marine diesel engines are very similar to the
comparable commercial engines, with implementation scheduled two years after the
commercial standards take effect. All the emission standards in these three rulemakings
targeted reductions in NOx and PM emissions. Finally, we adopted standards to control
NOx emissions at levels consistent with the requirements from the International Maritime
Organization (IMO), but we adopted these as EPA standards under the Clean Air Act to
make them mandatory for all engines with per-cylinder displacement above 2.5 liters
installed on U.S.-flag vessels starting in the 2004 model year. We are in the process of
reviewing the emission standards for all sizes of marine diesel engines and expect to
propose new requirements in the near future.
2-102
-------
Final Regulatory Impact Analysis
EPA is also investigating the possibility of designating U.S. coastal areas as SOx
(oxides of sulfur) Emission Control Areas (SECAs) under the IMO. Such a designation
would trigger a requirement for any vessel entering such an area to use reduced-sulfur
fuel or operate exhaust scrubbers to prevent SOx emissions.
2.4.3.4 Locomotives
Our regulations for locomotive engines consist of three tiers of standards,
applicable depending on the date a locomotive or a particular engine was originally
manufactured.90 The first set of standards (Tier 0) applies to locomotives and their
locomotive engines originally manufactured from 1973 through 2001, starting from the
time the engine was manufactured or later at "remanufacture."F The second set of
standards (Tier 1) applies to locomotives and their engines manufactured from 2002
through 2004 and again at engine manufacture or rebuild. The third set of standards (Tier
2) applies to locomotive engines manufactured in 2005 and later. The Tier 0 and Tier 1
regulations were primarily intended to reduce NOx emissions. The Tier 2 regulations are
projected to result in 50 percent reductions in VOC and diesel PM as compared to
unregulated engine emission levels, as well as additional NOx reductions beyond the Tier
0 and Tier 1 regulations. We are currently developing a new tier of more stringent
emissions standards for locomotive engines.
2.4.3.5 Aircraft
A variety of emission regulations have been applied to commercial gas turbine
aircraft engines, beginning with limits on smoke and fuel venting in 1974. In 1984, limits
were placed on the amount of unburned HC that gas turbine engines can emit per landing
and takeoff cycle. In 1997, we adopted standards that were equivalent to the existing
International Civil Aviation Organization (ICAO) NOx and CO emission standards for
gas turbine engines. In 2005, we tightened the NOx emission standards to levels that are
equivalent to the ICAO standards that became effective in 2004. These actions have
resulted in minimal emissions reductions, and have largely served to prevent increases in
aircraft emissions. We continue to explore ways to reduce emissions from aircraft
throughout the nation.
2.4.4 Voluntary Programs
In addition to the fuel and engine control programs described above, we are
actively promoting several voluntary programs to reduce emissions from mobile sources,
such as the National Clean Diesel Campaign, anti-idling measures, and Best Workplaces
for Commuters. While the stringent emissions standards described above apply to new
highway and nonroad diesel engines, it is also important to reduce emissions from the
existing fleet of about 11 million diesel engines. EPA has launched a comprehensive
initiative called the National Clean Diesel Campaign, one component of which is to
promote the reduction of emissions in the existing fleet of engines through a variety of
F To "remanufacture" an engine is to rebuild that engine to "new condition" at the end of four-to-
eight year long maintenance cycles.
2-103
-------
Final Regulatory Impact Analysis
cost-effective and innovative strategies. The goal of the Campaign is to reduce emissions
from the 11 million existing engines by 2014. Emission reduction strategies include
switching to cleaner fuels, retrofitting engines through the addition of emission control
devices and engine replacement. For example, installing a diesel paniculate filter
achieves diesel particulate matter reductions of approximately 90 percent (when
combined with the use of ultra low sulfur diesel fuel). The Energy Policy Act of 2005
includes grant authorizations and other incentives to help facilitate voluntary clean diesel
actions nationwide.
The National Clean Diesel Campaign is focused on leveraging local, state, and
federal resources to retrofit or replace diesel engines, adopt best practices and track and
report results. The Campaign targets five key sectors: school buses, ports, construction,
freight and agriculture. Almost 300 clean diesel projects have been initiated through the
Campaign. These projects will reduce more than 20,000 PM lifetime tons. PM and NOx
reductions from these programs will provide nearly $5 billion in health benefits.
Reducing vehicle idling provides important environmental benefits. As a part of
their daily routine, truck drivers often keep their vehicles running at idle during stops to
provide power, heat and air conditioning. EPA's SmartWay Transport Partnership is
helping the freight industry to adopt innovative idle reduction technologies and to take
advantage of proven systems that provide drivers with basic necessities without idling the
main engine. To date, there are 80 mobile and stationary idle-reduction projects
throughout the country. Emission reductions, on an annual basis, from these programs
are in excess of 157,000 tons of CO2, 2,000 tons of NOX and 60 tons of PM; over 14
million gallons of fuel are being saved annually. The SmartWay Transport Partnership
also works with the freight industry by promoting a wide range of new technologies such
as advanced aerodynamics, single-wide tires, weight reduction, speed control and
intermodal shipping.
Daily commuting represents another significant source of emissions from motor
vehicles. EPA's Best Workplaces for CommutersSM program is working with employers
across the country to reverse the trend of longer, single-occupancy vehicle commuting.
OTAQ recognizes employers that have met the National Standard of Excellence for
Commuter Benefits by adding them to the List of Best Workplaces for Commuters™.
These companies offer superior commuter benefits such as transit subsidies for rail, bus,
and vanpools and promote flexi-place and telework. Emergency Ride Home programs
provide a safety net for participants. More than 1,600 employers representing 3.5 million
U.S. workers have been designated Best Workplaces for Commuters™.
Much of the growth in the Best Workplaces for Commuters™ program has been through
metro area-wide campaigns. Since 2002, EPA has worked with coalitions in over 14
major metropolitan areas to increase the penetration of commuter benefits in the
marketplace and the visibility of the companies that have received this distinguished
designation. Another significant path by which the program has grown is through
Commuter Districts including corporate and industrial business parks, shopping malls,
business improvement districts and downtown commercial areas. To date EPA has
2-104
-------
Final Regulatory Impact Analysis
granted the Best Workplaces for Commuters™ "District" designation to over twenty
locations across the country including sites in downtown Denver, Houston, Minneapolis,
Tampa, and Boulder.
2-105
-------
Final Regulatory Impact Analysis
References for Chapter 2
1 Federal Register. Regulation of Fuels and Fuel Additives: Renewable Fuel Standard
Program; Proposed Rule, 70(184): Page 55551-55600, September 22, 2006.
2 Renewable Fuel Standard Program, Draft Regulatory Impact Analysis, EPA420-D-06-
008, September 2006. http://www.epa.gov/otaq/renewablefuels/index.htm.
3 http://www.epa.gov/otaq/nonrdmdl.htm#model
4 http://www.epa.gov/otaq/models/mobile6/m6tech.htm
5 Mullen, M., Neumann, J. Technical Memorandum: Documentation of 2003 VMT
Projection Methodology, Prepared by E. H. Pechan and Associates and Industrial
Economics, Inc. for James DeMocker, Office of Air and Radiation, Office of Policy
Analysis and Review, U. S. EPA, Contract No. 68-W-02-048, WA B-41, March 2004.
Included as Appendix G: Clean Air Interstate Rule: Emission Inventory Technical
Support Document, http://www.epa.gov/cair/technical.html. This document is available
in Docket EPA-HQ-OAR-2003-0053.
6 "VOC/PM Cold Temperature Characterization and Interior Climate Control
Emission/Fuel Economy Impact", Alan P. Stanard , Southwest Research Institute Project
No. 03.11382.01, October 2005
7 Brzezinski, D.; Fernandez; A. (2005) Cold Temperature Effects on Vehicle HC
Emissions. U. S. EPA, Assessment and Standards Division, National Vehicle and Fuel
Emissions Laboratory, Ann Arbor, MI, Report No. EPA420-D-06-001.
8 http://www.epa.gov/otaq/models/mobile6/m6tech.htm
9 Landman, L. C. (2007) Estimating Emissions Associated with Portable Fuel Containers
(PFCs). U. S. EPA, Assessment and Standards Division, National Vehicle and Fuel
Emissions Laboratory, Ann Arbor, MI, Report No. EPA420-R-07-001. This document is
available in Docket EPA-HQ-OAR-2003-0053.
10 Watson,!., Fujita,E., Chow,J., Zielinska,B., Richards,L., Neff,W., Dietrich,D. Northern
Front Range Air Quality Study Final Report: Volume 1. June 30, 1998. Colorado State
University. This document is available in the EPA Docket as EPA-HQ-OAR-2005-0036-
1179.
11 Stanard, A.P. (2005) VOC/PM cold temperature characterization and interior climate
control emissions/fuel economy impact. Draft final report volume I. EPA contract 68-C-
06-018. Work assignment No. 0-4. Southwest Research Institute project No.
03.11382.04.
2-106
-------
Final Regulatory Impact Analysis
12 Bailey, C.R. (2005) Cold-temperature exhaust particulate matter emissions.
Memorandum to Docket EPA-HQ-OAR-2005-0036.
13 Energy Information Agency, Annual Energy Outlook 2006, Table 17.
http://www.eia.doe.gov/oiaf/forecasting.html
14 Pollack, A. K.; Lindhjem, C.; Stoeckenius, T. E.; Tran, C.; Mansell, G.; Jiminez, M.;
Wilson, G.; Coulter-Burke, S. 2004. CRC Project E-64: Evaluation of the U. S. EPA
MOBILE6 Highway Vehicle Emission Factor Model. Prepared by Environ Corp. for
Coordinating Research Council and U. S. EPA. March, 2004.
http://www.epa.gov/otaq/m6.htmtfm60
15 National Research Council. 2000. Modeling Mobile Source Emissions. National
Academy Press; Washington, D.C. http://www.nap.edu/books/0309070880/html/
16 National Research Council. 2000. Modeling Mobile Source Emissions. National
Academy Press; Washington, D.C. http://www.nap.edu/books/0309070880/html/.
17 Strum, M., Cook, R., Pope, A., Palma, T., Shedd, S., Mason, R., Michaels, H.,
Thurman, J., Ensley, D. 2006. Projection of Hazardous Air Pollutant Emissions to
Future Years. Science of the Total Environment, 366: 590-601.
18 Michaels, H., Brzezinski, D., Cook, R.. EPA's National Mobile Inventory Model
(NMIM), A Consolidated Emissions Modeling System for MOBILE6 and NONROAD.
U. S. Environmental Protection Agency, Office of Transportation and Air Quality,
Assessment and Standards Division, Ann Arbor, MI, March 2005; Report No. EPA-420-
R-05-003. http://www.epa.gov/otaq/nmim.htm. This document is available in Docket
EPA-HQ-OAR-2005-0036.
19 Cook, R., Glover, E., Michaels, H., Brzezinski, D. 2004. Modeling of Mobile Source
Air Toxic Emissions Using EPA's National Mobile Inventory Model. Proceedings, 2004
Emission Inventory Conference, Clearwater Beach, FL.
http://www.epa.gov/ttn/chief/conference/ei 13/index.html. This document is available in
Docket EPA-HQ-OAR-2005-0036.
20 U. S. EPA. NMEVI2005. http://www.epa.gov/otaq/nmim.htm
21 Stump, F., Tejada, S., Ray, W., Dropkin, D., Black, F., Snow, R., Crews, W., Siudak,
P., Davis, C., Baker L., Perry, N. 1989. The Influence of Ambient Temperature on
Tailpipe Emissions from 1984 to 1987 Model Year Light-Duty Gasoline Vehicles.
Atmospheric Environment 23: 307-320.
22 Stump, F., Tejeda, S., Ray, W., Dropkin, D., Black, F., Snow, R., Crews, W., Siudak,
P., Davis C., Carter, P. 1990. The Influence of Ambient Temperature on Tailpipe
2-107
-------
Final Regulatory Impact Analysis
Emissions from 1985-1987 Model Year Light-Duty Gasoline Vehicles — II. Atmospheric
Environment 24A: 2105-2112.
23 Stump, F., Tejeda, S., Dropkin, D. Loomis, C. Unpublished. Characterization of
Emissions from Malfunctioning Vehicles Fueled with Oxygenated Gasoline-MTBE Fuel
- Part I. U. S. EPA, National Exposure Research Laboratory, Office of Research and
Development. This document is available in Docket EPA-HQ-OAR-2005-0036.
24 Stump, F., Tejeda, S., Dropkin, D. Loomis, C., Park, C. Unpublished. Characterization
of Emissions from Malfunctioning Vehicles Fueled with Oxygenated Gasoline-Ethanol
(E10) Fuel - Part II. U. S. EPA, National Exposure Research Laboratory, Office of
Research and Development. Unpublished. This document is available in Docket EPA-
HQ-OAR-2005-0036.
25 Stump, F., Tejeda, S., Dropkin, D. Loomis, C., Park, C. Unpublished. Characterization
of Emissions from Malfunctioning Vehicles Fueled with Oxygenated Gasoline-Ethanol
(E10) Fuel - Part III. U. S. EPA, National Exposure Research Laboratory, Office of
Research and Development. This document is available in Docket EPA-HQ-OAR-2005-
0036.
26 Stanard, Alan P. 2005. VOC/PM Cold Temperature Characterization and Interior
Climate Control Emissions/Fuel Economy Impact. Prepared for U. S. EPA, Office of
Transportation and Air Quality, Ann Arbor, MI, by Southwest Research Institute.
Southwest Research Project No. 03.11382.04. This document is available in Docket
EPA-HQ-OAR-2005-0036.
27 U. S. EPA. 2004. 1999 National Emissions Inventory, Final Version 3.
http://www.epa.gov/ttn/chief/net/1999inventory. html.
28 MathPro. 1998. Costs of Alternative Sulfur Content Standards for Gasoline in PADD
IV. Final Report. Prepared for the National Petrochemical and Refiners Association.
December 30. This document is available in Docket EPA-HQ-OAR-2005-0036.
29 Mathpro, 1999. Costs of meeting 40 ppm Sulfur Content Standard for Gasoline in
PADDs 1-3, Via MOBIL and CD Tech Desulfurization Processes. Final Report.
Prepared for the American Petroleum Institute. February 26. This document is available
in Docket EPA-HQ-OAR-2005-0036.
30 MathPro. 1999. Analysis of California Phase 3 RFG Standards. Prepared for the
California Energy Commission. December 7. This document is available in Docket
EPA-HQ-OAR-2005-0036.
31 Mullen, M., Neumann, J. Technical Memorandum: Documentation of 2003 VMT
Projection Methodology, Prepared by E. H. Pechan and Associates and Industrial
2-108
-------
Final Regulatory Impact Analysis
Economics, Inc. for James DeMocker, Office of Air and Radiation, Office of Policy
Analysis and Review, U. S. EPA, Contract No. 68-W-02-048, WA B-41, March 2004.
Included as Appendix G: Clean Air Interstate Rule: Emission Inventory Technical
Support Document, http://www.epa.gov/cair/technical.html. This document is available
in Docket EPA-HQ-OAR-2003-0053.
32 Wyborny, Lester; Memorandum to the Docket; Effect of Benzene Control on Gasoline
Quality, February 22, 2006.
33 Fritz, S. 2000. Diesel Fuel Effects on Locomotive Exhaust Emissions. Prepared by
Southwest Research for the California Air Resources Board. This document is available
in Docket EPA-HQ-OAR-2003-0053.
34 Eastern Research Group and E. H. Pechan and Associates. 2005. Documentation for
Aircraft, Commercial Marine Vessels, Locomotive and Other Nonroad Components of
the National Emissions Inventory. Prepared for U. S. EPA, Office of Air Quality
Planning and Standards, http://www.epa.gov/ttn/chief/net/1999inventory.html. This
document is available in Docket EPA-HQ-OAR-2005-0036.
35 Federal Aviation Administration, 2004. Terminal Area Forecast System.
http://www.apo.data.faa.gov/main/taf.asp.
36 Hester, Charles. 2006. Review of Data on HAP Content in Gasoline. Memorandum
from MACTEC to Steve Shedd, U. S. EPA, March 23, 2006. This document is available
in Docket EPA-HQ-OAR-2003-0053.
37 Haskew, H. M.; Liberty, T. F.; McClement, D. 2004. Fuel Permeation from
Automotive Systems. Prepared for the Coordinating Research Council by Harold
Haskew and Associates and Automotive Testing Laboratories, Inc. September 2004.
CRC Project No. E-65. http://www.crcao.com. This document is available in Docket
EPA-HQ-OAR-2005-0036.
38 Hester, Charles. 2006. Review of Data on HAP Content in Gasoline. Memorandum
from MACTEC to Steve Shedd, U. S. EPA, March 23, 2006. This document is available
in Docket EPA-HQ-OAR-2003-0053.
39 U. S. EPA. 2006. Determination that Gasoline Distribution Stage 1 Area Source (GD
AS) Category Does Not Need to Be Regulated Under Section 112(c)6. Memoradum
from Stephen Shedd to Kent Hustvedt, May 9, 2006. This document is available in
Docket EPA-HQ-OAR-2005-0036.
40 U. S. EPA. 2005. Documentation for the Final 2002 Mobile National Emissions
Inventory. Prepared by U. S. EPA, Assessment and Standards Division and E. H. Pechan
and Associates for U. S. EPA, Office of Air Quality Planning and Standards. September
2005.
2-109
-------
Final Regulatory Impact Analysis
41 ERG, Inc. 2003. Documentation for the for the Final 1999 Nonpoint Area Source
National Emission Inventory for Hazardous Air Pollutants (Version 3). Prepared for U.
S. EPA, Office of Air Quality Planning and Standards, August 26, 2003.
http://www.epa.gov/ttn/chief/net/1999inventory.html. This document is available in
Docket EPA-HQ-OAR-2005-0036.
42 Driver, L. 2005. Memorandum to State/Local Agencies that submitted 2002 gasoline
distribution emissions requesting review/comments on EPA plans for final 2002 NEI.
ftp://ftp.epa.gov/EmisInventory/draftnei2002/nonpoint/gasoline_marketing/gasdistributio
ninstructions.pdf This document is available in Docket EPA-HQ-OAR-2005-0036.
43 Strum, M., Cook, R., Pope, A., Palma, T., Shedd, S., Mason, R., Michaels, H.,
Thurman, J., Ensley, D. 2006. Projection of Hazardous Air Pollutant Emissions to
Future Years. Science of the Total Environment., 366: 590-601.
44 ERG, Inc. 2003. Documentation for the Final 1999 Point Source National Emissions
Inventory for Hazardous Air Pollutants (Version 3). Prepared for U. S. Environmental
Protection Agency, Office of Air Quality Planning and Standards, July 2003.
http://www.epa.gov/ttn/chief/net/1999inventory.html. This document is available in
Docket EPA-HQ-OAR-2005-0036.
45 ERG, Inc. 2003. Documentation for the Final 1999 Nonpoint Source National
Emissions Inventory for Hazardous Air Pollutants (Version 3). Prepared for U. S.
Environmental Protection Agency, Office of Air Quality Planning and Standards, August
2003. http://www.epa.gov/ttn/chief/net/1999inventory.html. This document is available
in Docket EPA-HQ-OAR-2005-0036.
46 U. S. Environmental Protection Agency. User's Guide for the Emissions Modeling
System for Hazardous Air Pollutants (EMS-HAP) Version 3.0 EPA-454/B-03-006
August 2004 by: Madeleine Strum, Ph.D. U.S. EPA, Office of Air Quality Planning and
Standards Emissions, Monitoring and Analysis Division Research Triangle Park, NC
And, Under Contract to the U.S. This document is available in Docket EPA-HQ-OAR-
2005-0036 and at http://www.epa.gov/scram001/userg/other/emshapv3ug.pdf
47
Regional Economic Models, Inc. 2004. REMI Policy Insight, http://www.remi.com.
48 Fan W, Treyz F, Treyz G. An evolutionary new economic geography model. J
Regional Sci 2000 40: 671- 696.
49 Energy Information Adminstration. Annual Energy Outlook 2005 with Projections to
2025. 2005, Report No. DOE/EIA-0383. http://www.eia.doe.gov/oiaf/aeo/index.html.
2-110
-------
Final Regulatory Impact Analysis
50 U. S. Environmental Protection Agency. Clean Air Interstate Rule: Emissions
Inventory Technical Support Document, 2005b. Available from:
http://www.epa.gov/cair/technical.html. This document is available in Docket EPA-HQ-
OAR-2003-0053.
51 U. S. Environmental Protection Agency. SPECIATE, Version 3.2. Available at:
http://www.epa.gov/ttn/chief/software/speciate/index.html
52 Cook, R., Beidler, A., Touma, J., Strum, M. 2006. Preparing Highway Emissions
Inventories for Urban Scale Modeling: A Case Study in Philadelphia. Transportation
Research Part D: Transport and Environment, 11: 396-407.
53 U. S. EPA. 2001. National-Scale Air Toxics Assessment for 1996: Draft for EPA
Science Advisory Board Review. Report No. EPA-453/R-01-003 This document is
available in Docket EPA-HQ-OAR-2005-0036.
http://www.epa.gov/ttn/atw/sab/natareport.pdf
54 Taylor, M. Memorandum: Revised HAP Emission Factors for Stationary Combustion
Turbines, Prepared by Alpha-Gamma Technologies, Inc for Sims Roy, EPA OAQPS
ESD Combustion Group. August, 2003.Docket ID: OAR-2003-0189.
http://docket.epa.gov/edkpub/do/EDKStaffItemDetailView?objectId=090007d480271237
55 U. S. EPA. 2006. Regulatory Impact Analysis for the Review of the Particulate Matter
National Ambient Air Quality Standards. Office of Air Quality Planning and Standards,
Research Triangle Park, NC. http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_cr_ria.html. This
document is available in Docket EPA-HQ-OAR-2006-0834.
56 Glover, E.; Brzezinski, D. 2001. Soak Length Activity Factors for Start Emissions. U.
S. EPA, Office of Transportation and Air Quality, Ann Arbor, MI. Report No. EPA420-
R-01-011. http://www.epa.gov/otaq/models/mobile6/r01011 .pdf This document is
available in Docket EPA-HQ-OAR-2005-0036.
57 Enns, P.; Brzezinski, D. 2001. Comparison of Start Emissions in theLA92 and ST01
Test Cycles. U. S. EPA, Office of Transportation and Air Quality, Ann Arbor, MI.
Report No. EPA420-R-01 -025. http://www.epa.gov/otaq/models/mobile6/rO 1025.pdf
This document is available in Docket EPA-HQ-OAR-2005-0036.
58 Glover, E; Carey, P. 2001. Determination of Start Emissions as a Function of
Mileage and Soak Time for 1981-1993 Model Year Light-Duty Vehicles. U. S. EPA,
Office of Transportation and Air Quality, Ann Arbor, MI. Report No. EPA420-R-01-
058. http://www.epa.gov/otaq/models/mobile6/r01058.pdf. This document is available
in Docket EPA-HQ-OAR-2005-0036.
59 Glover, E.; Brzezinski, D. 2001. Exhaust Emission Temperature
2-111
-------
Final Regulatory Impact Analysis
Correction Factors for MOBILE6: Adjustments for Engine Start and Running LA4
Emissions for Gasoline Vehicles. U. S. EPA, Office of Transportation and Air Quality,
Ann Arbor, MI. Report No. EPA420-R-01-029.
http://www.epa.gov/otaq/models/mobile6/r01029.pdf. This document is available in
Docket EPA-HQ-OAR-2005-0036.
60 Cook, R.; Glover, E. "Technical Description of the Toxics Module for MOBILE6.2
and Guidance on Its Use for Emission Inventory Preparation", U.S. EPA, Office of
Transportation and Air Quality, Ann Arbor, MI, November 2002; Report No., EPA420-
R-02-011. http://www.epa.gov/otaq/m6.htm. This document is available in Docket
EPA-HQ-OAR-2005-0036.
61 MOBILE6 Vehicle Emissions Modeling Software, http://www.epa.gov/otaq/m6.htm.
62EPA. 1993. Final Regulatory Impact Analysis for Reformulated Gasoline. December
13, 1993. http://www.epa.gov/otaq/regs/fuels/rfg/. This document is available in Docket
EPA-HQ-OAR-2005-0036.
63 Weatherby, M. F., Fincher, S., DeFries, T. H., Kishan, S. 2005. Comparison of Toxic
to Hydrocarbon Ratios from In-Use New Technology Vehicles to Those Used in
MOBILE6.2. Prepared by Eastern Research Group, Austin, TX, for Rich Cook, Office of
Transportation and Air Quality, U. S. EPA. ERG Report No. ERG No.:
0136.04.001.001. This document is available in Docket EPA-HQ-OAR-2005-0036.
64 Baldauf, R. W., P. Gabele, W. Crews, R. Snow, and R. Cook. 2005. Criteria and Air
Toxic Emissions from In-Use Automobiles in the National Low-Emission Vehicle
Program. Journal of the Air and Waste Management Association., 55:1263 -1268.
65 Durbin,T., Miller,!.W., Younglove,T., Huai,T. and Cocker,K. "Effects of Ethanol and
Volatility Parameters on Exhaust Emissions: CRC Project No. E-67." January 30, 2006.
Coordinating Research Council # 06-VE-59596-E67.
66 Clark, N., Gautam, M., Wayne, W., Lyons, D., & Thompson, G., "California Heavy
Heavy-Duty Diesel Truck Emissions Characterization for Program E-55/59: (Draft) Final
Report" (Short Title: "E-55/59 All Phases") Nov. 11, 2005 West Virginia University
Research Corporation, Morgantown, WV. EPA document # 420-D-06-005.
67 Censullo, A. 1991. Development of Species Profiles for Selected Organic Emission
Sources. California Polytechnic State University. San Luis Obispo, CA. This document
is available in Docket EPA-HQ-OAR-2005-0036.
68 Gabele, Peter. 1997. Exhaust emissions from four-stroke lawn mower engines. J. Air
& Waste Manage. Assoc. 47:945-950.
2-112
-------
Final Regulatory Impact Analysis
69 Hare, C. T.; Carroll, J. N. 1993. Speciation of Organic Emissions to Study Fuel
Dependence of Small Engine Exhaust Photochemical Reactivity. Report for Advisory
Committee on Research, Southwest Research Institute, July 1993. This document is
available in Docket EPA-HQ-OAR-2005-0036.
70 Hare, C.T.; White, JJ. 1991. Toward the Environmentally Friendly Small Engine:
Fuel, Lubricant, and Emission Measurement Issues. SAE Paper No. 9 11222. This
document is available in Docket EPA-HQ-OAR-2005-0036.
71 Carroll, J. N. 1991. Emission Tests of In-use Small Utility Engines: Task III Report,
Nonroad Source Emission Factors Improvement. Prepared for U. S. EPA by Southwest
Research Institute. Report No. SwRI 3426-0006. This document is available in Docket
EPA-HQ-OAR-2005-0036.
72 "Characterization of Emissions from Small Hand-Held, In-Use 2- Cycle Engines",
Richard Snow and William Crews, BKI, Inc , EPA/600/X-4/191, October 2004. This
document is available in Docket EPA-HQ-OAR-2005-0036.
73 Starr, M. February, 2004 "Air Toxic Emissions from In-Use Nonroad Diesel
Equipment; Final Report" (SwRI-08.05004.04). Prepared for U. S. EPA, Office of
Transportation and Air Quality, Ann Arbor, MI by Southwest Research Institute. (EPA
document # 420-R-04-019). This document is available in Docket EPA-HQ-OAR-2005-
0036.
74 Starr, M. February, 2004 "Nonroad Duty Cycle Testing for Toxic Emissions; Final
Report" (SwRI-08.05004.05). Prepared for U. S. EPA, Office of Transportation and Air
Quality, Ann Arbor, MI by Southwest Research Institute. (EPA document # 420-R-04-
018). This document is available in Docket EPA-HQ-OAR-2005-0036.
75 Starr, M. May, 2003 "Transient and Steady-State Emissions Testing of Ten Different
Nonroad Diesel Engines Using Four Fuels; Final Report" (SwRI-08.03316). Prepared for
California Air Resources Board, El Monte, CA by Southwest Research Institute. (EPA
document # 420-R-03-901). This document is available in Docket EPA-HQ-OAR-2005-
0036.
76
40CFRPart80, SubpartD.
77 65 FR 6822 (February 10, 2000)
78 66 FR 5002 (January 18, 2001) http://www.epa.gov/otaq/diesel.html
79
69 FR 38958 (June 29, 2004)
65 FR 6697, February 10, 2000.
2-113
-------
Final Regulatory Impact Analysis
81 66 FR 5001, January 18, 2001.
82 69 FR 38958, June 29, 2004.
83 67 FR 68241, November 8, 2002.
84 67 FR 68241, November 8, 2002.
85 60 FR 34582, July 3, 1995.
86 64 FR 15208, March 30, 1999 and 65 FR 24267, April 25, 2000.
87 61 FR 52088, October 4, 1996.
88 67 FR 53050, August 14, 2002.
89 64 FR 73300, December 29, 1999 and 68 FR 9746, February 28, 2003.
90
63 FR 18978, April 16, 1998.
2-114
-------
Final Regulatory Impact Analysis
Chapter 3: Table of Contents
Chapter 3: Air Quality and Resulting Health and Welfare Effects of Air Pollution from Mobile
Sources 3
3.1 Air Quality and Exposure Measurements 3
3.1.1 Ambient Monitoring 3
3.1.2 Population-Based (Representative) Exposure Measurements 9
3.1.3 Elevated Concentrations and Exposures in Mobile Source-Impacted Areas 12
3.1.3.1 Concentrations Near Major Roadways 14
3.1.3.1.1 Particulate Matter 14
3.1.3.1.2 Gaseous Air Toxics 15
3.1.3.2 Exposures Near Major Roadways 16
3.1.3.2.1 In Vehicles 16
3.1.3.2.2 InHomesand Schools 19
3.1.3.2.3 Pedestrians and Bicyclists 21
3.1.3.3 Concentrations and Exposure in Homes with Attached Garages 22
3.1.3.4 Concentrations and Exposure in Parking Garages 25
3.1.3.5 Concentrations and Exposure at Service Stations 28
3.1.3.6 Occupational Exposure 29
3.1.4 Uncertainties in Air Toxics Measurements 31
3.2 Modeled Air Quality, Exposures, and Risks for Air Toxics 31
3.2.1 National-Scale Modeled Air Quality, Exposure, and Risk for Air Toxics 31
3.2.1.1 Air Quality Modeling 34
3.2.1.1.1 Methods 34
3.2.1.1.2 Air Quality Trends for Air Toxics: Reference Case 35
3.2.1.1.3 Distributions of Air Toxic Concentrations across the U. S.: Reference
Case 38
3.2.1.1.4 Impacts of Controls on Ambient Concentrations 39
3.2.1.2 Exposure and Risk Modeling 44
3.2.1.2.1 Methods 44
3.2.1.2.2 Exposure and Risk Trends for Air Toxics: Reference Case 48
3.2.1.2.3 Distributions of Air Toxics Risk across the U. S.: Reference Case 57
3.2.1.2.4 Impacts of Controls on Average Inhalation Cancer Risks and
Noncancer Hazards 67
3.2.1.3 Strengths and Limitations 74
3.2.1.4. Perspective on Cancer Cases 77
3.2.2 Local-Scale Modeling 79
3.3 Ozone 83
3.3.1 Science of Ozone Formation 84
3.3.2 Health Effects of Ozone 85
3.3.3 Current 8-Hour Ozone Levels 86
3.3.4 Projected 8-Hour Ozone Levels 88
3.3.4.1 CAIR Ozone Air Quality Modeling 88
3.3.4.2 Ozone Response Surface Metamodel Methodology 93
3-1
-------
Final Regulatory Impact Analysis
3.3.4.3 Ozone Response Surface Metamodel Results 96
3.3.5 Environmental Effects of Ozone Pollution 97
3.3.5.1 Impacts on Vegetation 97
3.4ParticulateMatter 99
3.4.1 Science of PM Formation 99
3.4.2 Health Effects of Paniculate Matter 100
3.4.2.1 Short-Term Exposure Mortality and Morbidity Studies 100
3.4.2.2 Long-Term Exposure Mortality and Morbidity Studies 101
3.4.2.3 Roadway-Related Pollution Exposure 102
3.4.3 Current and Projected PM Levels 102
3.4.3.1 Current PM2.5 Levels 102
3.4.3.2 Current PMio Levels 103
3.4.3.3 Projected PM2.5 Levels 104
3.4.3.3.1 PM Modeling Methodology 104
3.4.3.3.2 Areas at Risk of Future PM2.5 Violations 104
3.4.4 Environmental Effects of PM Pollution 107
3.4.4.1 Visibility Impairment 107
3.4.4.1.1 Current Visibility Impairment 108
3.4.4.1.2 Current Visibility Impairment at Mandatory Class I Federal Areas.. 108
3.4.4.1.3 Future Visibility Impairment 109
3.4.4.1.4 Future Visibility Impairment at Mandatory Class I Federal Areas.... 109
3.4.4.2 Atmospheric Deposition 110
3.4.4.2.1 Heavy Metals Ill
3.4.4.2.2 Polycyclic Organic Matter 112
3.4.4.3 Materials Damage and Soiling 112
3.5 Health and Welfare Impacts of Near-Roadway Exposure 113
3.5.1 Mortality 114
3.5.2 Non-Allergic Respiratory Symptoms 116
3.5.3 Development of Allergic Disease and Asthma 117
3.5.4 Cardiovascular Effects 118
3.5.5 Birth Outcomes 119
3.5.6 Childhood Cancer 120
3.5.7 Summary of Near-Roadway Health Studies 121
Appendix 3 A: Influence of Emissions in Attached Garages on Indoor Air Benzene
Concentrations and Human Exposure 124
Appendix 3B: 8-Hour Ozone Nonattainment 144
Appendix 3C: PM Nonattainment 159
Appendix 3D: Visibility Tables 162
5-2
-------
Final Regulatory Impact Analysis
Chapter 3: Air Quality and Resulting Health and Welfare Effects of
Air Pollution from Mobile Sources
3.1 Air Quality and Exposure Measurements
3.1.1 Ambient Monitoring
Ambient air toxics data are useful for identifying pollutants of greatest concern, areas of
unhealthy ambient air toxics concentrations, and air toxics trends; evaluating and improving
models; and assessing the effectiveness of air toxics reduction strategies. Ambient air toxics data
though have limitations for use in risk assessments. While EPA, states, tribes, and local air
regulatory agencies collect monitoring data for a number of toxic air pollutants, both the
chemicals monitored and the geographic coverage of the monitors vary from state to state.1 In
recent years, the US EPA and states have initiated more extensive monitoring of air toxics to
assist in air pollution management through measurement and mitigation.2 EPA is working with
its regulatory partners to build upon the existing monitoring sites to create a national monitoring
network for a number of toxic air pollutants. The goal is to ensure that those compounds that
pose the greatest risk are measured. In 2004, EPA published a draft National Air Toxics
Monitoring Strategy to advance this goal.3 The National Air Toxics Trends Station (NATTS)
monitoring network is currently in place, consisting of 23 sites in 22 urban areas nationally.4
The available monitoring data help air pollution control agencies track trends in toxic air
pollutants in various locations around the country. EPA conducted a pilot city monitoring
project in 2001 that included sampling in four urban areas and six small city/rural areas (see
Figure 3.1-1). This program helped answer several important national network design questions
(e.g., sampling and analysis precision, sources of variability, and minimal detection levels).
Figure 3.1-1. Map of Ten Cities in Monitoring Pilot Project
(Seattle
Detroit •
Cedar Rapids ^ Providence
Grand Junction
Charleston
San Jacinto Rio Rancho
San Juan
Tampa • |
5-3
-------
Final Regulatory Impact Analysis
Building on the pilot program, the US EPA and states established a national air toxics
monitoring program beginning with a 10-city pilot program, which now consists of the NATTS,
and numerous community-scale monitoring studies.5 To guide development of the monitoring
program, a qualitative data analysis project was begun in 2001 and the first phase was completed
in 2004. The analysis showed that typical urban concentration ranges for most VOCs are
approximately an order of magnitude (or more) higher than the background concentrations.
Because air toxics concentrations vary spatially, other monitoring networks are needed to
provide additional, especially rural, concentrations. Extrapolation for most air toxics beyond the
urban scale is not recommended without a network of rural measurements capable of capturing
gradients between urban and rural areas. For the latest information on national air toxics
monitoring, see www.epa.gov/ttn/amtic/airtxfil.html.
Figure 3.1-2 shows measurements of benzene taken from 95 urban monitoring sites
around the country. These urban areas generally have higher levels of benzene than other areas
of the country. Measurements taken at these sites show, on average, a 47% drop in benzene
levels from 1994 to 2000. During this period, EPA phased in new (so-called "tier 1") car
emission standards; required many cities to begin using cleaner-burning gasoline; and set
standards that required significant reductions in benzene and other pollutants emitted from oil
refineries and chemical processes.
Figure 3.1-2. Ambient Benzene, Annual Average Urban Concentrations, Nationwide, 1994-
2000
c 4
o
o
90% of sites have concentrations below this line
of sites have concentrations below this line
94 95 Q6 97 99
1994-00: 47% decrease
99
00
Following is a summary of analyses recently performed on ambient measurements of air
toxics to identify pollutants and geographic areas of concern and to evaluate trends. Use of
monitoring data to evaluate and improve models is discussed in Section 3.2.
EPA recently completed a study of the spatial and temporal trends in ambient air toxics
data within the NATTS and other networks from 1990-2003.6 Most data came from urban
monitors. Nationally, citywide average annual concentrations of benzene, formaldehyde, and
3-4
-------
Final Regulatory Impact Analysis
acetaldehyde, varied by about a factor of five, and 1,3 -butadiene by more than 10 times. The
coefficient of variation21 of annual average concentrations between different monitors within the
same city averaged 0.37 for benzene, 0.45 for 1,3-butadiene. Between cities, the coefficient of
variation could vary substantially. Different pollutants showed different seasonal trends, with
average concentrations of benzene and 1,3-butadiene being highest in colder seasons, while
average concentrations of formaldehyde and acetaldehyde were higher during warm seasons,
reflecting the high photochemical production of aldehydes. The concentrations of benzene,
butadiene, and acetaldehyde fell substantially over different time periods. From 1990-2003,
benzene concentrations fell by 57%. Insufficient data existed in earlier years to analyze 1,3-
butadiene and acetaldehyde. Formaldehyde increased by 134% over this period, although
changes in sampling methodology at some sites around 1995 make this quantification suspect.
From 1995-2003, the average annual changes in benzene, 1,3-butadiene, formaldehyde, and
acetaldehyde were -47%, -54%, +11%, and -12%. From 1998-2003, the changes were -21%, -
46%, +17%, and -4%, respectively. For benzene, these trends were statistically significant, but
formaldehyde and acetaldehyde trends after 1995 were not.
One recent publication evaluated the trends in ambient concentrations of benzene and
1,3-butadiene in the Houston, TX metropolitan area.7 Using data from two air monitoring
networks, a state-based network and the Photochemical Assessment Monitoring Systems, the
study constructed a statistical model, controlling for meteorology and seasonality, to evaluate
trends in ambient toxics over the 1997-2004 time period. Averaged over state monitoring sites
with data across the time period, the model estimated 1.7% and 3.7% average annual decrease in
ambient benzene and 1,3,-butadiene, respectively. Mobile source and point source emission
reductions contributed roughly equally to this change. Examining long-term average
concentration differences across monitoring sites, benzene varied by roughly two-fold across
monitors while 1,3-butadiene varied roughly six-fold across monitors. This may be attributable
to the substantial contribution of industrial sources to the local 1,3-butadiene inventory, while the
benzene inventory is dominated by mobile sources. The study also evaluated differences in
weekday and weekend concentration, with the model predicting significant meteorologically-
adjusted concentration weekday increases relative to weekend only during the 6-9 A.M. morning
rush hour period.
A recent study from San Francisco, CA evaluated trends in ambient benzene emissions
and air quality throughout the 1990's.8 The study noted substantial decreases in benzene
emissions and ambient concentrations. Unique to the study was the attribution of components of
these reduction to specific regulatory programs related to vehicles and fuels. In particular, the
study attributed a 1-year drop of 54% in benzene emission rates to a combination of the
introduction of California phase 2 RFG (attributed a 50% decrease) and fleet turnover (attributed
a 4% decrease). During the same year (1995-1996), a 42% reduction in the ambient
concentration of benzene was also observed. Fleet turnover effects were shown to be cumulative
over time. The study indicates that in San Francisco both fuel and vehicle effects are important
a A "coefficient of variation" is a measure of variability, and for a set of data is defined as the standard deviation
over the mean.
3-5
-------
Final Regulatory Impact Analysis
contributors to changes in emissions and ambient concentrations of benzene.
New York State has a systematic program in place that has been measuring air toxics
since the 1990s.9 The network of monitors is located throughout urban, industrial, residential
and rural locations. The New York State Department of Environmental Conservation recently
examined the spatial and temporal characteristics of benzene by analyzing five of the 32 total
network sites across the state (see Table 3.1-1). Spatial trends show a wide range of annual
average benzene concentrations, with the lowest value at a rural site and the highest at an
industrial site. The recent 3-year period of 2001-2003 was also compared with the longer 1990-
2003 period. The 3-year period exhibits a decrease in mean concentration compared to the entire
period, indicating that benzene concentrations are decreasing over New York State throughout
this period. The mean annual rate of change in the period 1990 to 2003 was determined using
linear regression and moving average (KZ filter) on the concentration data. The analysis
indicated that site-specific ambient concentration levels of benzene decreased by 50% or more
during 1990 to 2003. These decreases occurred in ozone nonattainment areas that had
reformulated gasoline (RFG) requirements as well as in the rest of the state. The downward
trend can be attributed to regulatory measures aimed at reducing toxic emissions from industrial
sources, replacement of older higher emitting vehicles with vehicles meeting more stringent EPA
standards for hydrocarbon emissions, as well as the adoption of RFG in 1995 and 1999 for the 1-
hour ozone nonattainment areas in New York State. Since trends were observed for sites that
were not part of the RFG program, decreases may also be attributed to the improvement in
vehicle emissions technology and the state-wide adoption of the California Low Emission
Vehicle program.
A similar downward trend was observed in California. In California, the Air Resources
Board (ARB) maintains an Almanac of Emissions and Air Quality.10 The Almanac summarizes
statewide emissions, statewide annual average concentrations (calculated as a mean of monthly
means), and statewide average health risks for selected air toxics. Currently there are data
available for ten air toxics in California, including benzene. The ARB network consists of 18 air
quality monitoring stations. The data collected, analyzed, and reported reflect a spatial average;
therefore, ambient concentrations for individual locations may be higher or lower. Estimates
show that approximately 84% of the benzene emitted in California comes from motor vehicles,
including evaporative leakage and unburned fuel exhaust. The predominant sources of total
benzene emissions in the atmosphere are gasoline fugitive emissions and gasoline motor vehicle
exhaust. Approximately 49% of the statewide benzene emissions can be attributed to on-road
motor vehicles, with an additional 35% attributed to other mobile sources such as recreational
boats, off-road recreational vehicles, and lawn and garden equipment. Currently, the benzene
content of gasoline is less than 1%. Some of the benzene in the fuel is emitted from vehicles as
unburned fuel. Benzene is also formed as a partial combustion product of larger aromatic fuel
components. Industry-related stationary sources contribute 15% and area-wide sources
contribute 1% of the statewide benzene emissions. The primary stationary sources of reported
benzene emissions are crude petroleum and natural gas mining, petroleum refining, and electric
generation. The primary area-wide sources include residential combustion of various types such
as cooking and water heating. The primary natural sources are petroleum seeps that form where
3-6
-------
Final Regulatory Impact Analysis
oil or natural gas emerge from subsurface sources to the ground or water surface. The statewide
benzene levels have shown generally steady improvement since 1990. To examine the trend in
benzene while minimizing the influences of weather on the trend, the statewide average benzene
concentration for 1990-1992 was compared to that for 2001-2003. The result was a 72%
decrease in benzene concentration. These downward trends for benzene and other air toxics are
a result of many control measures implemented to reduce emissions.
Another recent evaluation of hazardous air pollutant (HAP) trends was conducted for
selected metropolitan areas.11 Researchers retrieved historical concentration and emissions data
from the US EPA for Boston, New York City, Philadelphia, Tampa Bay, Detroit, Dallas, St.
Louis, Denver, Los Angeles, and Seattle, chosen for each of EPA's ten regions. Annual and
seasonal trends were generated to evaluate reductions in HAP emissions and ambient
concentrations during the time period 1990-2003. Several air toxics were targeted, including
benzene. To evaluate the trends, average concentrations from 1990-1994 were compared to
2002-2003 (these time periods were chosen due to availability of data). The results showed that
over 85% of the metropolitan area-HAP combinations decreased in their HAP concentrations,
while less than 15% realized an increase. For example, Table 3.1-2 shows that benzene
concentrations decreased in seven of the ten metropolitan areas (range 19 to 79%).
Each of these analyses consistently illustrates the significant reductions in national annual
average concentrations of benzene and other air toxics. The air pollution management efforts of
the US EPA and states have been effective in reducing ambient concentrations of air toxics over
time. Additional reductions are expected with the implementation of additional regulatory
measures such as this one. It should be noted that due to the limited spatial and temporal
coverage of air toxics monitoring networks, using ambient monitors to represent exposure adds
substantial uncertainty in exposure assessment.
5-7
-------
Final Regulatory Impact Analysis
Table 3.1-1. Site Descriptions of the Monitoring Stations Along with Mean Benzene Concentration from 1990-2003 and 2001-
2003, for Monitoring Stations in New York State.
Site Character
Location Area
2000 Population
(thousands)
Annual Vehicle
Miles Traveled
(million miles)
Period 1990-2003
Mean Concentration
(ns/m3)
Period 200 1-2003
Mean Concentration
(ug/m3)
Lackawanna
Industrial
Buffalo
950
8250
5.09
2.26
Eastern District
High School
Urban
Brooklyn
2465
4246
2.85
2.05
Troy
Small Urban
Hudson Valley
153
1413
2.31
1.68
Niagara Falls
Urban Industrial
Niagara
220
1546
1.80
1.08
Whiteface
Mountain Base
Lodge
Rural
Essex
39
577
0.86
0.54
-------
Final Regulatory Impact Analysis
3.1.2 Population-Based (Representative) Exposure Measurements
In addition to measurements of outdoor concentrations, an important component of
understanding human exposure to air toxics is the body of studies that employ survey techniques
to assess microenvironmental and representative populations' exposures. Typically, these
studies are designed to represent a discrete geographic area. The personal exposure
concentration summaries from these studies are shown in Table 3.1-3.
The National Human EXposure Assessment Survey (NHEXAS) was a series of
population-based exposure studies. The states in EPA Region 5 were the focus of one NHEXAS
study, which was conducted in mid-1990.12 Nearly 400 personal and indoor air samples were
obtained from both smokers and non-smokers, along with a smaller number of outdoor air
samples in residential areas. Measurements took place over 6 days per subject. Overall, average
personal exposure to benzene was 7.52 ug/m3, with indoor air concentrations averaging 7.21
ug/m3. Outdoor air concentrations averaged 3.61 ug/m3. Personal air concentrations were
significantly associated with indoor air concentrations, as well as blood concentrations. The
preliminary results of the NHEXAS pilot study in Arizona, another study area, indicate that
among the 179 statistically-sampled homes, median indoor concentrations were 1.3 ug/m3 during
the mid-1990's, while outdoor concentrations were 1.0 ug/m3.13 Furthermore, reported results
from the Arizona study indicate that fuel-related VOCs are elevated in homes with attached
garages.
In another study based on a random population-based sample of an urban population, 37
non-smoking residents of South Baltimore, MD were equipped with passive monitors to assess
3-day average personal exposure to VOCs, in addition to indoor and outdoor air.14 Monitoring
took place in 2000 and 2001. Modeled air quality data from the ASPEN dispersion model,
employed in EPA's National Air Toxics Assessment for 1996, were also obtained for the study
area. Overall, median outdoor modeled concentrations of benzene and other fuel-related VOCs
corresponded well with measured data in the area (correlation coefficient of median VOC
concentrations = 0.97). Average personal exposure to benzene was 4.06 ug/m3, while 95th
percentile values were 7.30 ug/m3. For indoors, the respective values were 3.70 and 8.34 ug/m3,
while for outdoors the values were 1.84 and 3.14 ug/m3. Overall, the study provides evidence
that modeling outdoor benzene concentrations using ASPEN, as is done in this rule, provides
adequate representation of outdoor values. However, indoor and personal exposures are also
influenced by other sources, as is described in the section on attached garages.
While not a population-based study, the recently-completed Relationships of Indoor,
Outdoor and Personal Air (RIOPA) study provides a depiction of indoor, outdoor, and personal
concentrations of benzene and other toxics in three regions with differing source mixtures.15 100
non-smoking homes in each of Los Angeles, CA, Houston, TX, and Elizabeth, NJ were selected
for sampling in areas representing locations dominated by emissions from mobile sources,
stationary sources, and a mixture of sources, respectively. In the adult sample, average personal
exposures to benzene were 3.64 ug/m3, with a 95th percentile of 10.7 ug/m3. Respective statistics
3-9
-------
Final Regulatory Impact Analysis
for indoor air were 3.50 ug/m3 and 10.0 ug/m3, while outdoor statistics were 2.15 and 5.16
ug/m3.
Few studies have systematically addressed exposures among representative samples of
children. Several have been done in Minnesota, with others in New York, Los Angeles, and
Baltimore areas.
For the Minnesota Children's Pesticide Exposure Study (MNCPES), conducted in urban
and rural areas in the vicinity of Minneapolis-St. Paul, MN,16 all monitoring used the same 6-day
monitoring duration as used in the Region 5 NHEXAS study. In the first phase of the study, a
statistically representative sample of 284 homes with children underwent air monitoring for
VOCs. Low-income and minority homes were over sampled to ensure representation. Indoor
benzene concentrations averaged 4.6 ug/m3, with the data skewed toward higher concentrations.
The 95th percentile concentration was 12.7 ug/m3. Homes with attached garages had
significantly higher concentrations of benzene indoors (p < 0.0001). In the second phase of the
study, a subset of 100 children underwent intensive monitoring of personal, indoor, and outdoor
air as well as activity tracking via diary. Overall personal exposures were 4.8 ug/m3, with a 95th
percentile of 9.1 ug/m3. Indoor concentrations in the intensive period averaged 3.9 ug/m3 and
outdoor averaged 3.3 ug/m3. Regression analysis indicated that personal exposures generally
were higher than the time-weighted average of indoor and outdoor air. Furthermore, personal
exposures to benzene and toluene were elevated for children living in a home with an attached
garage, but only the relationship for toluene was significant at the 95% confidence level.
3-10
-------
Final Regulatory Impact Analysis
Table 3.1-2. Benzene Emission (Tons Per Year) and Concentration (ug/m3) Comparison
Metropolitan
Area
Boston
New York City
Philadelphia
Tampa Bay
Detroit
Dallas
St. Louis
Denver
Los Angeles
Seattle
1990
Emissions
6262
16653
5961
3103
6480
7933
4358
2800
19762
5844
2002
Emissions
2229
7512
2577
2408
4388
2832
2304
1913
4168
4315
% Change
in
Emissions
-64.4
-54.9
-56.8
-22.4
-32.3
-64.3
-47.1
-31.7
-78.9
-26.2
1990-1994
Average
Concentration
3.93
3.24
3.60
NA
4.19
1.21
5.16
NA
8.97
NA
2002-2003
Average
Concentration
0.81
1.35
1.26
NA
3.40
0.78
1.43
2.75
2.34
1.39
% Change
in
Concentration
-79.5
-58.5
-64.9
NA
-18.7
-35.8
-72.3
NA
-73.9
NA
5-11
-------
Final Regulatory Impact Analysis
In another study, students recruited from an inner-city school in Minneapolis, MN
participated in an exposure study called SHIELD.17 Students were recruited using stratified
random sampling, with a total of 153 children participating between two seasons. Home and
personal samples were collected and averaged over two continuous days of sampling using
passive methods. School measurements took place during school hours only, over the course of
5 days, and outdoor measurements were set up to run continuously outside the school through
each week sampled (Monday through Friday). The study reported median, 10th, and 90th
percentile concentrations. In personal samples, median benzene concentrations were 1.5 ug/m3
in spring and 2.1 ug/m3 in winter.18
The TEACH exposure study tracked inner-city high school students' exposures in New
York, NY and Los Angeles, CA. In the New York City study, 42 students underwent personal,
indoor home, and outdoor home air quality monitoring during two seasons.19 Average winter
benzene personal concentrations were 4.70 ug/m3, while indoor and outdoor concentrations
averaged 5.97 and 2.55 ug/m3. Average indoor concentrations exceeding average personal
concentrations is unique to the TEACH winter results. Summer values were 3.09, 1.75, and 1.31
ug/m3, respectively. The authors noted that VOC concentrations within the city tracked traffic
patterns. There was no substantial evidence for indoor sources of benzene.20 In a subsequent
publication, personal exposure concentrations for both cities were reported, averaged across both
seasons. New York City average exposure concentrations were 3.82 ug/m3, while Los Angeles
average exposure concentrations were 4.64 ug/m3.21
Overall, these studies show that personal and indoor concentrations of benzene and other
VOCs are substantially higher than those found outdoors (see Table 3.1-3). In general, these
differences are statistically significant. Some of the factors leading to these elevated
concentrations are likely a result of motor vehicle impacts such as exhaust and evaporative
emissions in attached garages, exposures during on-road commutes and exposures during vehicle
re-fueling. These and other factors are discussed in more detail in Section 3.1.3. This suggests
that risk reductions from the controls in this proposal will be greater than can currently be
estimated using national-scale modeling tools.
3.1.3 Elevated Concentrations and Exposures in Mobile Source-Impacted Areas
Air quality measurements near roads often identify elevated concentrations of air toxic
pollutants at these locations. The concentrations of air toxic pollutants near heavily trafficked
roads, as well as the pollutant composition and characteristics, differ from those measured distant
from heavily trafficked roads. Thus, exposures for populations residing, working, or going to
school near major roads are likely different than for other populations. Following is an overview
of concentrations of air toxics and exposure to air toxics in areas experiencing elevated pollutant
concentrations due to the impacts of mobile source emissions.
3-12
-------
Final Regulatory Impact Analysis
Table 3.1-3. Personal Exposure to Benzene from Population-Based Studies"
Location
EPA Region
5
Baltimore,
MD
Elizabeth,
NJ,
Houston, TX,
Los Angeles
CA
Elizabeth,
NJ,
Houston, TX,
Los Angeles
CA
Minneapolis
St. Paul, MN
Minneapolis,
MN
New York,
NY
Los Angeles,
CA
Year(s)
1995-
1996
2000-
2001
1999-
2001
1999-
2001
1997
2000
1999-
2000
1999-
2000
Includes
Smokers
Yes
No
No
No
Yese
Yese
No
No
Personal
Average
(ug/m3)
7.52
4.06
3.64
4.16
4.8
2.1
Winter
1.5
Spring
4.7
Winter
3.1
Summer
3.8
Total
4.64
"Upper
Bound"
(Hg/m3)
13.71b
7.30C
10.7C,
27.4g
12.0C,
43.6s
9.1
6.5
Winter"
4.2
Spring
11.4
Winte/
7.0
Summer
12.3
Totaf
11.27
Indoor
Average
(Hg/m3)
7.21
3.70
3.50
N/Rh
3.9
2.2
Winter
2.1
Spring
6.0
Winter
1.8
Summer
3.6
Total
3.87
Outdoor
Average
(Hg/m3)
3.61
1.84
2.15
N/Rh
3.3
1.3
Winter
1.1
Spring
2.5
Winter
1.3
Summer
1.8
Total
3.32
Reference
Clayton et
al. (1999)
Payne-
Sturges et
al. (2004)
Weisel et
al. (2005)
Weisel et
al. (2005)
Adgate et
al. (2004a)
Adgate et
al. (2004b)
Kinney et
al. (2002);
Sax et al.
(2006)
Sax et al.
(2006)
a Children's studies in italics
b 90th percentile
c 95th percentile
d Mean +2 standard deviations
e Smoking in homes
f Maximum measured value
g 99th percentile
h Not reported
3-13
-------
Final Regulatory Impact Analysis
3.1.3.1 Concentrations Near Major Roadways
3.1.3.1.1 Parti culate Matter
Mobile sources influence temporal and spatial patterns of criteria pollutants, air toxics,
and PM concentrations within urban areas. Motor vehicle emissions may lead to elevated
concentrations of pollutants near major roads. Since motor vehicle emissions generally occur
within the breathing zone, near-road populations may be exposed to "fresh" primary emissions as
well as combustion pollutants "aged" in the atmosphere. For paniculate matter, these fresh
versus aged emissions can result in the presence of varying particle sizes near roadways,
including ultrafine, fine, and coarse particle modes.
The range of particle sizes of concern is quite broad and is divided into smaller
categories. Defining different size categories is useful since particles of different sizes behave
differently in the atmosphere and in the human respiratory system. Table 3.1-4 lists the four
terms for categorizing particles of different sizes as defined by the US EPA.22
Table 3.1-4. Descriptions and Particle Sizes of Each Category of Particles
Description
Supercoarse
Coarse (or Thoracic Coarse Mode)
Fine (or Accumulation Mode)
Ultrafine (or Nuclei Mode)a
Particle Size, dp (um)
dp>10
2.5 < dp < 10
0.1
-------
Final Regulatory Impact Analysis
ultrafme mode particles dominate the number concentration in close proximity to the roadway,
while fine mode dominates farther from the road. Particle size distributions, mass and elemental
carbon concentrations have been examined near roads in Los Angeles.25'26 Researchers observed
a four-fold increase in particle number concentrations, when comparing measurements 300 m
and 20 m from LA highways. Other studies have similarly shown that ultrafme mode particles
show a sharp decrease in particle number concentrations as the distance from major roadways
increases.27'28 Evidence was recently found of increased exposures to ultrafme particles near
roads when it was discovered that children living near major roads had elevated levels of
particle-containing alveolar macrophages.29 Additionally, roadside monitoring has shown that
particle number varies with vehicle type and vehicle operating conditions. For example, elevated
ultrafme mode particle concentrations have been identified when operating speeds on the road
increase as well as when the proportion of heavy-duty diesel vehicles increases.30'31
An increase in coarse particles near roads could originate from engine deterioration,
brake and tire wear, and secondary aerosol formation.32'33'34'35 Engine deterioration is generally
a function of vehicle age and maintenance condition. Brake wear emissions are highly
dependent on brake pad materials.36 Secondary aerosol formation is dependent on fuel
composition, emission rates, atmospheric chemistry, and meteorology. Re-entrained road dust,
as well as brake and tire wear will also contribute to increased concentrations of coarse PM.
Meteorological factors can affect exposures to motor vehicle emissions near the road.
Researchers have noted that particle number concentrations changed significantly with changing
wind conditions, such as wind speed, near a road.37 Studies suggest that ambient temperature
variation can also affect particle number gradients near roads substantially.38 Wind direction
also affects traffic-related air pollution mass concentrations inside and outside of schools near
motorways.39'40 Diurnal variations in mixing layer height will also influence both near-road and
regional air pollutant concentrations. Decreases in the height of the mixing layer (due to
morning inversions, stable atmosphere, etc.) will lead to increased pollutant concentrations at
both local and regional scales.
3.1.3.1.2 Gaseous Air Toxics
Concentrations of mobile source air toxics have been estimated by a number of different
methods such as the NATA National-Scale Assessment, local-scale modeling assessments, and
from air quality monitoring in locations in immediate proximity to busy roadways. Each
approach offers a different level of representation of the concentrations of air toxics near
roadways.
Air quality monitoring is one way of evaluating pollutant concentrations at locations near
sources such as roadways. Ambient VOC concentrations were measured around residences in
Elizabeth, NJ, as part of the Relationship among Indoor, Outdoor, and Personal Air (RIOPA)
study. Data from that study was analyzed to assess the influence of proximity of known ambient
emission sources on residences.41 The ambient concentrations of benzene, toluene,
ethylbenzene, and xylene isomers (BTEX) were found to be inversely associated with: distances
from the sampler to interstate highways and major urban roads; distance from the sampler to
gasoline stations; atmospheric stability; temperature; and wind speed. The data indicate that
3-15
-------
Final Regulatory Impact Analysis
BTEX concentrations around homes within 200 m of roadways and gas stations are 1.5 to 4
times higher than urban background levels. In a subsequent study, proximity to major roadways,
meteorology, and photochemistry were all found to be significant determinants of ambient
concentration of a range of aldehyde species, including formaldehyde, acetaldehyde, acrolein,
and others. For most aldehydes, spring and summer concentrations were significantly higher
than those from colder seasons.42 However, formaldehyde concentrations were significantly
lower in summertime, suggesting greater photochemical destruction than production. On colder
days, when photochemical activity was lower, concentrations of formaldehyde, acetaldehyde,
acrolein, and other aldehydes were significantly higher with increasing proximity to high-traffic
roads.
Several other studies have found that concentrations of benzene and other mobile source
air toxics are significantly elevated near busy roads compared to "urban background"
concentrations measured at a fixed site 43>44>45>46>47>48 por example, measurements near a
tollbooth in Baltimore observed mean benzene concentrations to vary by time of day from 3 to
22.3 ug/m3 depending on traffic volume, vehicle type, and meteorology.49 In comparison with
ambient levels, Maryland's Department of Environment reported the range of benzene annual
averages measured at seven different monitoring sites in 2000 between 0.27-0.71 ug/m3.50
Another study measured the average benzene concentration in a relatively high traffic density (~
16000 automobiles/day) sampling area at 9.6 ug/m3 and in rural areas with hardly any traffic (<
50 automobiles/day) at 1.3 ug/m3.51 The concentration of benzene, along with several other air
toxics (toluene and the isomeric xylenes), in the urban area far exceeded those in the rural area.
According to Gaussian dispersion theory, pollutants emitted along roadways will show
highest concentrations nearest a road, and concentrations exponentially decrease with increasing
distance downwind. These near-road pollutant gradients have been confirmed by measurements
of both criteria pollutants and air toxics.52'53'54'55'56 Researchers have demonstrated exponential
reductions in concentrations of CO, as well as PM number, and black carbon (as measured by an
aethalometer), with increasing downwind distance from a freeway in Los Angeles.57'58 These
pollutants reached background levels approximately 300 m downwind of the freeway.
3.1.3.2 Exposures Near Major Roadways
The modeling assessments and air quality monitoring studies discussed above have
increased our understanding of ambient concentrations of mobile source air toxics and potential
population exposures. Results from the following exposure studies reveal that populations
spending time near major roadways likely experience elevated personal exposures to motor
vehicle related pollutants. In addition, these populations may experience exposures to differing
physical and chemical compositions of certain air toxic pollutants depending on the amount of
time spent in close proximity to motor vehicle emissions. Following is a detailed discussion on
exposed populations near major roadways.
3.1.3.2.1 In Vehicles
Several studies suggest that people may experience significant exposures while driving in
vehicles. A recent in-vehicle monitoring study was conducted by EPA and consisted ofin-
3-16
-------
Final Regulatory Impact Analysis
vehicle air sampling throughout work shifts within ten police patrol cars used by the North
Carolina State Highway Patrol (smoking not permitted inside the vehicles).59 Troopers operated
their vehicles in typical patterns, including highway and city driving and refueling. In-vehicle
benzene concentrations averaged 12.8 ug/m3, while concentrations measured at an "ambient" site
located outside a nearby state environmental office averaged 0.32 ug/m3. The study also found
that the benzene concentrations were closely associated with other fuel-related VOCs measured.
The American Petroleum Institute funded a screening study of "high-end" exposure
microenvironments as required by section 21 l(b) of the Clean Air Act.60 The study included
vehicle chase measurements and measurements in several vehicle-related microenvironments in
several cities for benzene and other air toxics. In-vehicle microenvironments (average
concentrations in parentheses) included the vehicle cabin tested on congested freeways (17.5
ug/m3), in parking garages above-ground (155 ug/m3) and below-ground (61.7 ug/m3), in urban
street canyons (7.54 ug/m3), and during refueling (46.0 ug/m3). It should be noted that sample
sizes in this screening study were small, usually with only one to two samples per
microenvironment. The final report of this study is expected to be released in 2007.
In 1998, the California Air Resources Board published an extensive study of
concentrations of in-vehicle air toxics in Los Angeles and Sacramento, CA.61 The data set is
large and included a variety of sampling conditions. On urban freeways, in-vehicle benzene
concentrations ranged from 3 to 15 ug/m3 in Sacramento and 10 to 22 ug/m3 in Los Angeles. In
comparison, ambient benzene concentrations ranged from 1 to 3 ug/m3 in Sacramento and 3 to 7
ug/m3 in Los Angeles.
Studies have also been conducted in diesel buses, such as the one recently conducted of
LA school buses.62'63 In the study, five conventional diesel buses, one diesel bus equipped with
a catalytic particle filter, and one natural gas bus were monitored for benzene, among other
pollutants. These buses were driven on a series of real school bus routes in and around Los
Angeles, CA. Average benzene concentrations in the buses were 9.5 ug/m3, compared with 1.6
ug/m3 at a background urban fixed site in west Los Angeles. Type of bus, traffic congestion
levels, and encounters with other diesel vehicles contributed to high exposure variability between
runs.
The same researchers additionally determined the relative importance of school bus-
related microenvironments to children's pollutant exposure.64 Real-time concentrations of black
carbon (BC), particle-bound PAH, nitrogen dioxide (NC^), particle counts (0.3-0.5 um size
range), and PM2.5 mass were measured inside school buses during long commutes, at bus stops
along the routes, at bus loading and unloading zones, and at nearby urban background sites.
Across all the pollutants, mean concentrations during bus commutes were higher than in any
other microenvironment. Mean exposures in bus commutes were 50 to 200 times more than for
loading and unloading zones at the school, and 20 to 40 times more than for bus stops along the
route, depending on the pollutant. The in-cabin exposures were dominated by the effect of
surrounding traffic when windows were open and by the bus' own exhaust when the windows
were closed. The mean pollutant concentrations in the three school bus commute-related
environments and background air are presented in the Table 3.1-5.
3-17
-------
Final Regulatory Impact Analysis
Table 3.1-5. Mean Concentrations of Black Carbon (BC), Particle Bound PAH, NO2,
Particle Count (PC), and PM2.5 in Three School Bus Commute Microenvironments and
Background Air
BC fag/m3)
Particle
Bound -PAH
(ug/m3)
N02 (ppb)
PC
(count/cm3)
PM2 5 (ug/m3)
Mean Concentrations
Background
2 ±0.1
0.027 ± 0.0015
49 ± 1.0
83 ± 3.1
20 ± 2.4
(Un)Loading
Zone
2 ±0.3
0.015 ± 0.0003
35 ± 0.2
Not collected
Not collected
Bus Stops
4 ±0.4
0.044 ± 0.0045
54 ± 1.9
62 ± 1.8
25"
Bus
Commutes21
3-19 (8)
0.064-0.400
(0.134)
34-110 (73)
77-236 (130)
21-62 (43)
aRanges are associated with different bus types and window positions. Values in
parenthesis are the mean for all runs.
bNot enough data to establish a confidence interval.
In another recent study of commuter buses, concentrations of benzene and other VOCs
were measured in buses on several routes in Detroit, MI.65 The average in-bus concentration of
benzene was 4.5 ug/m3, while the average concentrations at three fixed sites taken during the
study period ranged from 0.9-2.0 ug/m3. In this study, daily bus/ambient concentration ratios
were reported, and ranged from 2.8-3.3 on the three reported study days. The in-bus
concentrations were found to be most influenced by local traffic sources. A number of other
studies similarly observe that passenger car commuters are exposed to elevated pollutant
concentrations while driving on busy roads.66'67'68'69'70'71
Older studies that examine in-vehicle concentrations in older model year vehicles are
difficult to apply for regulatory analyses, due to the relatively rapid changes in vehicle emission
controls over the last 15 years. In general, these studies indicate that concentrations in vehicles
are significantly higher than ambient concentrations.72'73'74 The average benzene measurements
of these older in-vehicle studies (Raleigh, NC and CA South Coast Air Basin) are in Table 3.1-6
along with the more recent studies for comparison.
Overall, these studies show that concentrations experienced by commuters and other
roadway users are substantially higher than ambient air measured in typical urban air. As a
result, the time a person spends in a vehicle will significantly affect their overall exposure.
3-18
-------
Final Regulatory Impact Analysis
Table 3.1-6. Benzene Concentrations (ug/m3) Measured in Vehicles and in Ambient Air
Study
Raleigh, NC (1 989) a
CA South Coast Air Basin (1989) b
Boston, MA (1991) c
Los Angeles, C A (1998)
Sacramento, CA (1998)
Detroit, MI (2000) d
API Gasoline Screening (2002)
LA, CA School Buses (2003)
NC State Highway Patrol (2003)
In- Vehicle
Mean
11.6
42.5
17.0
10-22
3-15
4.5
17.5
9.5
12.8
Max
42.8
267.1
64.0
—
—
10.8
—
—
43.1
Ambient Air
Mean
1.9
9.3-16.9
—
3-7
1-3
0.9-2.0
—
1.6
0.32
Max
8.5
—
—
—
—
—
—
—
1.92
3.1.3.2.2
a A one-hour measurement was taken for each experimental trip.
b The estimated sampling time period was 1.5 hours/round-trip. n=191.
c In-vehicle measurement includes both interstate and urban driving, n=40.
d Measurements taken from interiors of urban buses.
In Homes and Schools
The proximity of schools to major roads may result in elevated exposures for children
due to potentially increased concentrations indoors and increased exposures during outdoor
activities. Here we discuss international studies in addition to the limited number of US studies,
because while fleets and fuels outside the U.S. can be much different, the spatial distribution of
concentrations is relevant.
There are many sources of indoor air pollution in any home or school. These include
indoor sources and outdoor sources, such as vehicle exhaust. Outdoor air enters and leaves a
house by infiltration, natural ventilation, and mechanical ventilation. In infiltration, outdoor air
flows into the house through openings, joints, and cracks in walls, floors, and ceilings, and
around windows and doors. In natural ventilation, air moves through opened windows and doors.
Air movement associated with infiltration and natural ventilation is caused by air temperature
differences between indoors and outdoors and by wind. Finally, there are a number of
mechanical ventilation devices, from outdoor-vented fans that intermittently remove air from a
single room, such as bathrooms and kitchen, to air handling systems that use fans and duct work
to continuously remove indoor air and distribute filtered and conditioned outdoor air to strategic
points throughout the house. The concentrations of outdoor pollutants can therefore influence
indoor concentrations. A review of the literature determined that approximately 100% of
gaseous compounds, such as benzene, and 80% of diesel PM can penetrate indoors.75'76
In the Fresno Asthmatic Children's Environment Study (FACES), traffic-related
pollutants were measured on selected days from July 2002 to February 2003 at a central site, and
inside and outside of homes and outdoors at schools of asthmatic children.77 Preliminary data
indicate that PAH concentrations are higher at elementary schools located near primary roads
than at elementary schools distant from primary roads (or located near primary roads with
3-19
-------
Final Regulatory Impact Analysis
limited access). PAH concentrations also appear to increase with increase in annual average
daily traffic on nearest major collector.
The East Bay Children's Respiratory Health Study studied traffic-related air pollution
outside of schools near busy roads in the San Francisco Bay Area in 2001,78 Concentrations of
the traffic pollutants PMio, PM2 5, black carbon, total NOX, and NC>2 were measured at ten school
sites in neighborhoods that spanned a busy traffic corridor during the spring and fall seasons.
The school sites were selected to represent a range of locations upwind and downwind of major
roads. Differences were observed in concentrations between schools nearby (< 300 m) versus
those more distant (or upwind) from major roads. Investigators found spatial variability in
exposure to black carbon, NOx, NO, and (to a lesser extent) NO2 associated with roads with
heavy traffic within a relatively small geographic area.
A study to assess children's exposure to traffic-related air pollution while attending
schools near roadways was performed in the Netherlands.79 Investigators measured PM2.5, NO2
and benzene inside and outside of 24 schools located within 400 m of roadways. The indoor
average benzene concentration was 3.2 ug/m3, with a range of 0.6-8.1 ug/m3. The outdoor
average benzene concentration was 2.2 ug/m3, with a range of 0.3-5.0 ug/m3. Overall results
indicate that indoor pollutant concentrations are significantly correlated with traffic density and
composition, percentage of time downwind, and distance from major roadways.
In another study performed in the Netherlands, investigators measured indoor
concentrations of black smoke, PMio, and NO2 in twelve schools between the periods of May
and August 1995.80 The schools were located at varying distances from the motorway (35-645
m). Results indicate that black smoke and NO2 concentrations inside the schools were
significantly correlated with truck and/or car traffic intensity as well as percentage of time
downwind from the motorway and distance of the school from the motorway. PMio
concentrations measured in classrooms during school hours were highly variable and much
higher than those measured outdoors, but they did not correlate with any of the distance or traffic
parameters.
In another Dutch study, researchers monitored children's personal exposure
concentrations, and home indoor and home outdoor levels of "soot" (particle blackness), NO,
and NO2.81 Four-month average concentrations were calculated for each pollutant. Personal
exposure to "soot" was 35-38% higher in students living within 75 meters of roads with 10,000
average annual daily traffic, a statistically significant result. Nonsignificant elevations in
personal exposure to NO, NO2, and NOx were also found.
The TEACH study (Toxic Exposure Assessment - Columbia/Harvard) measured the
concentrations of VOCs, PM2.s, black carbon, and metals outside the homes of high school
Qn
students in New York City. The study was conducted during winter and summer of 1999 on 46
students and in their homes. Average winter (and summer) indoor concentrations exceeded
outdoor concentrations by a factor of 2.3 (1.3). In addition, spatial and temporal patterns of
MTBE concentrations, used as a tracer for motor vehicle pollution, were consistent with traffic
patterns.
3-20
-------
Final Regulatory Impact Analysis
Average benzene concentrations were determined in a recent evaluation of the exposure
of urban inhabitants to atmospheric benzene in Athens, Greece.83 Home and personal levels of
50 non-smokers in six monitoring campaigns varied between 6.0-13.4 and 13.1-24.6 ug/m3,
respectively. Urban levels varied between 15.4 and 27.9 ug/m3 with an annual mean of 20.4
ug/m3. The highest values were observed during the first two sampling periods in fall and
winter, when wind speed was low. The low summer values were attributed to decreased vehicle
traffic. Among home factors, only proximity to busy roads was determined to be an important
influence on indoor benzene levels.
Children are exposed to elevated levels of air toxics not only in their homes, classrooms,
and outside on school grounds, but also during their commute to school. See above discussion of
in-vehicle (school bus and passenger car) concentrations of air toxics for one method of
commuting. The discussion below also presents potential exposures to children from another
commuting method.
3.1.3.2.3 Pedestrians and Bicyclists
Researchers have noted that pedestrians and cyclists along major roads experience
elevated exposures to motor vehicle related pollutants. Although commuting near roadways
leads to higher levels of exposure to traffic pollutants, the general consensus is that exposure
levels of those commuting by walking or biking is lower than for those who travel by car or bus,
(see discussion on in-vehicle exposure in previous section above). For example, investigators
found that personal measurements of exposure to PMio concentrations were 16% higher inside
the car than for the walker on the same route, but noted that a walker may have a larger overall
exposure due to an increase in journey time.84 Similarly, researchers found that traffic-related
pollutant exposure concentrations of car drivers were higher than for cyclists.85 Cyclists are
typically on the border of the road or on dedicated bike paths and therefore further away from the
vehicle emissions and are less delayed by traffic jams. However, after accounting for cyclists'
higher ventilation, the uptake of CO, benzene, toluene, and xylenes by cyclists sometimes
approached that of car drivers, and for NO2 it was significantly higher.
In the early 1990's, researchers studied the in-vehicle concentrations of a large number of
compounds associated with motor vehicle use and the exposure to VOCs of a pedestrian on an
urban sidewalk (50 m from roadways) in Raleigh, NC.86 The mean concentration of benzene in
the six pedestrian sidewalk samples was 6.8 ug/m3. This concentration was lower than the in-
vehicle measurement (11.6 ug/m3), but higher than the fixed-site measurement (1.9 ug/m3) on
urban roadways 100-300 m from streets.
The same researchers studied the exposure of commuters in Boston to VOCs during car
driving, subway travel, walking, and biking.87 For pedestrians, mean time-weighted
concentrations of benzene, toluene, and xylenes of 10.6, 19.8, and 16.7 ug/m3, respectively, were
reported. For cyclists, the time-weighted concentrations were similar to those of pedestrians, at
9.2, 16.3, and 13.0 ug/m3, respectively. In-vehicle exposure concentrations were higher as
discussed above.
Numerous other studies which were conducted in Europe and Asia yield similar results.
3-21
-------
Final Regulatory Impact Analysis
A survey of CO concentration was conducted for various transport modes along heavy traffic
routes in Athens, Greece.88 Results showed that mean CO levels for trips of 30 min were 21.4
ppm for private car, 10.4 ppm for bus, and 11.5 ppm for pedestrians. In Northampton, UK
during the winter 1999, personal measurements of exposure to PMi0, PM2.s, and PMi were made
during walking and in-car journeys on two suburban routes.89 In-car measurements were highest
(43.16, 15.54, and 7.03 ug/m3 for PMi0, PM2.5, and PMI, respectively) followed by walking
(38.18, 15.06, and 7.14 ug/m3, respectively). Background levels were only available for PMio
(26.55 ug/m3), but were significantly lower than the walking exposure levels. Researchers found
similar results for CO exposure levels of schoolchildren commuters.90 So although personal
exposures are greater for in-vehicle commutes, pedestrians and bicyclists in proximity to heavy
traffic are exposed to elevated pollutant levels relative to background.
3.1.3.3 Concentrations and Exposure in Homes with Attached Garages
Residential indoor air quality is a major determinant of personal exposure, with most
people spending the majority of their time indoors at home. According to the National Human
Activity Pattern Survey, nationally, people spend an average of 16.68 hours per day indoors in a
residence.91 The large fraction of time spent in this microenvironment implies that sources that
impact indoor air are likely to have a substantial effect on personal exposure.
Indoor air quality is in large part determined by ventilation of indoor spaces. Natural
ventilation occurs as a result of two factors: wind-induced pressure and the "stack effect." The
latter occurs when hot air rises in a home, causing a pressure drop in the lower part of the home,
which then creates airflow into the home from higher-pressure locations outside the home.
Natural ventilation can also be influenced by opening of windows and doors. Mechanical
ventilation employs fans and sometimes ductwork to manage ventilation within a home.
Air can be drawn into a home from either outdoors, or in a home with an attached garage,
from the garage. Air from the garage can have higher concentrations of VOCs and other
pollutants as a result of the storage of vehicles, other engines and equipment, fuel (gasoline in
gas cans), solvents, or cleaning products. As a result, homes with a greater fraction of airflow
from the garage are more susceptible to air quality decrements from in-garage emissions.
Several studies have examined homes with attached garages to determine the fraction of
residential air intake from the garage. A recent study from Fairbanks, Alaska used
perfluorocarbon tracer (PFT) gases to estimate that 12.2% of air entering a mechanically
ventilated energy efficient home and 47.4% of the air entering the living spaces of an older
passively ventilated home originated in the homes' attached garages.92 In an Ann Arbor,
Michigan home, researchers used PFT gases to estimate that 16% of the air entering the home
entered through the garage.93 A recent study of a representative sample of homes in Anchorage,
Alaska employing PFT estimated that in homes with a forced air furnace in an attached garage,
36.7% of indoor air originated in the garage.94 In homes that had forced air furnaces indoors or
hytronic heat, 17.0% and 18.4% of indoor air originated in the garage, respectively. A study
from Minnesota examined homes constructed in 1994, 1998, and 2000.95 Homes built in 1994
had 17.4% of airflow originating in the garage. Homes built in 1998 and 2000 had 10.5% and
9.4% of airflow from the garage, respectively. In another study conducted in Ottawa, Ontario, an
3-22
-------
Final Regulatory Impact Analysis
average of 13% of home air intake came from the garage.96 That study also found that the
house-garage interface area was as leaky as the rest of the building envelope. In another study
from Washington, D.C., the house-garage interface was found to be 2.5 times as permeable as
the rest of the house.97 This discrepancy may indicate that homes built in colder climates are
built more tightly than homes in warmer regions as a result of weather-sealing. However, there
is no evidence that in regions with cold weather, colder temperatures lead to elevated indoor
concentrations of VOCs.98
Several studies have examined the influence of attached garages on indoor air and
personal exposure. In the 1980's researchers identified attached garages as a major source of
benzene and other VOCs in residences. The Total Exposure Assessment Methodology (TEAM)
Study was completed in 1985.99 The goal of this study was to develop methods to measure
individual total exposure (through air, food and water) and resulting body burden to toxic and
carcinogenic chemicals, and then to apply these methods with a probability-based sampling
framework to estimate the exposures and body burdens of urban populations in several U.S.
cities. The study measured personal exposures of 600 people to a number of air toxics. The
subjects were selected to represent residents of cities in New Jersey, North Carolina, North
Dakota, and California. In the study, a large fraction of an average nonsmoker's benzene
exposure originated from sources in attached garages.100 Work done as part of the TEAM study
also identified stored gasoline as an important source of elevated benzene levels indoors.101 This
stored gasoline can be found primarily in gas cans as well as the fuel tanks of lawn and garden
equipment, such as lawn mowers and string trimmers. Lawn and garden equipment fuel tank
emissions, however, are significantly lower than evaporative emissions from gas cans, because
the fuel tanks are much smaller than gas cans, typically 0.3 to 0.4 gallons. Emissions are also
higher from gas cans because vents and spouts are left open.
These early studies have highlighted the role of evaporative emissions within the garage
as contributors to indoor air pollution. Since then, major changes have affected emissions from
vehicles, including additional controls on evaporative emissions, on- board diagnostics, and state
inspection and maintenance programs addressing evaporative emission controls. Several
researchers have subsequently conducted air measurements in homes and in attached garages to
evaluate the effects on indoor air.
Garage concentrations of benzene and other VOCs are generally much higher than either
indoor or outdoor air, and constitute one of the highest-concentration microenvironments to
which a person might typically be exposed outside the occupational setting. The garage also
supplies contaminated air to the home to which it is attached. One recent study from Michigan
found average garage benzene concentrations of 36.6 ug/m3, with a standard deviation of 38.5
ug/m3, compared to mean and standard deviation concentrations of 0.4 ug/m3 and 0.12 ug/m3 in
ambient air.102 In Alaska, where fuel benzene levels tend to be very high and homes may be
built very airtight, garage concentrations have been measured at even higher levels. One study
from Anchorage measured average garage benzene concentrations of 103 ug/m3, with a standard
deviation of 135 ug/m3.103 More recently, a two-home study in Fairbanks found garage benzene
average concentrations of 119 ug/m3 during summer and 189 ug/m3 during winter in one well-
ventilated home with an air-to-air heat exchanger.104 In an older home with passive ventilation
summer and winter garage benzene concentrations were 421 and 103 ug/m3, respectively.
3-23
-------
Final Regulatory Impact Analysis
Other studies have studied the effect of garages or the sources within them on indoor air
quality. Most prominently, a group of Canadian investigators conducted source apportionment
of indoor non-methane hydrocarbons (NMHC) in 16 Ontario homes in the late 1990's.105 They
also assembled source profiles from hot soak and cold start emissions, which they used to
conduct source apportionment of total indoor air NMHC. All emissions samples and house
testing were conducted using the same 1993 model year vehicle. Overall, while the vehicle was
hot-soaking in the garage over a four hour sampling period, between 9 and 71% of the NMHC
inside the house could be attributable to that vehicle's emissions. Similarly, in the two hours
following a cold start event, between 13 and 85% of indoor NMHC could be attributed to the
vehicle cold start. Prior to the hot soak testing, average indoor benzene concentrations were 3.77
ug/m3, while during the hot soak, concentrations averaged 13.4 ug/m3. In the garage,
concentrations averaged 121 ug/m3 during the cold start. Prior to a cold start, indoor benzene
concentrations averaged 6.98 ug/m3, while for the two hours following cold start, concentrations
averaged 25.9 ug/m3. In the garage, concentrations averaged 422 ug/m3 over the two hours
following cold start.
The study also conducted real-time monitoring of CO and total hydrocarbons (THC)
within the house and garage. Overall, concentrations of CO and THC were relatively constant
during hot-soaks, but following a cold start, indoor concentrations of CO and THC tended to rise
sharply, and fall over the next two hours. This study provides direct evidence that a high fraction
of indoor NMHC (or VOCs) are directly attributable to emission events occurring in the garage.
Other studies have examined the influence of attached garages by comparing homes with
and without attached garages. In another study from Alaska, 137 Anchorage homes underwent
indoor air quality monitoring for benzene and other VOCs.106 Homes with attached garages had
significantly higher concentrations of indoor benzene compared to homes without attached
garages (70.8 ug/m3 vs. 8.6 ug/m3). In addition, elevated benzene indoors was also associated
with the presence of a vehicle in the garage, fuel being opened in the garage, and the use of
forced-air heaters.
In another Alaska study, concentrations of benzene and toluene in indoor air were found
to be not significantly associated with their urinary biomarkers, but indoor concentrations were
associated with the number of gasoline-powered engines stored in the garage.107 In a recent
follow-up to the study, ventilation patterns in two homes were evaluated using perfluorocarbon
tracers and a multi-zone indoor air quality model.108 In the study, average garage concentrations
were consistently elevated relative to the home. Furthermore, the study calculated the "virtual"
source strengths for benzene and toluene within the garage, and the garage was the only major
source of benzene within the home. Median garage source strengths for benzene ranged from
14-126mg/h.
Several population-based surveys have also found evidence of the influence of attached
garages. The National Human Exposure Assessment Survey (NHEXAS) Phase I pilot study in
Arizona was a representative exposure survey of the population. It found that in non-smoking
homes with attached garages, distribution of toluene concentrations indoors was shifted
1 flQ
significantly higher in homes with attached garages. Homes with attached garages had
3-24
-------
Final Regulatory Impact Analysis
median toluene levels of 24 ug/m3, while homes without garages had median concentrations of 5
ug/m3. The NHEXAS study in EPA Region 5 states was of similar design, but covering the
states of the upper Midwest. Using multivariate statistics, investigators found that VOCs
including benzene were associated with the storage of gasoline-powered equipment in an
attached garage.110
In one study from New Jersey, investigators evaluated the indoor air effects of a vehicle
fueled with "M85" - an 85% methanol, 15% gasoline blend - parking in the garage of a single
home.1U Testing was undertaken with both normally-functioning and malfunctioning
evaporative emissions controls, as well as with the HVAC system on and off. Garage benzene
concentrations exceeded indoor concentration by approximately 10-fold. Furthermore, the room
adjacent to the garage had substantially higher concentrations than a room on the opposite side of
the house. This study provides evidence that the garage is a major source of benzene inside the
house.
Appendix 3 A presents an EPA analysis of the effect of attached garages on indoor air
under various scenarios. This study was undertaken to evaluate the magnitude of exposure
underestimation using the national-scale exposure modeling techniques discussed above. Using
a mass balance model, steady-state concentrations of benzene were calculated as a function of
the concentration of air in the garage, the concentration of outdoor air, and the fraction of house
air intake from a garage. Data were obtained from studies discussed above. Because it is
unclear how well the homes studied to date represent the housing nationally, it is not currently
feasible to provide a highly precise estimate of the effect of attached garages on benzene
exposure nationally. Depending on how the available data are summarized, overall modeled
exposure concentrations would be expected to increase between 1.2 and 6.6 ug/m3 above average
inhalation exposure concentrations to benzene from ambient sources (1.4 ug/m3, as discussed in
Section 3.2). It should be noted that there is considerable uncertainty associated with this
estimated range, as discussed in Appendix 3 A.
Proposed reductions in fuel benzene content, new standards for cold temperature exhaust
emissions during vehicle starts, and reduced emissions from gas cans are all expected to
significantly reduce this major source of exposure.
3.1.3.4 Concentrations and Exposure in Parking Garages
Relatively limited air quality data for parking garages is available in the literature. The
following are results of air quality studies performed in parking garages, all of which indicate
that air toxics and criteria pollutants measured in these environments are substantially higher
than found in outdoor air. Because of the limited amount of data, we include results from some
non-U.S. studies, although differences in fuels and control technology limited their applicability
to the U.S.
In November 1990, a study of microenvironments, partially funded by the US EPA,
evaluated the potential range in concentrations of selected air toxics.112 Ten parking garages,
along with gasoline stations and office buildings, were randomly chosen for sampling since they
were among the least studied of the potentially important exposure microenvironments. The
principal air contaminants monitored were benzene, formaldehyde, and CO. Additional
3-25
-------
Final Regulatory Impact Analysis
compounds included toluene, xylenes, 1,2-dichloroethane, chloroform, carbon tetrachloride,
perchloroethylene, 1,1,1-trichloroethane, 1,3-butadiene, and trichloroethylene. The majority of
the compounds measured were significantly higher inside the garage compared to the ambient
sample. For example, the median 5-minute concentration of benzene was 67.1 ug/m3 in the
parking garage and 12.8 ug/m3 in ambient air. CO was 11000 ppb in the parking garage and
2000 ppb in ambient air. The researchers identified elevated levels of selected air toxics in
parking garages and pointed out the potential contribution from cold starts at the end of the work
day.
A more recent 2002 study was funded by The American Petroleum Institute to screen
"high-end" exposure microenvironments as required by section 21 l(b) of the Clean Air Act.113
An interim report is available. The study included measurements at underground parking
garages and surface parking lots in several cities. Air toxics quantified included hydrocarbons
(HCs), carbonyl compounds, BTEX, total VOC, and CO. When sampling at parking lot exits,
spikes in pollutant concentrations were observed when vehicles accelerated out of the parking
lot, while presumably prior to full catalyst warm-up. In underground garages, the levels of
BTEX and other compounds of interest varied with traffic level and reached concentrations that
were significantly higher than ambient levels outside the garage. The final report of the 21 l(b) is
expected in 2007.
A comparative study of indoor air quality in Hong Kong showed that the levels of CO,
NOx, and nonmethane hydrocarbons (NMHC) detected in a local park garage were the highest
among 13 other indoor sampling locations.114 The study did not specify the type or size of the
chosen parking garage, but indicated that it was located in an urban commercial area. High
indoor/outdoor ratios indicated that the air quality was mainly affected by indoor sources,
namely the vehicle exhaust. They also concluded that the pollution generated might cause health
hazards to the users and workers using such an environment.
Another assessment of the air quality in indoor park garages was performed in Hong
Kong in August through December 2000.115 Air samples were collected in two different garages
(an enclosed and semi-enclosed parking garage) as well as outdoors (within 10m of each
parking garage) and analyzed for one hundred different C3-C12 VOCs. Other compounds
measured included CO, CO2, PMio, and PM2.5. The CO levels in the enclosed garage were more
than in the semi-enclosed garage, and double the levels of the outdoor air. The PMio and PM2.5
concentrations were also found to be higher in the parking garage environments than outdoors.
High mass fractions of aliphatic and aromatic compounds detected in the enclosed garage
showed that fuel evaporation and motor vehicular exhaust were the major contributors to the
VOCs. The total concentrations of NMHC in the enclosed and semi-enclosed garages ranged
from 580 to 4610 ug/m3 and 43.1 to 175 ug/m3, respectively. The mean concentration of NMHC
measured in the enclosed garage (1910 ug/m3) was about 17 times higher than in the semi-
enclosed garage (94.6 ug/m3), and 3 times higher than measured at the outdoor sites. Not only
was the level of VOCs higher in the enclosed garage, but also the abundance of species
identified. The most abundant species in similar ranking order for both garages was toluene, 2-
methylbutane, w/p-xylenes, w-pentane, 2-methylpentane, w-hexane, and w-butane. Other major
gasoline components such as benzene, xylenes, and C4-C7 saturated HCs were also very high in
the enclosed garage. The difference between the two sites could be associated with the
3-26
-------
Final Regulatory Impact Analysis
ventilation and location, since the occupancy rates and fleet mixes were similar. The authors
also noted that the absence of sunlight in the enclosed garage would result in a slower or
negligible photochemical depletion rate of unsaturated hydrocarbons, and consequently an
increased abundance of the species observed.
In another study of multi-level parking garages in an Athens urban area, CO levels were
characterized in autumn 1999.116 Samples were collected at the exit sites (ramp where the flow
of vehicles was concentrated), the indoor site (first underground level where the majority of cars
parked), and immediately outside of each garage. Results indicate that CO levels varied
significantly over site, time, and day of measurement. The peak 1-hour value at the indoor sites
ranged from 22.9 to 109.3 ppm. At the indoor site, levels showed little variation and remained
high over time. The peak 1-hour value at the exit sites ranged from 8.9 to 57.3 ppm. At the exit
sites, 15-minute maximum concentrations were 5-15 times higher than the maximum recorded
CO level immediately outside the garage. CO levels on Saturday were much lower than a typical
weekday due to the reduced traffic, and weekday values were highest during the afternoon
sampling times (12:00-16:00 hour) corresponding with peak traffic volumes.
In Mumbai, India, ambient levels of benzene were determined during different seasons at
several different locations, including two parking areas.117 Parameters of the parking areas were
not specified, but 24-hour geometric means of benzene measured 117.4 and 74.2 ug/m3 during
the summer, 94.5 and 75.4 ug/m3 during the monsoon, and 148.0 and 703.0 ug/m3 during the
winter seasons, respectively. These values were considerably higher in comparison to less
heavily trafficked residential locations. The mean benzene concentrations of four different
residential locations ranged from 4.7 to 32.9 ug/m3, 1.9 to 33.5 ug/m3, and 4.7 to 18.8 ug/m3,
respectively, for the summer, monsoon, and winter seasons. The high concentrations in parking
areas were attributed to cold start-up emissions of engines.
A study in the UK of twelve underground parking garages identified high pollutant levels
of NOx, CO, CO2, BTEX, and PM.118 The parking garages selected covered a cross-section of
sizes (1 to 8 decks), ventilation system (natural and mechanical), designs (50 to 690 spaces), and
usages (business, shopping, and/or residential). Monitoring sites were located inside and at the
exit of the parking garage. The highest 15-minute average CO levels were measured at the exit
of parking garages, but a number of the parking garages had CO levels consistently higher inside
than at their exit. The NO2 measurements showed similar trends. Weekday benzene
concentration measurements averaged over one hour inside the parking garage and at the exit
ranged from 60 to 870 ug/m3 and 10 to 350 ug/m3, respectively.
In Madrid, Spain, atmospheric pollution produced by vehicles in parking garages was
studied.119 Two parking garages of different design were chosen for measurements of PMio,
lead, 12 PAHs, and CO. In both garages, CO, NO, TSP, and lead concentrations directly
correlated with vehicle traffic flow into and out of the garage. Also, higher values were observed
on the weekdays than during the weekend, for CO, NO, PAHs, and TSP in both garages. For
example, in one garage, the average daily TSP concentrations were 78-122 ug/m3 on the
weekdays versus 39 ug/m3 on the weekend, which was similar to outdoor city average
measurement (50 ug/m3). The researchers conclude that maximum concentrations for NO were
observed during maximum parking garage exits and therefore due to vehicle cold-starts. They
3-27
-------
Final Regulatory Impact Analysis
also conclude that the mechanical ventilation used in both garages was not sufficient to disperse
the pollutants emitted by the vehicles.
3.1.3.5 Concentrations and Exposure at Service Stations
Although there is relatively limited air quality data for service stations available currently
in the literature, the general consensus is that exposures to air toxics at service stations
significantly exceed ambient background levels. The studies below measure personal exposures
and concentrations during refueling either inside or outside of vehicles throughout the United
States. Several studies conducted outside of the United States chronicle similar results but are
not presented here due to differences in fuels and control technologies.
The TEAM study from the 1980's, described above, pumping gas and exposure to auto
exhaust were significantly associated with elevated benzene exposure. People who filled their
tanks with gasoline had twice as much benzene in their breath as people who did not. Estimated
concentrations at the breathing zone could exceed 1000 ug/m3 (100 times the ambient level),
based on the median breath benzene value measured (n=67) for those who had worked at or been
in a service station during the past 24 hours. Since this study, implementation of fuel controls,
onboard vapor recovery, and Stage II vapor recovery have changed emission and concentration
levels as discussed in Section 3.1.1.
In March 1990, another study randomly sampled 100 self-service filling stations
throughout Southern California along with samples at 10 parking garages and 10 offices nearby
those garages.120 The study took five-minute samples of 13 motor vehicle air pollutants (CO,
formaldehyde, and VOCs) in each microenvironment and in the ambient environment. The
median benzene concentration measured at the service stations was 28.8 ug/m3 with the
maximum reported value of 323 ug/m3. The median benzene concentration in ambient air was
significantly lower at 12.8 ug/m3.
A 1993 National Institute for Occupational Safety and Health (NIOSH) study assessed
benzene and MTBE concentrations and service station attendant exposures at service stations
101
with and without Stage II vapor recovery in Cincinnati, Phoenix, and Los Angeles. The mean
(and maximum) benzene exposure measurements were 96 (927), 160 (1662), and 192 (607)
ug/m3, respectively. The study found that Stage II vapor recovery did not significantly reduce
exposure to benzene during refueling. However, the efficiency of Stage II vapor recovery has
improved over the years. Northeast States for Coordinated Air Use Management (NESCAUM)
has suggested that Stage II vapor recovery systems are greater than 90% effective at capturing
MTBE and benzene vapors during refueling.122 These systems would therefore be expected to
reduce exposure beyond that shown in the NIOSH exposure assessment.
In March 1996 to July 1997, concentrations of MTBE, benzene, and toluene were
1 Ol
determined inside automobile cabins during fueling. Air samples were collected at service
stations in New Jersey, and the mean benzene in-cabin concentration was 54.3 ug/m3 (n=46).
The background concentration at the pump island measured 9.6 ug/m3 (n=36). The highest in-
cabin concentrations for all three pollutants occurred in a car that had a malfunctioning vapor
recovery system and in a series of cars sampled on an unusually warm, calm winter day when the
3-28
-------
Final Regulatory Impact Analysis
fuel volatility was high, the evaporation maximal, and the wind dispersion minimal. The in-
cabin concentrations were also typically higher when the car window was opened during the
entire fueling process.
In a study conducted between summer 1998 and spring 1999, self-service gas station
customers took part in a study to measure personal and breath concentrations of benzene at gas
stations in New Jersey.124 Benzene exposure concentrations during refueling (with a median
duration of three minutes) averaged 2.9 mg/m3 (SD = 5.8 mg/m3). Breath concentrations
averaged 160 ug/m3 (SD = 260 ug/m3). Breath benzene concentrations were significantly
correlated with refueling exposure concentrations, which was itself significantly associated with
refueling duration, time of year, and fuel octane grade.
Most recently, as discussed in the section on in-vehicle and parking garage exposure and
concentrations, a screening study of "high-end" exposure microenvironments was performed by
the American Petroleum Institute.60 The study included several vehicle-related
microenvironments in Houston and Atlanta during summer 2002. Among the various
microenvironments examined, the highest short-term concentrations occurred during refueling.
The in-vehicle average concentration of benzene measured during refueling was 46.0 ug/m3.
3.1.3.6 Occupational Exposure
Occupational settings can be considered a microenvironment in which exposure to
benzene and other air toxics can occur. Occupational exposures to benzene from mobile sources
or fuels can be several orders of magnitude greater than typical exposures in the non-
occupationally exposed population. Several key occupational groups are discussed below.
Occupations that involve fuel distribution, storage, and tank remediation lead to elevated
exposure to mobile-source related air toxics. Researchers published a review of benzene and
total hydrocarbon exposures in the downstream petroleum industry, including exposure data
from the past two decades among workers in the following categories: refinery, pipeline, marine,
rail, bulk terminals, tank truck drivers, service stations, underground storage tanks, tank cleaning,
and site remediation.125 The studies reviewed indicate that benzene exposure can range from <1
to more than 10 mg/m3, which is approximately three orders of magnitude higher than typical
non-occupational exposures (although there are occurrences of high benzene exposures in non-
occupational settings as well). This review is relevant because of the potential for fuel benzene
reductions to reduce their exposures as well. This statement is echoed by researchers in the
occupational literature.126 Occupational exposures in this range have been associated with
increased risk of certain leukemias in occupational epidemiology studies (Section 1.3.1).
Handheld and non-handheld equipment operators may also be exposed to elevated
concentrations of air toxics. As discussed below, several studies were conducted in work
categories employing small engine equipment, such as lawn and garden workers, workers in
construction/demolition, and others. Many of these occupations require the use of personal
protective equipment to prevent high exposures to carbon monoxide or other species. At present,
there are no representative samples of exposures among these categories. Non-occupational
exposures from these equipment types may also be important contributors to overall exposure.
3-29
-------
Final Regulatory Impact Analysis
EPA recently conducted a study of occupational exposures among lawn and garden workers
using riding tractors, walk-behind lawn mowers, string trimmers, and chainsaws.127 Results
demonstrated that equipment operators can experience highly variable exposures, with short-
term personal concentrations of CO and PM2.s ranging over two orders of magnitude. The study
also reported operator breathing-zone concentrations of formaldehyde and acetaldehyde that
were higher than background levels in all tests. This study illustrated the role of operator's
activity in affecting exposure levels to fuel-related air toxics.
Another study provides some insight into the possible range of benzene exposures in
workers who operate gasoline-powered engines, particularly those with 2-stroke engine
cycles.128 A study of snowmobile rider exposures in Sweden found benzene concentrations
ranging from under 10 ug/m3 to 2.5 mg/m3, a range of at least two orders of magnitude.
Exposures measured on riders on the back of the vehicle ranged from 0.7-0.8 mg/m3. These
measurements illustrate the potential for relatively high exposures when operating 2-stroke
equipment, as used in this study. Yellowstone National Park commissioned a study in 2002 to
examine occupational exposures of park employees to benzene, other VOCs, PMi0, and CO.129
Work shift benzene concentrations at a snowmobile entry gate 176.7 ug/m3, while snowmobile-
bound mobile patrol officers' exposure concentrations averaged 137.20 ug/m3. The highest
observed work shift concentration in the study was 514.1 ug/m3. At major sites of tourist interest
where snowmobiles parked, such as the Old Faithful geyser, concentrations averaged 41.3 to
48.8 ug/m3. 15-minute "peak" samples of workers' personal air ranged from 46.8 ug/m3 to 842.8
ug/m3. This study provides an indication of the variability of occupational benzene exposure
concentrations with time, and highlights the potential for elevated work shift exposures over
several hours.
A preliminary report published by the Northeast States for Coordinated Air Use
Management further illustrates the occupational impact of nonroad heavy-duty diesel
equipment. 13° In-cabin and work site perimeter measurements were collected for diesel
equipment emissions from the agricultural, construction (building and roadway), and lumber
industries in the Northeast. Initial results indicate that PM2.5 concentrations were 1-16 times
greater than the average ambient concentrations in each monitoring area. In-cabin exposures to
PM2.5 for operators ranged from 2 ug/m3 to over 660 ug/m3. Additionally, measured
concentrations of acetaldehyde, benzene, and formaldehyde were found to be significantly
elevated, although concentrations were not presented.
In one recently-published study of diesel exhaust exposures in a representative sample of
trucking terminals nationally, investigators applied structural equation modeling to data on
personal exposure to diesel exhaust (as elemental carbon).131 The study found that worker
exposure to elemental carbon depended on work area concentrations and worker tobacco use.
Work area concentrations depended on the size and type of the trucking terminal, whether the
work site was a mechanical shop, work site ventilation, and terminal yard concentrations.
Terminal yard concentrations in turn were related to local meteorology, the proximity of
interstate highways, surrounding industrial land uses, and region of the country. This study is
valuable in showing how personal occupational exposures are a complicated function of many
factors. Sophisticated statistical methods are needed to properly estimate models with highly
complex covariance structures.
3-30
-------
Final Regulatory Impact Analysis
In addition, some occupations require that workers spend considerable time in vehicles,
which increases the time they spend in a higher-concentration microenvironment. In-vehicle
concentrations are discussed in Section 3.1.3.2.1 above.
3.1.4 Uncertainties in Air Toxics Measurements
A number of uncertainties limit our ability to fully describe the impacts of motor vehicle
emissions. As described above, most people in the U.S. experience some level of exposure to
emissions from motor vehicles. Thus, proper characterization of the level of these exposures is
critical. However, the exposure assessment techniques used may not adequately represent the
populations' true exposures to motor vehicle emissions.
Air quality and exposure measurements are expensive and therefore are limited. The
high costs of measurement techniques affect the quantity of samples that can be collected and
quantity of compounds that can be identified. As a result, measurements may only occur at
central monitoring sites, rather than in microenvironments impacted by motor vehicle emissions
or in personal breathing zones. Air quality monitoring at these central sites often do not
represent actual exposures, especially for populations living near roads or with substantial
occupational exposure.
Monitoring samples are often integrated and therefore lack time resolution. This can
result in difficulty in determining source contributions. Additionally, some compounds are hard
to measure accurately. For example, 1,3-butadiene is very reactive in the ambient atmosphere
and has a short atmospheric lifetime, estimated to be only two hours.132 Thus, this compound
can easily break down before samples are analyzed. Also, a vapor pressure of 3.3 atm at 25°C
makes it a very volatile compound. Secondary reactions are a confounding factor in air quality
measurements and can add additional uncertainty to measured ambient concentrations.
Personal exposure monitoring provides greater realism in describing a person's actual
exposure to air toxics. However, given the limitations on size of equipment, detection limits in
personal exposure monitoring studies are sometimes greater than those found in studies using
other techniques.
3.2 Modeled Air Quality, Exposures, and Risks for Air Toxics
3.2.1 National-Scale Modeled Air Quality, Exposure, and Risk for Air Toxics
EPA assesses human health impacts from outdoor, inhalation, chronic exposures to air
toxics in the National-Scale Air Toxics Assessment (NAT A). It assesses lifetime risks assuming
continuous exposure to levels of air toxics estimated for a particular point in time. The most
recent NATA was done for the year 1999.133 It had four steps:
1) Compiled a national emissions inventory of air toxics emissions from outdoor sources.
The 1999 National Emissions Inventory is the underlying basis for the emissions
information in the 1999 assessment.
3-31
-------
Final Regulatory Impact Analysis
2) Estimated ambient concentrations based on emissions as input to an air dispersion
model (the Assessment System for Population Exposure Nationwide, or ASPEN
model).134
3) Estimated population exposures based on a screening-level inhalation exposure model
(Hazardous Air Pollutant Exposure Model, version 5, or HAPEM5) and the estimated
ambient concentrations (from the ASPEN model) as input to the exposure model.135
4) Characterized 1999 potential public health risks due to inhalation of air toxics. This
included cancer and noncancer effects, using available information on air toxics health
effects, current EPA risk assessment and risk characterization guidelines, and estimated
population exposures.136
For this final rule, we have conducted air quality, exposure and risk modeling for the
years 1999, 2015, 2020, and 2030, using the same general approach as the 1999 NAT A. We
modeled all the pollutants in Table 2.2-1 for both the reference case, which includes all control
programs currently planned by EPA in regulations, and the control case, which includes the
cumulative impacts of the standards proposed in this rule. These pollutants
• Are on EPA's list of hazardous air pollutants in Section 112 of the Clean Air Act
• Are emitted by mobile sources
• Are included in the National Emissions Inventory
• Are included in the 1999 NATA
Note that the modeling did not include diesel PM and diesel exhaust organic gases. EPA has
previously done future-year projections of the mobile source contribution to air toxics
concentrations, exposure, and risk for selected air toxics,137'138'139' 140but prior to the proposal
for this rule, had never done a comprehensive assessment that includes projections for all mobile
source air toxics, as well as the stationary source contribution for those pollutants. It should be
noted that the reference case assessment results developed for the proposal have been published
in a peer reviewed journal article.141
As discussed in Chapter 2, a number of major revisions to inventory methodology have
been made relative to what was done for both the 1999 NATA, and air quality exposure and risk
modeling for the proposal. These include revisions to cold start emissions, use of NMIM2005
for nonroad equipment, addition of portable fuel container emissions, and changes to gasoline
distribution inventories. Also, this final rule modeling for 1999 does not include data submitted
by States for the 1999 NEI. In addition, the modeling for the final rule relied on an updated
version of the HAPEM model, HAPEM6.142 HAPEM6 improves on HAPEM5 by accounting
for the spatial variability of outdoor concentrations of air toxics within a census tract due to
higher outdoor concentrations at locations near major roadways. Other improvements to
HAPEM are discussed in section 3.2.1.2.1. This modeling work is discussed in more detail in an
EPA technical report, "National Scale Modeling of Air Toxics for the Final Mobile Source Air
Toxics Rule; Technical Support Document," Report Number EPA-454/R-07-002. It should be
noted that the control case modeling accounted only for the 0.62 percent standard, but not the 1.3
vol% maximum average. Thus, the emission reductions from highway vehicles and other
sources attributable to the fuel benzene standard are underestimated in many areas of the
country, particularly in areas where fuel benzene levels were highest without control, such as the
Northwest.
3-32
-------
Final Regulatory Impact Analysis
The NATA modeling framework has a number of limitations which prevent its use as the
sole basis for setting regulatory standards. Even so, this modeling framework is very useful in
identifying air toxic pollutants and sources of greatest concern, setting regulatory priorities, and
informing the decision making process.
Among the significant limitations of the framework is that it cannot be used to identify
ambient "hot spots," as mobile sources are not represented explicitly as roads or other locations
of mobile source activity. In addition, this kind of modeling assessment cannot address the kinds
of questions an epidemiology study might allow, such as the relationship between asthma or
cancer risk, and proximity of residences to point sources, roadways and other sources of air
toxics emissions. The framework also does not account for risk from potentially significant
sources of air toxics originating indoors, such as stoves or out-gassing from building materials or
evaporative benzene emissions from cars in attached garages. The ASPEN model performs well
for some pollutants, but has also been shown to systematically underestimate pollutant
concentrations relative to measured levels for certain pollutants such as metals and some reactive
compounds. The cancer unit risk estimates for most pollutants are "upper bound," meaning they
probably lead to overestimates of risk. It should be noted, however, that the unit risk estimate for
benzene is a maximum likelihood estimate, which is a best scientific estimate. The above
limitations are discussed in detail in Section 3.2.1.4.
Although we do not use it in this modeling, another tool that EPA uses to assess
distributions of concentrations of air toxics at the national scale is the Community Multiscale Air
Quality (CMAQ) modeling system.143 CMAQ can account for photochemical destruction and
production, deposition and regional transport of toxic air pollutants, and thus can be used to
predict the concentrations of HAPs with significant atmospheric production. In general,
predicted concentrations of air toxics from CMAQ were within a factor of 2 of measured values,
with a tendency to underpredict measured ambient concentrations.144 CMAQ underpredicts
monitored benzene levels more than ASPEN, because ASPEN values contain a large, added-on
concentration based on monitored values of benzene. CMAQ has sophisticated photochemistry,
but does not yet have the spatial resolution of dispersion models such as ASPEN, and thus
accounts for less of the total variability in levels of air toxics with localized concentration
gradients, such as benzene.145 Finally, CMAQ is requires more computational resources, which
makes it more difficult to use for evaluating trends in a large number of air toxics over many
years or impacts of control scenarios.
Details of the methods used and presentation of key results are discussed in the following
sections. Results do not account for other potentially significant sources of inhalation exposure,
such as benzene emissions from sources in attached garages (such as vehicles, snowblowers,
lawnmowers and gas cans).
3-33
-------
Final Regulatory Impact Analysis
3.2.1.1 Air Quality Modeling
3.2.1.1.1 Methods
Prior to performing air quality modeling of the projected emissions, the emissions from
the stationary and mobile inventories (discussed in Chapter 2) are processed in the Emissions
Modeling System for Hazardous Air Pollutants (EMS-HAP) Version 3 to create the emissions
input files used by ASPEN to calculate air quality concentrations.146 In addition to projecting
stationary and area source emissions to future years for some source categories, EMS-HAP
spatially allocates emissions inventoried at the county level to the census tract level, and
temporally allocates them to eight three-hour time periods throughout the day. Once the
emissions are processed, they are input into ASPEN to calculate air quality concentrations. In
addition to the emissions, ASPEN uses meteorological parameters and census tract centroid
locations for concentration calculations. ASPEN estimates do not account for day-of-week or
seasonal variations in emissions. The ASPEN model takes into account important determinants
of pollutant concentrations, such as: rate of release, location of release, the height from which the
pollutants are released, wind speeds and directions from the meteorological stations nearest to
the release, breakdown of the pollutants in the atmosphere after being released (i.e., reactive
decay), settling of pollutants out of the atmosphere (i.e., deposition), and transformation of one
pollutant into another. The model first estimates concentrations at receptors arranged in rings
around emission sources up to 50 kilometers away. The model then interpolates concentrations
to census tract centroids. For 1999, meteorological conditions in 1999 and 2000 census tract
data were used.
In using ASPEN to estimate projected concentrations in 2015, 2020, and 2030 for this
final rule, the same meteorology and census tract locations were used as for the 1999 NAT A.
Details of how ASPEN processed emissions data are provided in the technical document,
"National-Scale Modeling of Mobile Source Air Toxic Emissions, Air Quality, Exposure and
Risk for the Mobile Source Air Toxics Final Rule." ASPEN only accounts for sources within a
50-kilometer radius of each source when calculating ambient concentrations. Thus, the
contribution to ambient levels of air toxics from sources further away than 50-kilometers, as well
as the contribution of uninventoried sources, is addressed through the addition of a "background"
term.147 Mobile source pollutants which include a background component are 1,3-butadiene,
acetaldehyde, benzene, formaldehyde, and xylenes. Each of the three projection years used the
same 1999-based background. However, background levels are likely to change with emissions.
Thus, for the proposal, a sensitivity analysis was done to evaluate the potential impact of not
changing the background concentration (see Section 3.2.1.4).
It should be noted that in the control case scenarios, we have modeled the cumulative
impacts on air quality, exposure, and risk for all of the programs finalized today, not the impacts
of individual programs. Were we to model each program individually, we anticipate that
changes in air quality, exposure, and risk would track the patterns of emission changes closely.
Also, for the final rule, we estimated the contribution of secondary formation to ambient
concentrations of MSATs by applying ratios of secondary to primary concentrations from 1999
NATA to the modeled primary concentrations for this rule. This is different from the approach
3-34
-------
Final Regulatory Impact Analysis
used in the proposal where we projected precursor emissions and then modeled secondary
formation. When we applied the ratio approach to the proposal's primary concentrations, the
results were very similar to the full modeling approach (see Section 3.2.1.3). The comparisons
are discussed in the technical document cited above.
We estimated the contributions to ambient concentrations for the following source
sectors: major, area and other, onroad, nonroad, and background.13
3.2.1.1.2 Air Quality Trends for Air Toxics: Reference Case
Table 3.2-1 summarizes nationwide mean census tract ambient concentrations, without
the controls being finalized in this rule, of mobile source air toxics in 1999 and projection years
for the following source sectors: major sources, area and other sources, highway vehicles,
nonroad sources, and background. The behavior of benzene is typical of the projected trends.
Over 90% of the mobile source contribution to ambient benzene levels is attributable to gasoline
vehicles and engines. Figure 3.2-1 depicts the trend in nationwide average census tract
concentrations of benzene over this time period. The mobile source contribution to ambient
benzene concentrations is projected to decrease over 40% by 2015, with a decrease in ambient
benzene concentration from all sources of about 25%. Subsequently, increases in vehicle miles
traveled (VMT) are projected to produce increasing concentrations. Summary tables providing
data by State, and for reformulated and non-reformulated (i.e., conventional) gasoline areas, can
be found in the docket for this rule. Due to greater population and vehicle activity, the average
ambient benzene concentration in 1999 is much higher for counties in reformulated gasoline
areas than non-reformulated gasoline areas - about 1.9 |ig/m3 versus 1.2 |ig/m3. However the
percent reduction in average 2015 ambient concentration is similar regardless of fuel type - 22%
for non-reformulated gasoline counties versus 29% for reformulated gasoline counties.
b Major and "area and other" are stationary source emission sectors. Major sources, as defined by the Clean Air Act,
are those stationary facilities that emit or have the potential to emit 10 tons of any one toxic air pollutant or 25 tons
of more than one toxic air pollutant per year. Area and other sources include sources that generally have smaller
emissions on an individual basis than "major sources" and are often too small or ubiquitous in nature to be
inventoried as individual sources. "Area sources" include facilities that have air toxics emissions below the major
source threshold as defined in the air toxics sections of the Clean Air Act and thus emit less than 10 tons of a single
toxic air pollutant or less than 25 tons of multiple toxic air pollutants in any one year. Area sources include smaller
facilities, such as dry cleaners. "Other sources" include sources such as wildfires and prescribed burnings that may
be more appropriately addressed by other programs rather than through regulations developed under certain air
toxics provisions (section 112 or 129) in the Clean Air Act. For example, wildfires and prescribed burning are being
addressed through the burning policy agreed to by the Interim Federal Wildland Policy. "Background" includes
emissions from transport and uninventoried sources.
3-35
-------
Final Regulatory Impact Analysis
Table 3.2-1. Mean Ambient Concentrations of Mobile Source Air Toxics in 1999, 2015, 2020, and 2030, Without Controls in
this Rule.
Pollutant
1,3-Butadiene
2,2,4-Trimethylpentane
Acetaldehyde
Acrolein
Benzene
Chromium III
Chromium VI
Ethyl Benzene
Formaldehyde
Hexane
MTBE
Manganese
Naphthalene
Nickel
POM
Propionaldehyde
Styrene
Toluene
Xylenes
background
(MS m 3)
5.10E-02
O.OOE+00
5.17E-01
O.OOE+00
3.94E-01
O.OOE+00
O.OOE+00
O.OOE+00
7.62E-01
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
1.70E-01
1999 average concentrations (jig in 3)
major
1.97E-03
2.16E-02
2.94E-02
3.21E-03
2.20E-02
8.22E-04
1.07E-04
1.84E-02
3.99E-02
6.68E-02
1.30E-02
2.71E-03
4.56E-03
7.76E-04
4.93 E-03
1.01E-02
2.52E-02
2.03E-01
9.98E-02
area &
other
2.05E-02
2.32E-02
5.49E-02
2.93E-02
1.40E-01
4.53E-04
1.98E-04
9.00E-02
8.77E-02
4.30E-01
6.04E-02
2.22E-03
4.11E-02
1.42E-03
1.61E-02
2.33E-02
1.40E-02
8.05E-01
5.59E-01
onroad
5.20E-02
7.29E-01
6.78E-01
5.63E-02
6.89E-01
3.22E-05
2.15E-05
2.73E-01
4.65E-01
2.34E-01
4.00E-01
1.73E-05
1.46E-02
3.96E-05
1.73 E-03
1.68E-01
2.98E-02
1.81E+00
1.01E+00
nonroad
1.81E-02
1.96E-01
1.47E-01
2.27E-02
1.77E-01
5.53E-05
1.25E-05
9.73E-02
2.21E-01
8.56E-02
4.04E-01
5.46E-06
4.36E-03
9.98E-05
8.60E-04
4.27E-02
3.65E-03
4.18E-01
3.99E-01
total
(including
background)
1.44E-01
9.70E-01
1.43E+00
1.11E-01
1.42E+00
1.36E-03
3.39E-04
4.79E-01
1.58E+00
8.17E-01
8.77E-01
4.95E-03
6.46E-02
2.33E-03
2.37E-02
2.45E-01
7.27E-02
3.24E+00
2.23E+00
2015 annual average concentrations (jig m"3)
major
2.17E-03
1.09E-02
2.97E-02
3.53E-03
1.55E-02
1.04E-03
1.36E-04
1.24E-02
4.98E-02
5.94E-02
1.38E-02
3.23 E-03
3.97E-03
8.87E-04
3.79E-03
9.31E-03
3.00E-02
1.43E-01
8.22E-02
area &
other
2.05E-02
2.69E-02
5.71E-02
2.62E-02
1.63E-01
6.16E-04
2.72E-04
1.19E-01
9.82E-02
5.21E-01
6.52E-02
2.92E-03
5.01E-02
1.62E-03
1.86E-02
2.39E-02
1.89E-02
1.06E+00
7.60E-01
onroad
2.28E-02
3.66E-01
3.86E-01
2.42E-02
3.79E-01
4.40E-05
2.94E-05
1.35E-01
1.92E-01
1.16E-01
1.05E-01
2.36E-05
7.90E-03
5.43E-05
9.13E-04
8.24E-02
1.50E-02
9.00E-01
4.98E-01
nonroad
1.08E-02
1.15E-01
1.10E-01
1.81E-02
1.14E-01
5.85E-05
1.32E-05
5.66E-02
1.63E-01
5.93E-02
1.08E-01
6.46E-06
4.49E-03
1.15E-04
7.66E-04
2.83E-02
2.18E-03
2.50E-01
2.18E-01
total
(including
background)
1.07E-01
5.19E-01
1.10E+00
7.20E-02
1.07E+00
1.76E-03
4.50E-04
3.24E-01
1.27E+00
7.56E-01
2.93 E-01
6.17E-03
6.65E-02
2.67E-03
2.40E-02
1.44E-01
6.61E-02
2.35E+00
1.73E+00
-------
Final Regulatory Impact Analysis
Table 3.2-1 (cont'd). Mean Ambient Concentrations of Mobile Source Air Toxics in 1999, 2015, 2020, and 2030, Without
Controls in this Rule.
Pollutant
1,3-Butadiene
2,2,4-Trimethylpentane
Acetaldehyde
Acrolein
Benzene
Chromium III
Chromium VI
Ethyl Benzene
Formaldehyde
Hexane
MTBE
Manganese
Naphthalene
Nickel
POM
Propionaldehyde
Styrene
Toluene
Xylenes
background
(MS m 3)
5.10E-02
O.OOE+00
5.17E-01
O.OOE+00
3.94E-01
O.OOE+00
O.OOE+00
O.OOE+00
7.62E-01
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
1.70E-01
2020 annual average concentrations (jig m 3)
major
2.34E-03
1.17E-02
3.10E-02
3.96E-03
1.70E-02
1.17E-03
1.54E-04
1.39E-02
5.65E-02
6.53E-02
1.55E-02
3.59E-03
4.46E-03
9.61E-04
4.21E-03
9.35E-03
3.44E-02
1.60E-01
9.29E-02
area &
other
2.05E-02
2.84E-02
5.83E-02
2.54E-02
1.69E-01
6.96E-04
3.07E-04
1.31E-01
1.03E-01
5.62E-01
6.67E-02
3.21E-03
5.32E-02
1.78E-03
1.90E-02
2.45E-02
2.09E-02
1.16E+00
8.38E-01
onroad
2.37E-02
3.66E-01
3.98E-01
2.50E-02
3.88E-01
4.84E-05
3.23E-05
1.35E-01
1.97E-01
1.07E-01
8.48E-02
2.60E-05
7.86E-03
5.97E-05
9.47E-04
8.45E-02
1.57E-02
9.11E-01
5.04E-01
nonroad
1.14E-02
1.14E-01
1.09E-01
1.91E-02
1.18E-01
5.90E-05
1.34E-05
5.78E-02
1.64E-01
6.13E-02
1.12E-01
6.83E-06
4.80E-03
1.20E-04
7.71E-04
2.78E-02
2.21E-03
2.50E-01
2.18E-01
total
(including
background)
1.09E-01
5.20E-01
1.11E+00
7.34E-02
1.09E+00
1.97E-03
5.07E-04
3.38E-01
1.28E+00
7.96E-01
2.79E-01
6.83E-03
7.03E-02
2.92E-03
2.49E-02
1.46E-01
7.32E-02
2.48E+00
1.82E+00
2030 annual average concentrations (jig m"3)
major
2.34E-03
1.17E-02
3.10E-02
3.96E-03
1.70E-02
1.17E-03
1.54E-04
1.39E-02
5.65E-02
6.53E-02
1.55E-02
3.59E-03
4.46E-03
9.61E-04
4.21E-03
9.35E-03
3.44E-02
1.60E-01
9.29E-02
area &
other
2.05E-02
2.84E-02
5.83E-02
2.54E-02
1.69E-01
6.96E-04
3.07E-04
1.31E-01
1.03E-01
5.62E-01
6.67E-02
3.21E-03
5.32E-02
1.78E-03
1.90E-02
2.45E-02
2.09E-02
1.16E+00
8.38E-01
onroad
2.78E-02
4.24E-01
4.69E-01
2.94E-02
4.54E-01
5.94E-05
3.96E-05
1.57E-01
2.31E-01
1.18E-01
8.42E-02
3.19E-05
9.11E-03
7.34E-05
1.12E-03
9.84E-02
1.85E-02
1.06E+00
5.86E-01
nonroad
1.30E-02
1.24E-01
1.18E-01
2.18E-02
1.32E-01
6.04E-05
1.37E-05
6.45E-02
1.80E-01
6.87E-02
1.25E-01
7.59E-06
5.51E-03
1.31E-04
8.57E-04
2.99E-02
2.47E-03
2.75E-01
2.40E-01
total
(including
background)
1.15E-01
5.88E-01
1.19E+00
8.05E-02
1.17E+00
1.98E-03
5.15E-04
3.66E-01
1.33E+00
8.14E-01
2.92E-01
6.84E-03
7.23E-02
2.95E-03
2.52E-02
1.62E-01
7.63E-02
2.65E+00
1.93E+00
-------
Final Regulatory Impact Analysis
Figure 3.2-1. Nationwide Average Benzene Concentration, 1999-2030, Without Controls in
this Rule.
1.60
1.40
1.20
1.00
0.80
0.60
0.40
0.20
0.00
1999
2015
2020
2030
Year
3.2.1.1.3 Distributions of Air Toxic Concentrations across the U. S.: Reference Case
Table 3.2-2 gives the distribution of census tract concentrations, summed across all
source sectors and background, for mobile source air toxics across the nation in 2020, absent the
controls being finalized in this rule. Distributions for other years are similar. Summary tables
providing distributions for other years, as well as distributions by State and for reformulated and
non-reformulated gasoline areas, can be found in the docket for this rule. From this table, it can
be seen that 95th percentiles of average census tract concentrations for mobile-source dominated
pollutants such as benzene and 1,3-butadiene are typically two to five times higher than the
median of census tract concentrations, even though mobile source emissions are widely
dispersed. For pollutants with large major source contributions (e.g., manganese), the 95th
percentile of census tract averages can be much higher than the median. In addition, average
census tract concentrations can span one to several orders of magnitude. Thus, there is
considerable variation in average concentrations across the U.S.
Figure 3.2-2 depicts the geographic distribution of county median concentrations of
benzene in 2020. Relatively high levels are seen in the Northeast, Southern California, Florida,
parts of Texas, and the Great Lakes Region, where there is high population density and thus high
vehicle and nonroad equipment activity. Relatively high levels are also seen in the Pacific
Northwest, parts of Alaska, and the upper Great Lakes region. Analysis of fuel survey data
3-38
-------
Final Regulatory Impact Analysis
Table 3.2-2. National Distribution of Census Tract Concentrations for Mobile Source Air
Toxics in 2020, Without Controls in this Rule.
Pollutant
1,3-Butadiene
2,2,4-Trimethylpentane
Acetaldehyde
Acrolein
Benzene
Chromium III
Chromium VI
Ethyl Benzene
Formaldehyde
Hexane
MTBE
Manganese
Naphthalene
Nickel
POM
Propionaldehyde
Styrene
Toluene
Xylenes
2020 concentration (ng m 3) distribution
5th
percentile
3.03E-03
3.83E-02
5.45E-01
6.04E-03
3.42E-01
5.73E-06
3.52E-06
2.04E-02
4.08E-01
3.27E-02
3.34E-03
1.33E-05
2.88E-03
1.38E-05
1.72E-03
1.24E-02
2.52E-03
1.54E-01
2.66E-01
10th
percentile
5.60E-03
7.00E-02
5.82E-01
9.78E-03
4.15E-01
1.52E-05
8.79E-06
3.79E-02
5.29E-01
6.16E-02
7.88E-03
4.35E-05
5.91E-03
3.80E-05
2.94E-03
2.13E-02
4.88E-03
2.83E-01
3.43E-01
25th
percentile
3.12E-02
1.74E-01
6.99E-01
2.09E-02
6.33E-01
6.40E-05
3.56E-05
1.01E-01
8.08E-01
1.90E-01
2.39E-02
2.04E-04
1.86E-02
1.67E-04
5.73E-03
4.81E-02
1.23E-02
7.34E-01
6.35E-01
Median
8.36E-02
3.79E-01
9.41E-01
4.41E-02
9.37E-01
2.41E-04
1.22E-04
2.30E-01
1.16E+00
4.76E-01
7.22E-02
8.68E-04
4.48E-02
6.65E-04
1.19E-02
1.07E-01
2.70E-02
1.64E+00
1.22E+00
75th
percentile
1.30E-01
6.80E-01
1.29E+00
8.64E-02
1.32E+00
7.31E-04
3.32E-04
4.06E-01
1.52E+00
8.93E-01
2.44E-01
3.53E-03
8.82E-02
2.01E-03
2.08E-02
1.93E-01
5.39E-02
2.96E+00
2.06E+00
90th
percentile
1.98E-01
1.12E+00
1.84E+00
1.71E-01
1.90E+00
2.34E-03
9.08E-04
6.70E-01
2.12E+00
1.70E+00
8.80E-01
1.42E-02
1.67E-01
4.78E-03
3.62E-02
3.26E-01
1.06E-01
5.31E+00
3.61E+00
95th
percentile
3.28E-01
1.50E+00
2.49E+00
2.71E-01
2.36E+00
4.89E-03
1.55E-03
9.60E-01
2.67E+00
2.81E+00
1.30E+00
2.10E-02
2.37E-01
8.17E-03
5.78E-02
4.33E-01
1.75E-01
7.43E+00
5.38E+00
indicate higher than average fuel benzene levels in these areas. These areas also have higher
benzene emissions in winter due to cold starts. Higher benzene levels in Idaho are not due to
fuel benzene levels, but are primarily due to wildfire emission estimates, which were determined
to be an error in the 1999 National Emissions Inventory and the subsequent projections.
Similar benzene median county concentration maps for 1999, 2015, and 2030 can be
found in the docket for this rule, along with maps for other mobile source air toxics and tables of
concentration distributions.
3.2.1.1.4
Impacts of Controls on Ambient Concentrations
The standards being finalized in this rule will substantially reduce ambient concentrations
of air toxics across the United States. As noted above, these results reflect the cumulative effects
of all of the programs finalized in today's rule, not the individual programs. Table 3.2-3 shows
the reduction in nationwide average census tract concentrations of MSATs from all sources in
2015, 2020 and 2030. Table 3.2-4 shows the reduction in the highway vehicle contribution to
nationwide average census tract concentrations of MSATs. Table 3.2-5 shows that in 2030, the
highway vehicle portion of ambient benzene concentrations will be reduced almost 45% across
the U.S., the nonroad equipment contribution will be reduced about 10%, and
3-39
-------
Final Regulatory Impact Analysis
Figure 3.2-2. Geographic Distribution of County Median Concentrations (ug/m ) of
Benzene in 2020 Without Controls in this Rule.
0.063 - 0.338
0.339 - 0.567
0.568 - 0.880
0.881 -1.316
1.317-2.114
2.115-4.929
-------
Final Regulatory Impact Analysis
Table 3.2-3. Nationwide Average Census Tract Concentrations of MSATs, With and Without Controls in this Rule, 2015,
2020, and 2030.
2015 2020 2030
Reference Control % Reduction Reference Control % Reduction Reference Control % Reduction
1,3-Butadiene
2,2,4-Trimethylpentane
Acetaldehyde
Acrolein
Benzene
Chromium III
Chromium VI
Ethyl Benzene
Formaldehyde
Hexane
MTBE
Manganese
Naphthalene
Nickel
POM
Propionaldehyde
Styrene
Toluene
Xylenes
1.07E-01
5.19E-01
1.10E+00
7.20E-02
1.07E+00
1.76E-03
4.50E-04
3.24E-01
1.27E+00
7.56E-01
2.93E-01
6.17E-03
6.65E-02
2.67E-03
2.40E-02
1.44E-01
6.61E-02
2.35E+00
1.73E+00
1.03E-01
4.53E-01
1.04E+00
6.79E-02
9.56E-01
1.76E-03
4.50E-04
2.99E-01
1.24E+00
7.37E-01
2.82E-01
6.17E-03
6.65E-02
2.67E-03
2.40E-02
1.33E-01
6.33E-02
2.18E+00
1.64E+00
3.6
12.7
5.8
5.7
10.3
0.0
0.0
7.5
2.3
2.5
3.5
0.0
0.0
0.0
0.0
7.8
4.3
7.1
5.3
1.09E-01
5.20E-01
1.11E+00
7.34E-02
1.09E+00
1.97E-03
5.07E-04
3.38E-01
1.28E+00
7.96E-01
2.79E-01
6.83E-03
7.03E-02
2.92E-03
2.49E-02
1.46E-01
7.32E-02
2.48E+00
1.82E+00
1.03E-01
4.19E-01
1.01E+00
6.69E-02
9.38E-01
1.97E-03
5.07E-04
3.01E-01
1.24E+00
7.70E-01
2.66E-01
6.83E-03
7.03E-02
2.92E-03
2.49E-02
1.28E-01
6.87E-02
2.22E+00
1.68E+00
5.7
19.5
9.1
8.9
13.6
0.0
0.0
11.1
3.6
3.2
4.6
0.0
0.0
0.0
0.0
12.2
6.2
10.4
7.8
1.15E-01
5.88E-01
1.19E+00
8.05E-02
1.17E+00
1.98E-03
5.15E-04
3.66E-01
1.33E+00
8.14E-01
2.92E-01
6.84E-03
7.23 E-02
2.95E-03
2.52E-02
1.62E-01
7.63E-02
2.65E+00
1.93E+00
1.04E-01
4.26E-01
1.03E+00
6.97E-02
9.50E-01
1.98E-03
5.15E-04
3.07E-01
1.26E+00
7.76E-01
2.74E-01
6.84E-03
7.23E-02
2.95E-03
2.52E-02
1.33E-01
6.89E-02
2.24E+00
1.70E+00
9.0
27.6
13.7
13.4
18.5
0.0
0.0
16.3
5.6
4.7
6.0
0.0
0.0
0.0
0.0
18.0
9.7
15.7
11.8
5-41
-------
Final Regulatory Impact Analysis
Table 3.2-4. Nationwide Highway Vehicle Contribution to Average Census Tract Concentrations of MSATs, With and
Without Controls in this Rule, 2015, 2020, and 2030.
2015 2020 2030
% %
Reference Control Reduction Reference Control % Reduction Reference Control Reduction
1,3-Butadiene
2,2,4-Trimethylpentane
Acetaldehyde
Acrolein
Benzene
Chromium III
Chromium VI
Ethyl Benzene
Formaldehyde
Hexane
MTBE
Manganese
Naphthalene
Nickel
POM
Propionaldehyde
Styrene
Toluene
Xylenes
2.28E-02
3.66E-01
3.86E-01
2.42E-02
3.79E-01
4.40E-05
2.94E-05
1.35E-01
1.92E-01
1.16E-01
1.05E-01
2.36E-05
7.90E-03
5.43E-05
9.13E-04
8.24E-02
1.50E-02
9.00E-01
4.98E-01
1.89E-02
3.06E-01
3.22E-01
2.01E-02
2.83E-01
4.40E-05
2.94E-05
1.14E-01
1.62E-01
1.05E-01
1.01E-01
2.36E-05
7.90E-03
5.43E-05
9.13E-04
7.12E-02
1.22E-02
7.47E-01
4.14E-01
17.0
16.3
16.5
17.0
25.3
0.0
0.0
16.0
15.3
9.8
4.3
0.0
0.0
0.0
0.0
13.6
18.8
17.1
16.9
2.37E-02
3.66E-01
3.98E-01
2.50E-02
3.88E-01
4.84E-05
3.23 E-05
1.35E-01
1.97E-01
1.07E-01
8.48E-02
2.60E-05
7.86E-03
5.97E-05
9.47E-04
8.45E-02
1.57E-02
9.11E-01
5.04E-01
1.74E-02
2.71E-01
2.97E-01
1.85E-02
2.55E-01
4.84E-05
3.23E-05
1.01E-01
1.50E-01
8.89E-02
7.77E-02
2.60E-05
7.86E-03
5.97E-05
9.47E-04
6.66E-02
1.12E-02
6.66E-01
3.69E-01
26.3
25.9
25.4
26.2
34.2
0.0
0.0
25.6
23.6
16.9
8.3
0.0
0.0
0.0
0.0
21.1
28.8
26.9
26.7
2.78E-02
4.24E-01
4.69E-01
2.94E-02
4.54E-01
5.94E-05
3.96E-05
1.57E-01
2.31E-01
1.18E-01
8.42E-02
3.19E-05
9.11E-03
7.34E-05
1.12E-03
9.84E-02
1.85E-02
1.06E+00
5.86E-01
1.75E-02
2.70E-01
3.06E-01
1.87E-02
2.54E-01
5.94E-05
3.96E-05
l.OOE-01
1.56E-01
8.84E-02
7.30E-02
3.19E-05
9.11E-03
7.34E-05
1.12E-03
6.92E-02
1.11E-02
6.62E-01
3.67E-01
37.0
36.4
34.8
36.6
44.0
0.0
0.0
36.1
32.4
25.0
13.4
0.0
0.0
0.0
0.0
29.6
39.8
37.7
37.5
5-42
-------
Final Regulatory Impact Analysis
Table 3.2-5. Contributions of Source Sectors to Nationwide Average Census Tract Concentrations of Benzene, With and
Without Controls in this Rule, 2015, 2020, and 2030.
Reference
Control
% Difference
2015 annual average concentrations I
major
1.55E-02
1.54E-02
0
area &
other
1.63E-01
1.61E-01
-1
highway
vehicles
3.79E-01
2.83E-01
-25
nonroad
1.14E-01
1.02E-01
-10
fig m'3)
total
(including
background)
1 .07E+00
9.56E-01
-10
2020 annual average concentrations (fig m"3)
major
1 .70E-02
1 .69E-02
0
Average Nationwide Difference in Ambient Benzene Concentration - Non RFC Areas
Reference
Control
% Difference
1.08E-02
1.08E-02
0
1.43E-01
1.41E-01
-2
2.96E-01
2.17E-01
-27
8.15E-02
6.82E-02
-16
8.93E-01
7.99E-01
-11
1 .20E-02
1 .20E-02
0
Average Nationwide Difference in Ambient Benzene Concentration - RFC Areas
Reference
Control
% Difference
2.39E-02
2.38E-02
0
1.99E-01
1.97E-01
-1
5.29E-01
4.02E-01
-24
1.72E-01
1.63E-01
-5
1 .38E+00
1 .24E+00
-10
2.58E-02
2.58E-02
0
area & other
1.69E-01
1.67E-01
-1
1.48E-01
1.46E-01
-2
2.08E-01
2.05E-01
-1
highway
vehicles
3.88E-01
2.55E-01
-34
3.06E-01
2.00E-01
-35
5.34E-01
3.54E-01
-34
nonroad
1.18E-01
1.05E-01
-10
8.34E-02
6.95E-02
-17
1.79E-01
1.70E-01
-5
total
(including
background)
1 .09E+00
9.38E-01
-14
9.11E-01
7.89E-01
-13
1 .40E+00
1.21E+00
-14
2030 annual average concentrations (fig m'3)
major
1 .70E-02
1 .69E-02
0
1 .20E-02
1 .20E-02
0
2.58E-02
2.58E-02
0
area&
other
1.69E-01
1.67E-01
-1
1.48E-01
1.46E-01
-2
2.08E-01
2.05E-01
-1
highway
vehicles
4.54E-01
2.54E-01
-44
3.57E-01
1.97E-01
-45
6.29E-01
3.57E-01
-43
nonroad
1.32E-01
1.18E-01
-10
9.29E-02
7.72E-02
-17
2.03E-01
1.92E-01
-5
total
(including
background)
1.17E+00
9.50E-01
-19
9.71 E-01
7.94E-01
-18
1 .52E+00
1 .23E+00
-19
5-43
-------
Final Regulatory Impact Analysis
the area source contribution will be reduced about 1 to 2%. The reduction for area sources is due
to the impacts of fuel benzene control on gasoline distribution emissions, and reductions in
portable fuel container (PFC) emissions from PFC and fuel benzene controls. Reductions in
non-reformulated gasoline areas are even larger. It should be noted that the estimated total
reductions in ambient concentrations from all sources are probably significantly underestimated,
since we could not account for the impacts of controls on background levels, which includes
transport of emissions from these sources. Figure 3.2-3 presents the distribution of percent
reductions in median ambient benzene concentrations for U.S. counties with the controls being
finalized in 2030. Again, since the 1.3% maximum average fuel benzene standard is not
included in the modeling, reductions in some parts of the country, including the Pacific
Northwest, are underestimated. Summary tables providing data by State, as well as maps of
MSAT concentrations with controls and percent reductions with controls, can be found in the
docket for the rule.
Figure 3.2-3. Distribution of Percent Reductions in Median Ambient Benzene
Concentrations, 2030, for U. S. Counties with the Controls in this Rule.
Percent difference
^ -13.547%--6.407%
^ -6.406%--4.241%
| -4.240% - -2.863%
| -2.862%--1.821%
| -1.820%--0.956%
-0.955% - -0
3.2.1.2 Exposure and Risk Modeling
3.2.1.2.1 Methods
The HAPEM6 exposure model used in this assessment is the most recent version in a
series of models that the EPA has used to model population exposures and risks at the urban and
national scale in a number of assessments.148'149' 15° HAPEM6 is designed to assess average
3-44
-------
Final Regulatory Impact Analysis
long-term inhalation exposures of the general population, or a specific sub-population, over
spatial scales ranging from urban to national. HAPEM6 uses the general approach of tracking
representatives of 6 specified age groups as they move among indoor and outdoor
microenvironments and among geographic locations (a total of 14, HAPEM5 had 37). The
estimated pollutant concentrations in each microenvironment visited are combined into a time-
weighted average concentration, which is assigned to members of the demographic group.
HAPEM calculates 30 replicates with different exposures for each demographic group. These
data can be used to develop a distribution of exposures for the entire U. S. population.
HAPEM6 uses five primary sources of information: year 2000 population data from the
U.S. Census, population activity data, air quality data, roadway locations, and
microenvironmental data. The population data used are obtained from the U.S. Census. Two
kinds of activity data are used: activity pattern data and commuting pattern data. The activity
pattern data quantify the amount of time individuals spend in a variety of microenvironments and
come from EPA's Consolidated Human Activity Database (CHAD).151 The commuting data
contained in the HAPEM6 default file were derived from the year 2000 U.S. Census, and
includes the number of residents of each tract that work in that tract and every other U.S. Census
tract, as well as data on commuting times and distances. The air quality data come from ASPEN
(after background has been added). The road locations are determined from geographic
information system files from the U.S. Census. The microenvironmental data consist of factors
that estimate air toxic concentrations in specific microenvironments, based on penetration of
outdoor air into the microenvironment, proximity of the microenvironment to the emission
source, and emission sources within the microenvironment. These factors vary among
pollutants.152
New to HAPEM6 are algorithms which account for the gradient in concentrations of
primary (directly emitted) mobile source air toxics within 200 meters of major roadways. 153
HAPEM6 adjusts ambient concentrations generated by ASPEN for each census tract using
concentration gradients developed with the CALPUFF dispersion model.154 For locations within
75 meters and from 75 to 200 meters from major roads, ambient concentrations are adjusted
upward, while locations further from major roadways are adjusted downward. These
adjustments are consistent with results from prior modeling studies that explicitly accounted for
concentration gradients around major roads within census tracts.155 These adjusted
concentrations are then employed in microenvironmental concentration calculations.
HAPEM6 has a number of other technical improvements over the previous version of
HAPEM. These improvements, along with other details of the model, are described in the
HAPEM6 User's Guide.156 In short, HAPEM6 reduces the number of demographic groups to 6
age-based groups from 10 age-gender groups in HAPEM5, and reduces the number of
microenvironments modeled, from 37 to 14. This reduces modeling run time significantly with
little impact on results. HAPEM6 also accounts for commuting time better, basing commute
times and travel modes for each census tract on distributions reported in the 2000 Census. The
HAPEM runs used year 2000 census data. Average lifetime exposure for an individual in a
census tract was calculated from data for individual demographic groups using a post-processing
routine. We estimated the contributions to ambient concentrations for the following source
sectors: major, area and other, onroad, nonroad, and background.
3-45
-------
Final Regulatory Impact Analysis
Once HAPEM runs were completed, cancer risk and noncancer risk were calculated for
each of the mobile source air toxic pollutants, based on population exposure distributions. In the
HAPEM6 output, for each source category, there are 30 replicate exposure concentrations for
each of the six demographic groups (180 concentrations per census tract for each source
category). For each source category and each of the 30 replicates, a lifetime exposure
concentration was calculated. A risk estimate was then calculated for each of the 30 replicates.
The resulting data were used to develop distributions of population risks at various summary
levels (census tract, county, state, national). More detail is provided in the technical support
document. Table 3.2-6 lists the pollutants with their respective unit risk estimates (UREs) for
cancer calculations and reference concentrations (RfCs) for noncancer calculations. These are
the same values used in the 1999 NAT A, and more detailed information on how dose-response
values were selected is provided at the website for that assessment. Also listed are the cancer
weight of evidence classifications and target organ system(s) for noncancer calculations.
Table 3.2-6. Dose-Response Values Use in Risk Modeling (Concentrations in ug/m3)
HAP
1,3 -Butadiene
2,2,4-
Trimethylpentane
Acetaldehyde
Acrolein
Benzene
Chromium III
Chromium VI
Ethyl Benzene
Formaldehyde
Hexane
Manganese
MTBE
Naphthalene
Nickel
POM1
POM2
POMS
POM4
POMS
POM6
POM7
POMS
Styrene
Toluene
Xylenes
Carcinogen
Class
A
N/A
B2
A
N/A
A
B
C
A
B2
B2
B2
B2
B2
B2
B2
B2
URE
(per jig/m3)
S.OxKT5
N/A
2.2xl(r6
0
7.8xl(r6*
N/A
1.2xl(r2
0
5.5xl(r9
N/A
N/A
N/A
3.4xlO'5
1.6xl(r4
5.5X10'5
5.5xlO'5
.OxlO'1
.OxlO'2
.OxlO'3
.OxlO'4
.OxlO'5
2.0xlO'4
N/A
N/A
N/A
Source
IRIS
IRIS
IRIS
IRIS
CUT
CAL
EPA/
OAQPS
OAQPS
OAQPS
OAQPS
OAQPS
OAQPS
OAQPS
OAQPS
OAQPS
Organ
Systems
Reproductive
N/A
Respiratory
Respiratory
Immune
N/A
Respiratory
Developmental
Respiratory
Respiratory,
Neurological
Neurological
Liver, Kidney,
Ocular
Respiratory
Respiratory,
Immune
Neurological
Respiratory,
Neurological
Neurological
RfC(mg/
m3)
2.0xl(r3
N/A
9.0xl(r3
2.0xl(r5
S.OxKT2
N/A
l.OxKT4
1.0
9.8xl(r3
2.0X10-1
S.OxlO'5
3.0
S.OxlO'3
6.5xlO'5
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
1.0
4.0X10'1
l.OxlO'1
Source
IRIS
IRIS
IRIS
IRIS
IRIS
ATSDR
IRIS
IRIS
IRIS
IRIS
CAL
IRIS
IRIS
IRIS
*represents upper end of a range of MLE values
3-46
-------
Final Regulatory Impact Analysis
The weight of evidence classifications provided in this table were developed under EPA's 1986
risk assessment guidelines where:
A = Known human carcinogen
Bl = Probable human carcinogen, based on incomplete human data
B2 = Probable human carcinogen, based on adequate animal data
C = Possible human carcinogen
Dose-response values were selected using the following hierarchy:
1) EPA IRIS assessments.
2) Agency for Toxic Substances and Disease Registry (ATSDR) minimum risk levels
(MRLs) for noncancer effects - used as RfC.
3) California Office of Environmental Health Hazard Assessment (OEHHA) values.
There are a number of exceptions to this hierarchy:
1) Formaldehyde — EPA no longer considers the formaldehyde URE reported in IRIS,
which is based on a 1987 study, to represent the best available science in the peer-
reviewed literature. Accordingly, the 1999 risk estimates for formaldehyde are based on a
dose-response value developed by the CUT Centers for Health Research (formerly the
Chemical Industry Institute of Toxicology) and published in 1999. This issue is
discussed in Chapter 1 of the RIA.
2) Nickel — The IRIS URE for nickel inhalation shown in Table 3.2.-6 was derived from
evidence of the carcinogenic effects of insoluble nickel compounds in crystalline form.
Soluble nickel species, and insoluble species in amorphous form, do not appear to
produce genotoxic effects by the same toxic mode of action as insoluble crystalline
nickel. Nickel speciation information for some of the largest nickel-emitting sources
(including oil combustion, coal combustion, and others) suggests that at least 35% of
total nickel emissions may be soluble compounds. The remaining insoluble nickel
emissions are not well-characterized, however. Consistent with this limited information,
this analysis has conservatively assumed that 65% of emitted nickel is insoluble, and that
all insoluble nickel is crystalline. On this basis, the nickel URE (based on nickel
subsulfide, and representative of pure insoluble crystalline nickel) was adjusted to reflect
an assumption that 65% of the total mass of nickel may be carcinogenic. The ATSDR
MRL in Table 3.2.-6 was not adjusted, however, because the noncancer effects of nickel
are not thought to be limited to the crystalline, insoluble form.
3) POM — POM was divided into eight toxicity categories to cover the range of unit
risks of the individual POM species and POM groups contained in the 1999 NEI. The
unit risks for those eight categories were based on the midpoint of the range of unit risks
defining the toxicity category. More details on the development of these unit risks can be
found on the website for the 1999 NATA and in Appendix H of the 2001 EPA draft
report to the Science Advisory Board on the 1996 National-Scale Assessment.157
Individual cancer risk estimates (the product of unit risk estimates and exposure levels)
for various pollutants were assumed to be additive, since there was no evidence of non-additive
3-47
-------
Final Regulatory Impact Analysis
interactions for any of the pollutants. Most of the estimates are based on the statistical upper
confidence limit (UCL) of the fitted dose-response curve, but the estimates for hexavalent
chromium, nickel, and benzene are based on the statistical best fit ("maximum likelihood
estimate," or MLE). Except for benzene and chromium, where risks are based on maximum
likelihood dose-response values, risks from mobile source air toxics should all be considered
upper-bound values. True risks could be greater, but are likely to be lower, and could be zero.
To express chronic noncancer hazards, we used the RfC as part of a calculation called the
hazard quotient (HQ), which is the ratio between the concentration to which a person is exposed
and the RfC. A value of the HQ less than one indicates that the exposure is lower than the RfC
and that no adverse health effects would be expected. A value of the HQ greater than one
indicates that the exposure is higher than the RfC. However, because many RfCs incorporate
protective assumptions in the face of uncertainty, an HQ greater than one does not necessarily
suggest a likelihood of adverse effects. Furthermore, the HQ cannot be translated to a probability
that adverse effects will occur and is not likely to be proportional to risk. A HQ greater than one
can best be described as indicating that a potential exists for adverse health effects. However
one should evaluate the weight of evidence supporting the RfC value for a particular chemical
before determining potential risks. Following the approach used in the 1999 NAT A, combined
noncancer hazards were calculated using the hazard index (HI), defined as the sum of hazard
quotients for individual air toxics compounds that affect the same organ or organ system. The HI
is only an approximation of the combined effect, because some of the substances may affect the
target organs in different (i.e., non-additive) ways. As with the HQ, a value of the HI below 1.0
will likely not result in adverse effects over a lifetime of exposure. However, a value of the HI
greater than 1.0 does not necessarily suggest a likelihood of adverse effects. Furthermore, the HI
cannot be translated to a probability that adverse effects will occur and is not likely to be
proportional to risk. An HI greater than one can be best described as indicating that a potential
may exist for adverse health effects.
3.2.1.2.2 Exposure and Risk Trends for Air Toxics: Reference Case
Tables 3.2-7 and 3.2-8 summarize nationwide averages of median and 90th percentile
census tract exposure concentrations of mobile source air toxics in 1999, 2015, 2020, and 2030,
without the controls being finalized in this rule. It should be noted that all the other non-
inventoried sources, as well as the contribution from transport, contribute to background levels.
Overall, exposure to ambient concentrations tends to be less than ambient concentrations because
penetration rates to indoor microenvironments are typically less than one.c However, highway
vehicles make a larger contribution to overall average population exposures than they do to
ambient levels. This is largely because of elevated exposures experienced inside vehicles.
c In the exposure monitoring studies discussed in section 3.1.2, average measured personal exposure concentrations
are greater than those in both indoor and outdoor air. These differences may be attributable to several factors. First,
HAPEM6 does not include pollution sources within indoor microenvironments, such as attached garages,
environmental tobacco smoke, and solvent storage. Second, measured personal breathing zone concentrations are
integrated measurements that account for time-weighted average (TWA) concentrations that incorporate every
source, activity, and location with which a monitor comes into contact. Microenvironmental models like HAPEM6
simplify individual time budgets so they fit within the microenvironments modeled or monitored.
3-48
-------
Final Regulatory Impact Analysis
Table 3.2-9 summarizes national average population cancer risk across census tracts for
these years by pollutant, as well as total cancer risk across pollutants. The total cancer risk from
mobile source air toxics (including the stationary source contribution) was about 25 in a million
in 1999.
In all projection years, benzene emissions are by far the largest contributor to cancer risk
from mobile sources (see Figure 3.2-4). Other significant contributors to cancer risk from
mobile source air toxics include 1,3-butadiene, acetaldehyde, naphthalene, and hexavalent
chromium. It should be noted, however, that we have no actual measurements of hexavalent
chromium emissions from mobile sources, and that the risk estimate for this pollutant is based on
an assumption that forty percent of the chromium from highway vehicles and eighteen percent of
the chromium from nonroad sources was assumed to be the highly toxic hexavalent form. The
estimate for highway vehicles is based on data from utility boilers,158 and the estimate for
nonroad equipment is, based on combustion data from stationary combustion turbines that burn
diesel fuel.159 Thus there is a great deal of uncertainty in estimates for this pollutant.
Despite significant reductions in risk from mobile source air toxics, average inhalation
cancer risks for these pollutants in 2030, accounting for both mobile and stationary source
contributions, remain well above 20 in 1,000,000 (Figure 3.2-5). In addition, average risk from
exposure to benzene remains above 9 in 1,000,000.
3-49
-------
Final Regulatory Impact Analysis
Table 3.2-7. National Means of Census Tract Median Population Exposure Concentrations of Mobile Source Air Toxics in
1999, 2015, 2020, and 2030, Without Controls in this Rule.
Pollutant
1,3-Butadiene
2,2,4-Trimethylpentane
Acetaldehyde
Acrolein
Benzene
Chromium III
Chromium VI
Ethyl Benzene
Formaldehyde
Hexane
MTBE
Manganese
Naphthalene
Nickel
POM
Propionaldehyde
Styrene
Toluene
Xylenes
background
(Hg m 3)
3.96E-02
O.OOE+00
4.00E-01
O.OOE+00
3.05E-01
O.OOE+00
O.OOE+00
O.OOE+00
6.12E-01
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
1.28E-01
1999 annual average concentrations (ng in 3)
major
1.54E-03
1.70E-02
2.34E-02
2.56E-03
1.76E-02
3.23E-04
4.25E-05
1.45E-02
3.29E-02
5.50E-02
1.05E-02
1.05E-03
3.82E-03
3.02E-04
2.87E-03
7.73E-03
2.04E-02
1.61E-01
8.08E-02
area&
other
1.66E-02
1.86E-02
4.33E-02
2.35E-02
1.16E-01
1.79E-04
7.94E-05
7.49E-02
7.20E-02
3.60E-01
4.84E-02
8.93E-04
3.37E-02
5.78E-04
l.OOE-02
1.80E-02
1.14E-02
6.57E-01
4.66E-01
onroad
6.39E-02
8.23E-01
8.08E-01
6.62E-02
8.08E-01
1.93E-05
1.30E-05
3.22E-01
5.78E-01
2.85E-01
4.61E-01
1.08E-05
1.79E-02
2.38E-05
1.56E-03
1.93E-01
3.40E-02
2.14E+00
1.21E+00
nonroad
1.64E-02
1.57E-01
1.18E-01
1.83E-02
1.51E-01
2.21E-05
5.06E-06
8.02E-02
1.88E-01
7.13E-02
3.40E-01
2.40E-06
3.85E-03
4.17E-05
5.48E-04
3.35E-02
3.03E-03
3.42E-01
3.33E-01
total
(including
background)
1.38E-01
1.02E+00
1.39E+00
1.10E-01
1.40E+00
5.43E-04
1.40E-04
4.91E-01
1.48E+00
7.71E-01
8.59E-01
1.96E-03
5.92E-02
9.46E-04
1.50E-02
2.52E-01
6.88E-02
3.30E+00
2.22E+00
2015 annual average concentrations (|j,g m"3)
major
1.72E-03
8.68E-03
2.41E-02
2.91E-03
1.25E-02
4.11E-04
5.40E-05
9.92E-03
4.15E-02
4.94E-02
1.26E-03
1.13E-02
3.37E-03
3.47E-04
2.26E-03
7.24E-03
2.40E-02
1.16E-01
6.79E-02
area&
other
1.69E-02
2.18E-02
4.60E-02
2.14E-02
1.37E-01
2.43E-04
1.09E-04
l.OOE-01
8.26E-02
4.41E-01
1.17E-03
5.35E-02
4.18E-02
6.50E-04
1.16E-02
1.89E-02
1.56E-02
8.80E-01
6.43E-01
onroad
2.88E-02
4.16E-01
4.70E-01
2.90E-02
4.53E-01
2.64E-05
1.78E-05
1.62E-01
2.46E-01
1.44E-01
1.48E-05
1.24E-01
9.89E-03
3.29E-05
8.33E-04
9.56E-02
1.73E-02
1.09E+00
6.11E-01
nonroad
1.01E-02
9.26E-02
9.07E-02
1.49E-02
9.87E-02
2.34E-05
5.38E-06
4.69E-02
1.38E-01
4.98E-02
2.84E-06
8.90E-02
4.02E-03
4.80E-05
4.97E-04
2.28E-02
1.83E-03
2.06E-01
1.85E-01
total
(including
background)
9.71E-02
5.39E-01
1.03E+00
6.83E-02
1.01E+00
7.03E-04
1.86E-04
3.19E-01
1.12E+00
6.85E-01
2.45E-03
2.78E-01
5.91E-02
1.08E-03
1.52E-02
1.45E-01
5.86E-02
2.29E+00
1.63E+00
5-50
-------
Final Regulatory Impact Analysis
Table 3.2-7 (cont'd). National Means of Census Tract Median Population Exposure Concentrations of Mobile Source Air
Toxics in 1999, 2015, 2020, and 2030, Without Controls in this Rule.
Pollutant
1,3-Butadiene
2,2,4-Trimethylpentane
Acetaldehyde
Acrolein
Benzene
Chromium III
Chromium VI
Ethyl Benzene
Formaldehyde
Hexane
MTBE
Manganese
Naphthalene
Nickel
POM
Propionaldehyde
Styrene
Toluene
Xylenes
background
(M^m3)
3.96E-02
O.OOE+00
4.00E-01
O.OOE+00
3.05E-01
O.OOE+00
O.OOE+00
O.OOE+00
6.12E-01
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
1.28E-01
2020 annual average concentrations (jig in 3)
major
1.86E-03
9.37E-03
2.52E-02
3.27E-03
1.37E-02
4.59E-04
6.14E-05
1.11E-02
4.71E-02
5.44E-02
1.27E-02
1.40E-03
3.78E-03
3.77E-04
2.51E-03
7.27E-03
2.74E-02
1.30E-01
7.68E-02
area &
other
1.69E-02
2.31E-02
4.70E-02
2.07E-02
1.42E-01
2.74E-04
1.23E-04
1.10E-01
8.68E-02
4.77E-01
5.48E-02
1.29E-03
4.44E-02
7.15E-04
1.18E-02
1.94E-02
1.72E-02
9.68E-01
7.10E-01
onroad
2.98E-02
4.16E-01
4.85E-01
2.99E-02
4.64E-01
2.90E-05
1.96E-05
1.62E-01
2.52E-01
1.33E-01
1.01E-01
1.62E-05
9.84E-03
3.62E-05
8.63E-04
9.81E-02
1.80E-02
1.10E+00
6.18E-01
nonroad
1.07E-02
9.21E-02
9.01E-02
1.58E-02
1.02E-01
2.37E-05
5.45E-06
4.83E-02
1.38E-01
5.19E-02
9.25E-02
3.00E-06
4.31E-03
5.02E-05
5.01E-04
2.25E-02
1.87E-03
2.09E-01
1.87E-01
total
(including
background)
9.88E-02
5.41E-01
1.05E+00
6.97E-02
1.03E+00
7.86E-04
2.09E-04
3.32E-01
1.14E+00
7.17E-01
2.61E-01
2.71E-03
6.23E-02
1.18E-03
1.57E-02
1.47E-01
6.45E-02
2.41E+00
1.72E+00
2030 annual average concentrations (fig m"3)
major
1.86E-03
9.37E-03
2.52E-02
3.27E-03
1.37E-02
4.59E-04
6.14E-05
1.11E-02
4.71E-02
5.44E-02
1.27E-02
1.40E-03
3.78E-03
3.77E-04
2.51E-03
7.27E-03
2.74E-02
1.30E-01
7.68E-02
area &
other
1.69E-02
2.31E-02
4.70E-02
2.07E-02
1.42E-01
2.74E-04
1.23E-04
1.10E-01
8.68E-02
4.77E-01
5.48E-02
1.29E-03
4.44E-02
7.15E-04
1.18E-02
1.94E-02
1.72E-02
9.68E-01
7.10E-01
onroad
3.49E-02
4.81E-01
5.68E-01
3.51E-02
5.40E-01
3.56E-05
2.40E-05
1.87E-01
2.94E-01
1.46E-01
l.OOE-01
1.99E-05
1.14E-02
4.45E-05
1.02E-03
1.14E-01
2.13E-02
1.28E+00
7.17E-01
nonroad
1.21E-02
l.OOE-01
9.78E-02
1.79E-02
1.15E-01
2.43E-05
5.62E-06
5.41E-02
1.51E-01
5.83E-02
1.04E-01
3.35E-06
4.94E-03
5.47E-05
5.58E-04
2.42E-02
2.10E-03
2.30E-01
2.06E-01
total
(including
background)
1.05E-01
6.14E-01
1.14E+00
7.70E-02
1.12E+00
7.93 E-04
2.14E-04
3.62E-01
1.19E+00
7.36E-01
2.72E-01
2.71E-03
6.45E-02
1.19E-03
1.59E-02
1.65E-01
6.80E-02
2.61E+00
1.84E+00
5-51
-------
Final Regulatory Impact Analysis
Table 3.2-8. National Means of Census Tract 90 Percentile Population Exposure Concentrations of Mobile Source Air
Toxics in 1999, 2015, 2020, and 2030, Without Controls in this Rule.
Pollutant
1,3-Butadiene
2,2,4-Trimethylpentane
Acetaldehyde
Acrolein
Benzene
Chromium III
Chromium VI
Ethyl Benzene
Formaldehyde
Hexane
MTBE
Manganese
Naphthalene
Nickel
POM
Propionaldehyde
Styrene
Toluene
Xylenes
background
(Hg m 3)
5.88E-02
O.OOE+00
5.82E-01
O.OOE+00
4.50E-01
O.OOE+00
O.OOE+00
O.OOE+00
8.03E-01
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
2.04E-01
1999 annual average concentrations (ng in 3)
major
2.03E-03
2.65E-02
3.48E-02
3.68E-03
2.57E-02
4.55E-04
6.16E-05
2.33E-02
4.21E-02
7.54E-02
1.46E-02
1.44E-03
4.81E-03
4.25E-04
3.68E-03
1.30E-02
2.87E-02
2.52E-01
1.23E-01
area&
other
2.23 E-02
3.12E-02
6.34E-02
3.36E-02
1.71E-01
2.59E-04
1.15E-04
1.19E-01
9.22E-02
5.11E-01
7.13E-02
1.18E-03
4.39E-02
8.25E-04
1.21E-02
2.79E-02
1.78E-02
1.05E+00
7.05E-01
onroad
l.OOE-01
1.42E+00
1.27E+00
1.07E-01
1.24E+00
2.88E-05
1.92E-05
5.49E-01
7.89E-01
4.32E-01
7.22E-01
1.47E-05
2.44E-02
3.52E-05
2.04E-03
3.36E-01
5.90E-02
3.61E+00
1.95E+00
nonroad
2.49E-02
2.65E-01
1.80E-01
2.82E-02
2.28E-01
3.15E-05
7.20E-06
1.35E-01
2.52E-01
1.07E-01
5.16E-01
3.25E-06
5.09E-03
6.04E-05
7.05E-04
5.58E-02
5.23E-03
5.66E-01
5.25E-01
total
(including
background)
2.08E-01
1.75E+00
2.13E+00
1.72E-01
2.12E+00
7.74E-04
2.03E-04
8.27E-01
1.98E+00
1.13E+00
1.32E+00
2.64E-03
7.83E-02
1.35E-03
1.85E-02
4.33E-01
1.11E-01
5.48E+00
3.50E+00
2015 annual average concentrations (|j,g m"3)
major
2.15E-03
1.30E-02
3.32E-02
3.72E-03
1.70E-02
5.81E-04
7.88E-05
1.51E-02
4.93E-02
6.51E-02
1.45E-02
1.72E-03
4.07E-03
4.94E-04
2.89E-03
1.15E-02
3.31E-02
1.70E-01
9.59E-02
area&
other
2.16E-02
3.56E-02
6.27E-02
2.79E-02
1.91E-01
3.51E-04
1.57E-04
1.51E-01
9.74E-02
5.95E-01
7.24E-02
1.55E-03
5.13E-02
9.09E-04
1.38E-02
2.72E-02
2.31E-02
1.32E+00
9.14E-01
onroad
4.11E-02
7.08E-01
6.89E-01
4.36E-02
6.52E-01
3.97E-05
2.63E-05
2.63E-01
3.03E-01
2.04E-01
1.92E-01
2.01E-05
1.25E-02
4.77E-05
1.04E-03
1.60E-01
2.87E-02
1.71E+00
9.13E-01
nonroad
1.39E-02
1.54E-01
1.28E-01
2.11E-02
1.39E-01
3.35E-05
7.67E-06
7.53E-02
1.67E-01
7.06E-02
1.34E-01
3.85E-06
4.99E-03
6.89E-05
6.14E-04
3.57E-02
3.01E-03
3.21E-01
2.72E-01
total
(including
background)
1.38E-01
9.10E-01
1.49E+00
9.64E-02
1.45E+00
1.01E-03
2.70E-04
5.04E-01
1.42E+00
9.34E-01
4.13E-01
3.30E-03
7.29E-02
1.52E-03
1.84E-02
2.34E-01
8.79E-02
3.52E+00
2.40E+00
5-52
-------
Final Regulatory Impact Analysis
Table 3.2-8 (cont'd). National Means of Census Tract 90 Percentile Population Exposure Concentrations of Mobile Source
Air Toxics in 1999, 2015, 2020, and 2030, Without Controls in this Rule.
Pollutant
1,3-Butadiene
2,2,4-Trimethylpentane
Acetaldehyde
Acrolein
Benzene
Chromium III
Chromium VI
Ethyl Benzene
Formaldehyde
Hexane
MTBE
Manganese
Naphthalene
Nickel
POM
Propionaldehyde
Styrene
Toluene
Xylenes
background
(Hg in 3)
5.88E-02
O.OOE+00
5.82E-01
O.OOE+00
4.50E-01
O.OOE+00
O.OOE+00
O.OOE+00
8.03E-01
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
O.OOE+00
2.04E-01
2020 annual average concentrations (ng in 3)
major
2.32E-03
1.40E-02
3.47E-02
4.15E-03
1.86E-02
6.51E-04
8.98E-05
1.68E-02
5.60E-02
7.12E-02
1.61E-02
1.92E-03
4.55E-03
5.39E-04
3.21E-03
1.16E-02
3.78E-02
1.88E-01
1.08E-01
area&
other
2.16E-02
3.75E-02
6.42E-02
2.70E-02
1.99E-01
3.97E-04
1.77E-04
1.65E-01
1.02E-01
6.39E-01
7.34E-02
1.71E-03
5.44E-02
l.OOE-03
1.42E-02
2.78E-02
2.55E-02
1.44E+00
1.01E+00
onroad
4.28E-02
7.09E-01
7.14E-01
4.53E-02
6.68E-01
4.37E-05
2.90E-05
2.62E-01
3.11E-01
1.86E-01
1.52E-01
2.21E-05
1.24E-02
5.25E-05
1.08E-03
1.65E-01
2.99E-02
1.73E+00
9.22E-01
nonroad
1.48E-02
1.54E-01
1.28E-01
2.23 E-02
1.44E-01
3.39E-05
7.78E-06
7.71E-02
1.67E-01
7.27E-02
1.37E-01
4.07E-06
5.33E-03
7.19E-05
6.21E-04
3.52E-02
3.07E-03
3.23E-01
2.74E-01
total
(including
background)
1.40E-01
9.14E-01
1.52E+00
9.87E-02
1.48E+00
1.13E-03
3.04E-04
5.21E-01
1.44E+00
9.69E-01
3.78E-01
3.65E-03
7.66E-02
1.66E-03
1.91E-02
2.39E-01
9.64E-02
3.68E+00
2.51E+00
2030 annual average concentrations (jig m 3)
major
2.32E-03
1.40E-02
3.47E-02
4.15E-03
1.86E-02
6.51E-04
8.98E-05
1.68E-02
5.60E-02
7.12E-02
1.61E-02
1.92E-03
4.55E-03
5.39E-04
3.21E-03
1.16E-02
3.78E-02
1.88E-01
1.08E-01
area&
other
2.16E-02
3.75E-02
6.42E-02
2.70E-02
1.99E-01
3.97E-04
1.77E-04
1.65E-01
1.02E-01
6.39E-01
7.34E-02
1.71E-03
5.44E-02
l.OOE-03
1.42E-02
2.78E-02
2.55E-02
1.44E+00
1.01E+00
onroad
5.11E-02
8.25E-01
8.55E-01
5.37E-02
7.93E-01
5.40E-05
3.58E-05
3.06E-01
3.70E-01
2.06E-01
1.50E-01
2.72E-05
1.45E-02
6.45E-05
1.29E-03
1.94E-01
3.57E-02
2.04E+00
1.09E+00
nonroad
1.72E-02
1.69E-01
1.41E-01
2.56E-02
1.65E-01
3.48E-05
8.04E-06
8.71E-02
1.86E-01
8.21E-02
1.53E-01
4.54E-06
6.16E-03
7.86E-05
6.97E-04
3. 83 E-02
3.47E-03
3.61E-01
3.07E-01
total
(including
background)
1.51E-01
1.05E+00
1.68E+00
1.10E-01
1.63E+00
1.14E-03
3.11E-04
5.75E-01
1.52E+00
9.98E-01
3.93E-01
3.66E-03
7.96E-02
1.68E-03
1.94E-02
2.72E-01
1.03E-01
4.03E+00
2.71E+00
5-53
-------
Final Regulatory Impact Analysis
Table 3.2-9. National Average Cancer Risk Across Census Tracts for 1999, 2015, 2020, and 2030 by Pollutant, Without
Controls in this Rule.
Pollutant
1,3-Butadiene
Acetaldehyde
Benzene
Chromium VI
Formaldehyde
Naphthalene
Nickel
POM
1999 average Individual risk
major
4.36E-08
5.65E-08
1.49E-07
5.32E-07
1.81E-10
1.21E-07
4.81E-08
1.77E-07
area &
other
4.85E-07
1.10E-07
9.82E-07
9.43E-07
4.51E-10
1.22E-06
9.79E-08
1.06E-06
onroad
2.06E-06
1.96E-06
6.79E-06
1.69E-07
3.36E-09
6.38E-07
4.17E-09
1.05E-07
nonroad
5.39E-07
2.89E-07
1.30E-06
7.18E-08
1.11E-09
1.37E-07
6.20E-09
3.62E-08
total
(including
background)
4.43E-06
3.39E-06
1.18E-05
1.72E-06
8.69E-09
2.11E-06
1.56E-07
1.38E-06
2015 annual average individual risk
major
4.62E-08
5.59E-08
l.OOE-07
6.67E-07
2.10E-10
1.01E-07
5.53E-08
1.46E-07
area &
other
4.50E-07
1.16E-07
1.13E-06
1.25E-06
5.18E-10
1.46E-06
1.07E-07
1.25E-06
onroad
8.69E-07
1.08E-06
3.66E-06
2.29E-07
1.35E-09
3.43E-07
5.65E-09
5.39E-08
nonroad
3.20E-07
2.12E-07
8.25E-07
8.11E-08
7.69E-10
1.39E-07
6.87E-09
3.25E-08
total
(including
background)
2.97E-06
2.43E-06
8.33E-06
2.23E-06
6.43E-09
2.04E-06
1.75E-07
1.48E-06
Pollutant
1,3-Butadiene
Acetaldehyde
Benzene
Chromium VI
Formaldehyde
Naphthalene
Nickel
POM
2020 annual average individual risk
major
4.95E-08
5.80E-08
1.09E-07
7.53E-07
2.34E-10
1.12E-07
6.02E-08
1.61E-07
area &
other
4.38E-07
1.19E-07
1.17E-06
1.40E-06
5.47E-10
1.54E-06
1.16E-07
1.30E-06
onroad
8.92E-07
1.10E-06
3.71E-06
2.50E-07
1.38E-09
3.39E-07
6.19E-09
5.54E-08
nonroad
3.39E-07
2.08E-07
8.54E-07
8.34E-08
7.63E-10
1.48E-07
7.10E-09
3.27E-08
total
(including
background)
3.00E-06
2.46E-06
8.45E-06
2.49E-06
6.49E-09
2.14E-06
1.90E-07
1.55E-06
2030 annual average individual risk
major
4.82E-08
5.75E-08
1.08E-07
7.48E-07
2.28E-10
1.09E-07
6.01E-08
1.61E-07
area &
other
4.19E-07
1.19E-07
1.16E-06
1.38E-06
5.54E-10
1.52E-06
1.15E-07
1.31E-06
onroad
1.03E-06
1.28E-06
4.29E-06
3.05E-07
1.59E-09
3.91E-07
7.55E-09
6.52E-08
nonroad
3.86E-07
2.23E-07
9.59E-07
8.78E-08
8.22E-10
1.69E-07
7.60E-09
3.59E-08
total
(including
background)
3.16E-06
2.65E-06
9.13E-06
2.52E-06
6.76E-09
2.19E-06
1.90E-07
1.57E-06
5-54
-------
Final Regulatory Impact Analysis
Figure 3.2-4. Contributions to Average Inhalation Cancer Risk from Air Toxics Emitted
by Mobile Sources, 2020 (Not Including Diesel PM and Diesel Exhaust Organic Gases),
Without Controls in this Rule.
Naphthalene Nickel
6% ~
Formaldehyde
0% ~"\
Chromium VI
4%
1,3-Butadiene
15%
Acetaldehyde
16%
Benzene
58%
Figure 3.2-5. Average Nationwide Cancer Risk from Emissions of Mobile Source Air
Toxics from both Mobile and Stationary Sources across Census Tracts, 1999 to 2030 (Not
Including Diesel PM and Diesel Exhaust Organic Gases), Without Controls in this Rule.
O.OE+00
1999 2015 2020 2030
Year
3-55
-------
Final Regulatory Impact Analysis
It should also be noted that because of population growth projected to occur in the United
States, the number of Americans above cancer risk benchmarks will increase. Figure 3.2-6
depicts the U. S. population at various risk benchmarks for mobile source air toxics in 1999,
2015, 2020, and 2030, using population projections from EPA's BenMAP model, a tool the EPA
uses to estimate benefits of air pollution control strategies, and average census tract exposures.
(BenMAP was recently used for EPA's Clean Air Interstate Air Quality Rule (CAIR),160 and is
also discussed in Chapter 12 of the RIA). These statistics do not include populations in Alaska
and Hawaii; thus populations in these States were assumed to remain at year 2000 levels. More
details on the methodology used to project the U. S. population above various cancer risk
benchmarks are provided in the technical support document "National-Scale Modeling of Mobile
Source Air Toxic Emissions, Air Quality, Exposure and Risk for the Final Mobile Source Air
Toxics Rule." From Figure 3.2-6 it can be seen that, based on average census tract risks, the vast
majority of the population experiences risks between one in a million (IxlO"6) and one in ten
thousand (IxlO"4). However, the number of people experiencing risks above one in a hundred
thousand (IxlO"5) increases from 223 million in 1999 to 272 million in 2030.
Figure 3.2-6. U. S. Population at Various Cancer Risk Benchmarks due to Exposure to
Mobile Source Air Toxics, 1999 - 2030, Without Controls in this Rule.
Ann
fUU
OCfl
oou
= 1 E-04
Tables 3.2-10 and 3.2-11 summarize national average population hazard quotients for
chronic non-cancer effects across census tracts for these years by pollutant, as well as the
respiratory hazard index across pollutants. The respiratory system is the only target organ
system where the hazard index exceeds one. Although the average respiratory hazard index for
mobile source air toxics decreases by almost 33% between 1999 and 2030 (Figure 3.2-7), it is
still over 4 in 2030, indicating a potential for adverse health effects. The reduction in hazard
index occurs despite large increases in activity for highway and nonroad sources. In addition,
3-56
-------
Final Regulatory Impact Analysis
about 90% of this non-cancer risk is attributable to acrolein in all projection years. It should be
noted that the confidence in the RfC for acrolein is medium. About 25% of primary acrolein
emissions are from mobile sources, and about 70% of ambient concentrations of acrolein (and
about 75% of exposure) are attributable to mobile sources. The mobile source contribution to
concentrations and exposure is largely attributable to the contribution from mobile source 1,3-
butadiene, which is transformed to acrolein in the atmosphere. Moreover, projected growth in
the U. S. population and increasing vehicle miles traveled will increase the number of Americans
with a respiratory hazard index for mobile source air toxics above one, from 258 million in 1999
to 307 million in 2030 (Figure 3.2-8).
Detailed summary tables presenting cancer risk, hazard quotients and hazard indices by
State, and for reformulated and non-reformulated (i.e., conventional) gasoline areas, can be
found in the docket for this rule, along with statistics on number of individuals above various
cancer and non-cancer benchmarks, by source sector.
3.2.1.2.3 Distributions of Air Toxics Risk across the U. S.: Reference Case
Table 3.2-12 gives the distribution of nationwide individual cancer risks for mobile
source air toxics in 2020, absent the controls being finalized in this rule. Summary tables
providing distributions for other years, as well as distributions by State and for reformulated and
non-reformulated gasoline areas, can be found in the docket for this rule. Risk distributions are
broader than the distributions of ambient concentrations in Table 3.2-2. For instance, while the
95th percentile benzene concentration is about twice the median value, the 95th percentile cancer
risk is roughly three times the median risk. A key reason for this is the variability in activity
patterns, concentrations among microenvironments, and commuting patterns. Figures 3.2-9
through 3.2-12 depict the geographic distributions of median county cancer risks in 2020 for all
mobile source air toxics, and separately for benzene, acetaldehyde and 1,3-butadiene. These
geographic distributions closely track distributions of ambient concentrations, with the highest
risks in major population centers of the country where mobile source activity is the greatest.
Relatively high benzene risks are also seen in areas of the country where fuel benzene levels are
higher, such as the Pacific Northwest, parts of Alaska, and the upper Great Lakes region, since
higher fuel benzene levels lead to higher benzene emissions and higher exposures. Higher risks
are also seen in States with colder winters, due to elevated cold start emissions.
Previously discussed changes to the HAPEM exposure model, to account for near road
impacts, can impact distributions of risk. In order to evaluate the effect of switching to
HAPEM6 from HAPEM5 on individual risks nationally, we conducted model runs using
identical input data. Figure 3.2-13 depicts the national distribution of individual cancer risks
from benzene, comparing HAPEM6 and HAPEM5. Note that the graph is on a logarithmic
scale. As the graph illustrates, when HAPEM6 is used, there are fewer individuals with lower
benzene cancer risk levels (e.g. lxlO"4) is higher with HAPEM6 than HAPEM5. In general, the distribution of cancer
risks shifts slightly higher when comparing HAPEM6 to HAPEM5, but the largest effects are
observed in the populations with the highest and lowest risk levels, which are generally small
fractions of the total population.
3-57
-------
Final Regulatory Impact Analysis
Table 3.2-13 gives the distribution of nationwide individual hazard quotients for acrolein,
and hazard indices for the respiratory target system in 2020. Patterns for other years are similar.
The average respiratory hazard index at the 95th percentile is over 20 times that at the 5th
percentile, and about 4 times the median. Thus, some populations are experiencing much higher
hazard indices than others. Figure 3.2-14 depicts the geographic distribution of median county
respiratory hazard indices in 2020. The high hazard indices in Idaho are the result of high
inventory estimates for wildfires and reflect a known error in the Idaho inventory for this source.
This error was discovered at too late a date to produce and update emissions inventories for use
in the analyses undertaken for this rule. The errors are not expected to affect the analyses of the
impacts of controls undertaken for this rule.
3-58
-------
Table 3.2-10. National Average Population Hazard Quotient for Chronic Noncancer Effects Across Census Tracts, 1999 -
2030, Without Controls in this Rule.
Pollutant
1,3-Butadiene
Acetaldehyde
Acrolein
Benzene
Chromium VI
Ethyl Benzene
Formaldehyde
Hexane
MTBE
Manganese
Naphthalene
Nickel
Styrene
Toluene
Xylenes
Target System
Reproductive
Respiratory
Respiratory
Immunological
Respiratory
Developmental
Respiratory
Neurological, Respiratory
Liver, Kidney, Ocular
Neurological
Respiratory
Respiratory, Immunological
Neurological
Respiratory, Neurological
Neurological
1999 average Hazard Quotient
major
7.27E-04
2.86E-03
1.44E-01
6.35E-04
4.43E-04
1.60E-05
3.36E-03
2.76E-04
3.86E-06
2.04E-02
1.19E-03
4.62E-03
2.38E-05
4.55E-04
8.47E-04
area &
other
8.08E-03
5.54E-03
1.28E+00
4.20E-03
7.86E-04
8.09E-05
8.37E-03
1.89E-03
1.76E-05
1.93E-02
1.19E-02
9.42E-03
1.28E-05
1.82E-03
5.00E-03
onroad
3.43E-02
9.92E-02
3.70E+00
2.90E-02
1.41E-04
3.60E-04
6.23E-02
1.55E-03
1.72E-04
2.27E-04
6.25E-03
4.01E-04
3.77E-05
5.96E-03
1.32E-02
nonroad
8.98E-03
1.46E-02
1.03E+00
5.55E-03
5.98E-05
9.17E-05
2.05E-02
3.95E-04
1.28E-04
4.59E-05
1.35E-03
5.96E-04
3.46E-06
9.69E-04
3.72E-03
total
(including
background)
7.39E-02
1.71E-01
6.16E+00
5.06E-02
1.43E-03
5.48E-04
1.61E-01
4.11E-03
3.21E-04
3.99E-02
2.07E-02
1.50E-02
7.78E-05
9.20E-03
2.43E-02
2015 average Hazard Quotient
major
7.69E-04
2.82E-03
1.58E-01
4.29E-04
5.56E-04
1.05E-05
3.90E-03
2.43E-04
3.94E-06
2.65E-02
9.88E-04
5.32E-03
2.85E-05
3.12E-04
6.85E-04
area &
other
7.49E-03
5.84E-03
1.13E+00
4.83E-03
1.04E-03
1.05E-04
9.62E-03
2.21E-03
1.88E-05
2.56E-02
1.43E-02
1.03E-02
1.76E-05
2.39E-03
6.69E-03
onroad
1.45E-02
5.46E-02
1.54E+00
1.56E-02
1.90E-04
1.74E-04
2.51E-02
7.58E-04
4.43E-05
3.07E-04
3.36E-03
5.43E-04
1.84E-05
2.88E-03
6.38E-03
nonroad
5.34E-03
1.07E-02
8.10E-01
3.53E-03
6.76E-05
5.30E-05
1.43E-02
2.71E-04
3.20E-05
5.32E-05
1.36E-03
6.61E-04
2.05E-06
5.72E-04
2.02E-03
total (including
background)
4.96E-02
1.23E-01
3.63E+00
3.56E-02
1.86E-03
3.42E-04
1.19E-01
3.48E-03
9.90E-05
5.24E-02
2.00E-02
1.68E-02
6.66E-05
6.16E-03
1.72E-02
3-59
-------
Final Regulatory Impact Analysis
Table 3.2-10 (cont'd). National Average Population Hazard Quotient for Chronic Noncancer Effects Across Census Tracts,
Without Controls in this Rule.
Pollutant
1,3-Butadiene
Acetaldehyde
Acrolein
Benzene
Chromium VI
Ethyl Benzene
Formaldehyde
Hexane
MTBE
Manganese
Naphthalene
Nickel
Styrene
Toluene
Xylenes
Target System
Reproductive
Respiratory
Respiratory
Immunological
Respiratory
Developmental
Respiratory
Neurological, Respiratory
Liver, Kidney, Ocular
Neurological
Respiratory
Respiratory, Immunological
Neurological
Respiratory, Neurological
Neurological
2020 average Hazard Quotient
major
8.25E-04
2.93E-03
1.78E-01
4.67E-04
6.28E-04
1.17E-05
4.34E-03
2.66E-04
4.36E-06
2.99E-02
1.09E-03
5.78E-03
3.29E-05
3.47E-04
7.69E-04
area&
other
7.30E-03
5.99E-03
1.09E+00
4.99E-03
1.17E-03
1.14E-04
1.02E-02
2.37E-03
1.91E-05
2.80E-02
1.51E-02
1.12E-02
1.96E-05
2.63E-03
7.35E-03
onroad
1.49E-02
5.58E-02
1.57E+00
1.58E-02
2.09E-04
1.72E-04
2.55E-02
6.92E-04
3.53E-05
3.37E-04
3.33E-03
5.95E-04
1.90E-05
2.89E-03
6.39E-03
nonroad
5.64E-03
1.05E-02
8.52E-01
3.65E-03
6.95E-05
5.44E-05
1.42E-02
2.82E-04
3.28E-05
5.59E-05
1.45E-03
6.83E-04
2.09E-06
5.78E-04
2.04E-03
total
(Including
background)
5.00E-02
1.24E-01
3.69E+00
3.61E-02
2.07E-03
3.52E-04
1.20E-01
3.61E-03
9.16E-05
5.83E-02
2.10E-02
1.83E-02
7.36E-05
6.45E-03
1.80E-02
2030 average Hazard Quotient
major
8.03E-04
2.90E-03
1.78E-01
4.63E-04
6.23 E-04
1.15E-05
4.23 E-03
2.65E-04
4.26E-06
3.08E-02
1.07E-03
5.78E-03
3.32E-05
3.44E-04
7.59E-04
area &
other
6.98E-03
6.02E-03
1.08E+00
4.96E-03
1.15E-03
1.12E-04
1.03E-02
2.32E-03
1.90E-05
2.81E-02
1.49E-02
1.10E-02
1.97E-05
2.63E-03
7.26E-03
onroad
1.72E-02
6.47E-02
1.82E+00
1.83E-02
2.54E-04
1.96E-04
2.95E-02
7.53E-04
3.44E-05
4. 11 E-04
3. 83 E-03
7.26E-04
2.22E-05
3.32E-03
7.33E-03
nonroad
6.43E-03
1.13E-02
9.62E-01
4.10E-03
7.32E-05
6.09E-05
1.53E-02
3.17E-04
3.62E-05
6.15E-05
1.65E-03
7.30E-04
2.32E-06
6.37E-04
2.25E-03
total (including
background)
5.26E-02
1.34E-01
4.04E+00
3.90E-02
2.10E-03
3.81E-04
1.25E-01
3.66E-03
9.38E-05
5.94E-02
2.14E-02
1.83E-02
7.74E-05
6.93E-03
1.90E-02
5-60
-------
Final Regulatory Impact Analysis
Table 3.2-11. National Respiratory Hazard Index for Chronic Noncancer Effects across
Census Tracts, Without Controls in this Rule.
Respiratory System Average Hazard Index
Year
1999
2015
2020
2030
background
0.12
0.12
0.12
0.11
major
0.16
0.17
0.19
0.19
area & other
1.32
1.17
1.14
1.13
on road
3.88
1.63
1.66
1.92
n on road
1.07
0.84
0.88
0.99
total (including
background)
6.54
3.92
3.99
4.35
Figure 3.2-7. Average Respiratory Hazard Index for U.S. Population (Aggregate of Hazard
Quotients for Individual Pollutants), Without Controls in this Rule.
7
«
Ł
x °
0
^ 4
C *r
-3
« °
N
T 9
-*- L.
1
n
u -
_
1999
i
2015
i
Year
2020
i
2030
...
D rnoDiie
• stationary
D background
3-61
-------
Final Regulatory Impact Analysis
Figure 3.2-8. U. S. Population at Various Non-Cancer Hazard Benchmarks due to
Exposure to Mobile Source Air Toxics, 1999 - 2030, Without Controls in this Rule.
400
DHQ<0.1
D0.1 <= HQ< 1
• 1 <= HQ< 10
DHQ>= 10
1999
2015
2020
2030
Year
Table 3.2-12. Distribution of Individual Cancer Risks for Mobile Source Air Toxics in
2020, Without Controls in this Rule.
Pollutant
Total Risk: All HAPs
1,3-Butadiene
Acetaldehyde
Benzene
Chromium VI
Formaldehyde
Naphthalene
Nickel
POM
2020 risk distribution
5th
percentile
4.71E-06
1.52E-07
1.09E-06
2.72E-06
3.85E-08
2.29E-09
1.59E-07
1.84E-09
1.26E-07
10th
percentile
6.08E-06
2.96E-07
1.19E-06
3.36E-06
7.93 E-08
2.89E-09
2.80E-07
4.09E-09
1.90E-07
25th
percentile
9.78E-06
1.06E-06
1.46E-06
4.84E-06
2.38E-07
4.12E-09
6.72E-07
1.39E-08
3.48E-07
Median
1.53E-05
2.30E-06
1.96E-06
6.93E-06
7.01E-07
5.75E-09
1.39E-06
4.60E-08
6.78E-07
75th
percentile
2.37E-05
3.60E-06
2.81E-06
l.OOE-05
1.81E-06
7.67E-09
2.61E-06
1.31E-07
1.19E-06
3.79E-05
5.47E-06
4.20E-06
1.48E-05
4.54E-06
1.05E-08
4.73E-06
3.04E-07
1.99E-06
95th
percentile
4.93E-05
7.70E-06
5.35E-06
1.86E-05
7.29E-06
1.29E-08
6.68E-06
5.06E-07
3.07E-06
3-62
-------
Final Regulatory Impact Analysis
Figure 3.2-9. 2020 County Median Cancer Risk for All Mobile Source Air Toxics, Without
Controls in this Rule.
Figure 3.2-10. 2020 County Median Cancer Risk for Benzene, Without Controls in this
Rule.
3-63
-------
Final Regulatory Impact Analysis
Figure 3.2-11. 2020 County Median Cancer Risk for Acetaldehyde, Without Controls in
this Rule.
Figure 3.2-12. 2020 County Median Cancer Risk for 1,3-Butadiene, Without Controls in
this Rule.
3-64
-------
Final Regulatory Impact Analysis
Figure 3.2-13. 1999 Comparison Between HAPEM6 and HAPEM5 Nationwide Individual
Benzene Cancer Risk, Without Controls in this Rule.
100,000,000
10,000,000
1,000,000
100,000
o 10,000
J5
§• 1,000
100
10
rfl
1
r^ r^ h
o o c
LJJ LJJ LJ
- t
3 t
J L
^ h
D C
JJ U
- r
3 C
J L
rfl
~- h-
•? *•?
JJ LLJ
r :
1^ 1^ CD CD
CD CD CD CD
LLJ LLJ LLJ LLJ
'
'
:
CD CD CD CŁ
CD CD CD C
LJJ LJJ LJJ U
3 C!
3 C
J U
•
'•
3 C!
3 C
J U
3 CŁ
3 C
J U
3 U
3 C
J U
'•
3 U
3 C
J U
3 U
3 C
J L
•3 u
p c
LJ U
3 U
3 C
J U
3 U
3 C
U U
3 LT
3 C
U U
3 U
3 C
J U
i n
3 LO •*
3 CD C
J LJJ L
a- ^r ^r
p cp cp
U LLJ LLJ
Population with Risk > Risk bin
IHAPEM6 DHAPEM5
3-65
-------
Final Regulatory Impact Analysis
Table 3.2-13. Distribution of Individual Hazard Quotients/Hazard Indices for Mobile
Source Air Toxics (from both Mobile and Stationary Sources) in 2020, Without Controls in
this Rule.
Pollutant
Acrolein
Respiratory System
2020 average Hazard Quotient or Hazard Index
5th
percentile
0.41
0.53
10th
percentile
0.61
0.75
25th
percentile
1.18
1.36
Median
2.31
2.57
75th
percentile
4.47
4.83
90th
percentile
8.05
8.54
95th
percentile
11.3
11.9
Figure 3.2-14. 2020 County Median Non-Cancer Hazard Index Respiratory Mobile Source
Air Toxics, Without Controls in this Rule.
3-66
-------
Final Regulatory Impact Analysis
3.2.1.2.4 Impacts of Controls on Average Inhalation Cancer Risks and Noncancer Hazards
The standards being finalized in this rule will substantially reduce inhalation cancer and
noncancer risk from exposure to air toxics emitted by mobile sources across the United States.
Table 3.2-14 shows that in 2030, the highway vehicle contribution to MSAT cancer risk will be
reduced on average 36% across the U.S., and the nonroad equipment contribution will be
reduced about 6%. In 2030, the highway vehicle contribution to benzene cancer risk will be
reduced on average by 43% across the U.S., and the nonroad contribution will be reduced by
11%. Table 3.2-15 summarizes the change in median and 95th percentile inhalation cancer risks
from benzene and all MSATs attributable to all outdoor sources in 2015, 2020, and 2030, with
the controls being finalized in this rule. Reductions are significantly larger for individuals in
the 95th percentile than in the 50th percentile. Thus, this rule is providing bigger benefits to
individuals experiencing the highest levels of risk. In states with high fuel benzene levels and
high cold start emissions, the cancer risk reduction from total MSATs is about 40% or higher
(Table 3.2-16).d Figure 3.2-15 depicts the impact on the mobile source contribution to
nationwide average population cancer risk from all MSATs and benzene in 2030. Nationwide,
the cancer risk attributable to total MSATs would be reduced by 30%, and the risk from mobile
source benzene would be reduced by 37%. Figures 3.2-16 and 3.2.-17 present the distribution of
percent reductions in average MSAT and benzene cancer risk, respectively, from all sources in
2030 with the controls being finalized in 2030. Table 3.2-17 shows reductions in hazard
quotients and hazard indices for acrolein and respiratory effects, respectively. Nationwide, the
mobile source contribution to the acrolein hazard quotient and respiratory hazard index would
both be reduced about 23%, and the highway vehicle contribution will be reduced about 35%.
Summary tables providing exposure and risk data by State, as well as maps of cancer risks and
noncancer hazards with controls and percent reductions with controls, can be found in the docket
for the rule.
It should be noted that the estimated total relative reductions are significant
underestimates, since we could not account for further reductions in emissions from transport,
i.e., background sources. In Section 3.2.1.4, we provide a quantitative estimate of the expected
reductions in background concentrations in future years. Again, as noted previously, since this
modeling did not include the 1.3 vol% maximum average fuel benzene level, reductions in risk
for some parts of the country, such as the Pacific Northwest, are underestimated.
d Reductions are likely to be higher than estimated by this modeling, due to the 1.3% maximum average fuel
benzene level.
3-67
-------
Final Regulatory Impact Analysis
Table 3.2-14. Contributions of Source Sectors to Nationwide Average Cumulative MSAT Cancer Risk, With and Without Controls,
2015, 2020, and 2030
Total MSATs
Reference
Control
% Difference
Benzene
Reference
Control
% Difference
201 5 Average Risks
major
1.17E-06
1.17E-06
0.0
1 .OOE-07
1 .OOE-07
0.3
area & other
5.76E-06
5.74E-06
0.3
1.13E-06
1.12E-06
1.3
total onroad
6.24E-06
4.98E-06
20.2
3.66E-06
2.73E-06
25.4
total
nonroad
1 .62E-06
1 .53E-06
5.3
8.25E-07
7.38E-07
10.5
total (including
background)
1 .97E-05
1 .83E-05
6.9
8.33E-06
7.30E-06
12.3
2020 Average Risks
major
1 .30E-06
1 .30E-06
0.0
1 .09E-07
1 .09E-07
0.3
area & other
6.08E-06
6.06E-06
0.3
1.17E-06
1.15E-06
1.3
total onroad
6.35E-06
4.58E-06
27.9
3.71 E-06
2.45E-06
34.0
total
nonroad
1 .67E-06
1 .58E-06
5.5
8.54E-07
7.62E-07
10.8
total (including
background)
2.03E-05
1 .84E-05
9.3
8.45E-06
7.09E-06
16.2
2030 Average Risks
major
1 .29E-06
1 .29E-06
0.0
1 .08E-07
1 .08E-07
0.3
area & other
6.02E-06
6.01 E-06
0.3
1.16E-06
1.15E-06
1.3
total onroad
7.37E-06
4.69E-06
36.3
4.29E-06
2.43E-06
43.4
total
nonroad
1 .87E-06
1 .77E-06
5.6
9.59E-07
8.54E-07
10.9
total (including
background)
2.14E-05
1 .86E-05
13.1
9.13E-06
7.15E-06
21.7
5-68
-------
-th
Table 3.2-15. Change in Median and 95 Percentile Inhalation Cancer Risk from Benzene
and all MSATs Attributable to Outdoor Sources in 2015, 2020, and 2030 with the Controls
Being Finalized in this Rule.
All
MSATs
Without
Controls
With
Controls
Percent
Change
Benzene
Without
Controls
With
Controls
Percent
Change
2015
median
l.SOxlO'5
1.41xlO'5
6
6.86xlO'6
6.17xlO'6
10
95th
4.75xlO'5
4.37xlO'5
8
1.82xlO'5
1.53xlO'5
16
2020
median
1.53xlO'5
1.40xlO'5
8
6.93xlO'6
6.02xlO'6
13
95th
4.93xlO'5
4.40xlO'5
11
1.86xlO'5
1.47xlO'5
21
2030
median
1.61xlO'5
1.42xlO'5
12
7.37xlO'6
6.06xlO'6
18
95th
5.28xlO'5
4.49xlO'5
15
2.06xlO'5
1.49xlO'5
28
Table 3.2-16. States with Highest Reductions in Average Benzene Cancer Risk Resulting
from Mobile Source Emissions, 2030.
State
Alaska
North Dakota
Washington
Minnesota
Wyoming
Montana
Idaho
Michigan
South Dakota
Oregon
Average Risk -
Reference Case
l.OlxlO'5
2.92xlO'6
1.39xlO'5
1.21xlO'5
2.38xlO'6
3.12xlO'6
5.03xlO'6
1.09xlO'5
2.73xlO'6
l.OlxlO'5
Average Risk -
Control Case
4.23xlO'6
1.68xlO'6
S.lOxlO'6
7.08xlO'6
1.39xlO'6
1.87xlO'6
3.02xlO'6
6.55xlO'6
1.66xlO'6
6.17xlO'6
Percent Difference
-58%
-42
-42
-42
-41
-40
-40
-40
-39
-39
3-69
-------
Figure 3.2-15. Contribution to Nationwide Average Population Cancer Risk from Mobile
Source MSATs and Benzene Emitted by Mobile Sources in 2030, Without and With
Controls in this Rule.
Q F OR
8P OR
v 7 P OR
i- b.b-Ub -
d)
o
O
d)
O) 4 F OR
35 ^t.t-uo -
5 ^ P OR
9 F OR
1 P OR
n F+nn
Total \
/I SATs
Ben
:ene
D Without Controls
• With Controls
Figure 3.2-16. Distribution of Percent Reductions in Median MSAT Cancer Risk, 2030, for
U.S. Counties with Controls in this Rule.
3-70
-------
Figure 3.2-17. Distribution of Percent Reductions in Median Benzene Cancer Risk, 2030,
for U.S. Counties With Controls in this Rule.
-46.519%--21.861%
-21.860%--15.843%
-15.842%--10.893%
-10.892%--5.932%
-5.931%-0.747%
As a result of the controls being finalized in this rule, the number of people above the 1 in
100,000 cancer risk level due to exposure to all mobile source air toxics from all sources will
decrease by over 11 million in 2020 and by about 17 million in 2030. The number of people
above the 1 in 100,000 increased cancer risk level from exposure to benzene from all sources
decreases by about 30 million in 2020 and 46 million in 2030 (Table 3.2-18).
3-71
-------
Table 3.2.-17. Reductions in Hazard Quotients and Hazard Indices for Acrolein and Respiratory Effects Due to MSAT Controls.
2015 Average Hazard Index/ Quotient
area &
other
1.17
1.17
0.0
1.13
totalonroad
1.63
1.35
16.7
1.54
total
nonroad
0.84
0.84
0.0
0.81
total (Including
background)
3.92
3.65
6.9
3.63
2020 Average Hazard Index/ Quotient
major
0.19
0.19
0.0
0.18
area &
other
1.14
1.14
0.0
1.09
total onroad
1.66
1.24
25.6
1.57
total
nonroad
0.88
0.88
0.0
0.85
total
(Including
background)
3.99
3.56
10.7
3.69
2030 Average Hazard Index/ Quotient
major
0.19
0.19
0.0
0.18
area &
other
1.13
1.13
0.0
1.08
total onroad
1.92
1.24
35.4
1.82
total
nonroad
0.99
0.99
0.0
0.96
total (Including
background)
4.35
3.67
15.6
4.04
5-72
-------
Table 3.2-18. Decrease in Number of People with Inhalation Exposure above the 1 in
100,000 Cancer Risk Level due to Inhalation Exposure from Ambient Sources, With
Controls in this Rule.
Year
2015
2020
2030
Benzene
21,697,000
30,031,000
46,360,000
All Mobile Source Air
Toxics
8,149,000
11,257,000
16,737,000
The standards being finalized will also impact on the number of people above various respiratory
hazard index levels (Table 3.2-19).
Table 3.2-19. Decrease in Number of People with Inhalation Exposure above a Respiratory
Hazard Index of One due to Inhalation Exposure from Ambient Sources, With Controls in
this Rule.
Year
2015
2020
2030
Decrease in Population
with Respiratory HI > 1
5,639,000
10,227,000
16,919,000
3-73
-------
3.2.1.3 Strengths and Limitations
Air quality, exposure, and risk were assessed using the best available suite of tools for
national-scale analysis of air toxics. The same general suite of tools was used in 1996 and 1999
NATA. The 1996 NATA was reviewed by EPA's Science Advisory Board, and the analyses
done for 1999 incorporate several changes in response to comments made in this peer review.
Among the improvements were:
• Improved emission inventory with detailed characterization of source categories within
the onroad and nonroad source sectors and more speciated data for some pollutant groups
(POM) within particular source categories.
• Speciation of chromium to hexavalent form based on emission sources rather than a
single number applied across all sources
• Improved surrogates for spatial allocation in EMS-HAP.
• Improved estimation of "background" concentrations for many pollutants. These
background levels were previously uniform across the country. Now, for many
pollutants, background levels are based on recent monitor data and spatially vary
depending on county population density.161
• Improved version of HAPEM, which includes more recent census data, commuting
algorithms and better characterization of exposure distributions through improvements in
modeling long-term activity patterns and variability in concentration levels in
microenvironments.
In addition to the improvements for the 1999 NATA, improvements were made in analyses
for this rule, including inventory improvements and updates to HAPEM discussed earlier.
The SAB expressed their belief that due to the limitations inherent in the analysis, the 1996
NATA should not be used to support regulatory action. However, the use of the improved
analyses in this rule does provide useful insight on the nature of the mobile source air toxics
problem and the possible public health improvements associated with this rule.
In addition to the strengths listed above, there are limitations due to uncertainty. The
inventory uncertainties are discussed in Chapter 2. There are a number of additional significant
uncertainties associated with the air quality, exposure and risk modeling. These uncertainties
result from a number of parameters including: development of county-level estimates from
broader geographic data (i.e., state, regional or national), surrogates used to allocate emissions to
census tracts, parameters used to characterize photochemical processes, long range transport,
terrain effects, deposition rates, human activity pattern parameters, assumptions about
relationships between ambient levels in different microenvironments, and dose-response
parameters. Uncertainties in dose-response parameters are discussed in Chapter 1 of the RIA.
The modeling also has certain key limitations: results are most accurate for large geographic
areas, exposure modeling does not fully reflect variation among individuals, non-inhalation
exposure pathways and indoor sources are not accounted for; and for some pollutants, the
ASPEN dispersion model may underestimate concentrations. Also, the 1999 NATA does not
3-74
-------
include default adjustments for early life exposures recently recommended in the Supplemental
Guidance for Assessing Susceptibility from Early-Life Exposure to Carcinogens.162 If
warranted, incorporation of such adjustments would lead to higher estimates of lifetime risk.
EPA will determine as part of the IRIS assessment process which substances meet the criteria for
making adjustments, and future assessments will reflect them.
As part of the 1999 NAT A, EPA compared ASPEN-modeled concentrations with
available, but geographically limited, ambient air quality monitoring data for 1999. For each
monitor-pollutant combination, EPA compared the annual average concentration estimated by
the ASPEN model at the exact geographical coordinates of the monitor location with the annual
average monitored value to get a point-to-point comparison between the model and monitor
concentrations. The agreement between model and monitor values for benzene was very good,
with a median model to monitor ratio of 0.95, and 74% of sites within a factor of 2. Agreement
for acetaldehyde was almost as good as benzene, but data suggest that ASPEN could be
underpredicting for other mobile source air toxics (see Table 3.2-20).
More detailed discussion of modeling limitations and uncertainties can be found on the 1999
NATA website.
Table 3.2-20. Agreement of 1999 Model and Monitors by Pollutant on a Point-to-Point
Basis Pollutants listed were Monitored in at least 30 Sites and in a Broad Geographical
Area (Several States)
Pollutant
Acetaldehyde
Benzene
Formaldehyde
Chromium
Manganese
Nickel
No. of
Sites
68
115
68
42
34
40
Median of
Ratios
0.92
0.95
0.64
0.29
0.4
0.53
Within
Factor of 2
74%
72%
60%
26%
44%
48%
Within
30%
44%
43%
28%
5%
15%
18%
Underestimated
56%
52%
76%
95%
91%
75%
In addition to the limitations and uncertainties associated with modeling the 1999 base
year, there are additional ones in the projection year modeling. For instance, the modeling is not
accounting for impacts of demographic shifts that are likely to occur in the future. Assumptions
about future-year meteorology introduce additional uncertainty in ambient concentrations and
resulting exposures. Another limitation is the use of 1999 "background" levels to account for
mid-range to long-range transport. However, since background is related to emissions far away
from receptors, these levels should decrease as those emissions decrease. For the proposed rule
we performed a sensitivity analysis for benzene, formaldehyde, acetaldehyde and 1,3-butadiene
to evaluate the potential bias introduced by this assumption. We used background estimates
scaled by the change in the proposed rule inventory for a future year relative to 1999. The
scaling factors applied to the background level for an individual county were based on emissions
for counties within 300 kilometers of that county's centroid. Our analysis indicated that using a
scaled background reduced benzene concentrations about 15% on average across the U. S in
2015, 2020, and 2030. Table 3.2-21 compares national average total concentrations from the
3-75
-------
proposed rule using 1999 versus scaled backgrounds. More details are provided in the technical
support document for the proposed rule.
163
Table 3.2-21. National Average Total Concentrations (All Sources and Background) for
2015, 2020, and 2030 using both the 1999 Background and the Scaled Backgrounds (Data
from Proposed Rule).
HAP
1,3 -Butadiene
Acetaldehyde
Benzene
Formaldehyde
Xylenes
Total Concentrations (ug ni 3) using 1999
Background
2015
9.81X10'2
9.66X10"1
9.13X10'1
1.22
1.55
2020
9.77xlO"2
9.36X10"1
9.02X10"1
1.22
1.61
2030
l.OOxlO"1
9.56X10"1
9.24X10"1
1.25
1.65
Total Concentrations (jig m 3) using Scaled
Concentrations
2015
7.57xlO"2
7.77X10"1
7.57X10"1
9.56X10"1
1.50
2020
7.50xlO"2
7.47X10"1
7.40X10"1
9.68X10"1
1.56
2030
7.86xlO"2
7.78X10"1
7.71 xlO'1
1.01
1.60
The largest impacts were in the Midwest as can be seen in Figure 3.2-19, which depicts
ratios of the ASPEN-modeled ambient benzene concentrations with an adjusted background
versus the 1999 background in 2020. Data tables with results of the sensitivity comparison by U.
S. County, along with maps of pollutant concentrations with and without an adjusted background
can be found in the docket for the rule.
While accounting for impacts of emission reductions on background levels would reduce
estimated population risks, it would increase estimated reductions in risk of control strategies in
a given year, since background levels would be reduced. Also, if the modeling accounted for
equipment and fuels in attached garages and increased risks from early lifetime exposures,
estimated risks and risk reductions from fuel benzene control would be larger.
3-76
-------
Figure 3.2-19. Ratios of Benzene Concentrations with and without an Adjusted
Background, 2020 (from modeling done to support proposed rule).
0.516-0.638
0.639-0.716
0.717-0.786
0.787 - 0.893
0.894-1.068
3.2.1.4. Perspective on Cancer Cases
We have not quantified the cancer-related health benefits of expected MSAT reductions
in terms of avoided cancer cases or dollars. The EPA Science Advisory Board (SAB)
specifically commented in their review of the 1996 National Air Toxics Assessment (NATA)
that these tools were not yet ready for use in a national-scale benefits analysis, because they did
not consider the full distribution of exposure and risk, or address sub-chronic health effects.164
While EPA has since improved many of these tools, there remain critical limitations for
estimating cancer incidence. For the MSATs of greatest concern, for example, we are currently
unable to estimate cessation lag, which is the time between reduction in exposure and decline in
risk to "steady state level."165 We have also not resolved the analytical challenges associated
with quantifying partial lifetime probabilities of cancer for different age groups or estimating
changes in survival rates over time. Indeed, some of these issues are likely to remain highly
uncertain for the foreseeable future.
We can, however, present some perspective on how average individual risks could
translate into cumulative excess cancer cases across the U.S. population over a lifetime,
assuming continuous exposure at a given level for 70 years. Cancer cases were estimated by
summing the distribution of individual cancer risks from the national-scale modeling done to
support this rule.
To estimate annual incidence, this would be divided by 70. However, without knowing
when within a lifetime cancer is more likely to occur, and without accounting for time-varying
3-77
-------
exposure, any estimate of incidence for a given calendar year is highly uncertain. We also note
that a proper calculation would entail the use of a life table of incidence rates within discrete age
ranges and a dose-response formulation expressing rate ratios as a function of benzene inhalation
exposure concentration.
In 2030, the cumulative excess average individual cancer risk from outdoor emissions of
mobile source air toxics is estimated at 2. IxlO"5. If the entire U. S. population (projected to be
about 364 million)166 were exposed to this level of risk over a 70-year lifetime, it would result in
about 7700 cancer cases, which translates into 110 annual cancer cases.
In its review of the 1996 NAT A, SAB recommended that if cancer cases were calculated
for benefits assessment, a "best estimate" of risk (rather than an upper bound), should be used.
We believe that the maximum likelihood unit risk range for benzene represents a best estimate.
In our analyses, we have used the upper end of this range, as did the 1999 NATA. If we used the
lower end of this range, incidence estimates would be lower by a factor of about 3.5. Following
is a discussion related to benzene specifically, including a discussion of the potential
implications of the limitations of our national-scale modeling, which were noted in Section
3.2.1.4.
In 2030, the national average inhalation individual cancer risk from outdoor mobile and
stationary sources of benzene, in the absence of the standards being finalized in this rule, is
estimated at approximately 9.1xlO"6, based on the modeling done for this rule. If the entire U. S.
population were exposed to that level of risk over a 70-year lifetime, it would result in
approximately 47 excess cancer cases per year (Equation 1).
(1) Excess Cancer Cases at 2030 Exposure Level =
(Average Individual Cancer Risk] x (2030 Population)
= 9.1xlO~6 x3.64x!08 ^3300
Annual Cancer Cases = 3300/70 = 47
As discussed in Section 3.1.3.3, EPA's estimate of risk due to exposure to benzene could
increase significantly if the influence of attached garages were included. When the exposures for
people with attached garages are averaged across the population, time-weighted average
individual exposures to benzene could increase by roughly 1.2 to 6.6 |ig/m3 (Appendix 3 A).
There is a great deal of uncertainty associated with these estimates. This could result in about
another 3400 to 18700 excess cancer cases (equation 3). The numerical ranges expressed here
may not fully address all sources of uncertainty involved in making these projections.
(3) Attached Garage Excess Cancer Cases =
(Average Exposure] x (Benzene LIRE) x (Population)
= (\.2-6.6jUg/m3)x(7.8xW-6/jUg/m3) x(3.64x!08)= 3400-18700
Annual Cancer Cases = 49 - 268
3-78
-------
Thus, including attached garages would increase the number of benzene-related excess cancer
cases to somewhere between 96 and 315 annually. This estimate would still not include higher
exposure levels from occupational exposures, vapor emissions from leaking underground storage
tanks, or other accidental releases into the environment. Any population risk characterization
that does not account for these factors underestimates the excess cancer related to benzene.
With the controls being finalized in this rule, average individual risk, not including
attached garage exposures, is reduced to 7.3xlO"6, which results in approximately 37 cancer
cases per year. Thus, excess leukemia cases would be reduced by 10 annually. A roughly 40%
reduction in overall benzene emissions could reduce attached garage exposures by approximately
0.5-2.6 |ig/m3 as well, thus reducing excess annual cancer cases from this source of exposure by
another estimated 20 to 100 excess cancer cases. Thus, this rule would prevent roughly 30 to
110 benzene-related excess cancer cases annually, assuming continuous lifetime exposure to
2030 levels, given the assumptions of population size and lifetime above, and not including
excess leukemia from occupational exposure or from leaking underground storage tanks.
Emission reductions in 2030 would reduce cancer cases not just in 2030, but also well beyond
this period. There would also be further unquantified reductions in incidence due to the other air
toxics reductions.
Such estimates should be interpreted with extreme caution since they could imply an artificial
sense of precision. Serious limitations include:
• As discussed in Chapter 1, the current unit risk estimate for benzene may underestimate
risk from leukemia, because some recent epidemiology data, including key studies
published after the most recent IRIS assessment, suggest a supralinear rather than linear
dose-response at low doses. However, the studies published after the most recent IRIS
assessment have not yet been formally evaluated by EPA as part of the IRIS review
process, and it is not clear whether these data provide sufficient evidence to reject a linear
dose-response curve. A better understanding of the biological mechanism of benzene-
induced leukemia is needed.
• Geographically heterogeneous percentage emissions reductions do not translate directly
into changes in ambient levels, exposure, and risk.
• The U.S. population would have experienced higher average exposures in previous years,
but this is not accounted for.
• The extent to which available studies of indoor air homes in with attached garage are
representative of the national housing stock is unknown.
• Cessation lag between reduction in exposure and reduction in risk is not accounted for.
• Differences in risk among various age groups are not known, and the age structure of the
U.S. population is expected to change over time.
3.2.2 Local-Scale Modeling
Modeling at the national or regional scale, such the modeling done for the NATA
National-Scale Assessment described in Section 3.2.1, is designed to identify and prioritize air
toxics, emission source types and locations which are of greatest potential concern in terms of
contributing to population risk. Such assessments also help elucidate patterns of exposure and
risk across broad geographic areas, and can help characterize trends in air toxics risk and
3-79
-------
potential impacts of controls at a broad geographic scale, as demonstrated above. However,
more localized assessments are needed to characterize and compare risks at local levels, and
identify potential "hotspots."
National or regional-scale assessments typically rely on a "top down" approach to
estimate emissions. Under a "top down" approach, emissions are estimated at the county level,
typically starting from more aggregated information (e.g., state or national level) on activity.
Spatial surrogates are then used to allocate emissions to grid cells or census tracts for modeling.
Use of more local data can greatly improve the characterization of the magnitude and distribution
of air toxic emissions. Air quality modeling can also be conducted with better spatial resolution
than is computationally feasible in a regional or national-scale assessment. As a result, spatial
gradients of air toxic concentrations and locations where the highest risks are likely to occur can
be more accurately identified.
Local-scale modeling is typically done using steady-state plume dispersion models, such
as the Integrated Source Complex (ISC) Model, the newly promulgated AERMOD (AMS/EPA
Regulatory Model), or non-steady-state puff models such as CALPUFF. These models have a
limited ability to simulate chemical reactions in the atmosphere. As discussed in Section 3.2.1,
grid-based models, such as CMAQ, which better simulate chemical processes, do not yet have
the spatial resolution of dispersion models. Significant advances are being made, however, in
combining features of grid-based models and plume/puff models. These advances are described
in a recent paper.167 A case study of diesel exhaust particulate matter in Wilmington, CA was
recently conducting employing some of these advances.168 The researchers combined Gaussian
and regional photochemical grid models. They found that local data, when modeled, provided a
much more refined picture of the magnitude and distribution of possible community "hot spots"
than more traditional, regional data, which rely on more default assumptions. An evaluation of
the approach determined that spatial allocation and emission rates contribute most to uncertainty
in model results, and this uncertainty could be substantially reduced through the collection and
integration of site specific information about the location of emission sources, and the activity
and emission rates of key sources affecting model concentrations. They conclude that for
neighborhood assessments, incorporating site-specific data can lead to improvement in modeled
estimates of concentrations, especially where site-specific data are lacking in regulatory
databases.
The Wilmington study discussed above also allocated motor vehicle emissions to
individual road "links," rather than using spatial surrogates to allocate county level vehicle
emissions to grid cells. In using spatial surrogates to allocate emissions, high local
concentrations may not be captured for environments near major roadways, which are often
clustered in urban centers. One local-scale assessment done in the Minneapolis-St. Paul area of
Minnesota, using such an inventory with the ISC model, found that the model tended to
overpredict at low monitored benzene concentrations and underpredict at high monitored
concentrations.169 Local-scale modeling using activity data for individual road links can better
characterize distributions of concentrations, and differentiate between locations near roadways
and those further away, as observed in the following studies.
As discussed in Section 3.1.3.2, local-scale modeling in Houston assigned emissions to
3-80
-------
individual road links. 17° Researchers at US EPA developed a methodology which utilized a
Geographic Information System (GIS) to allocate benzene emissions in Houston to major road
segments in an urban area and model the segments as elongated area sources. The Industrial
Source Complex Short Term (ISCST) dispersion model used both gridded and link-based
emissions to evaluate the effect of improved spatial allocation of emissions on ambient modeled
benzene concentrations. Allocating onroad mobile emissions to road segments improved the
agreement between modeled concentrations when compared with monitor observations, and also
resulted in higher estimated concentrations in the urban center where the density of
neighborhood streets is greater and the largest amount of traffic found. The calculated annual
average benzene model concentrations at monitor sites are compared to the observed annual
average concentrations in Figure 3.2-20. Most of the gridded model emissions show lower
benzene concentrations than both the link-based and observed monitor concentrations.
Allocating the onroad mobile emissions to road segments resulted in an increase in the average
benzene concentration, resulting in values that more closely match concentrations reported by
monitors.
Recent air quality modeling in Portland, OR using the CALPUFF dispersion model
assigned emissions to specific roadway links.171 The resulting data were used to develop a
regression model to approximate the CALPUFF predicted concentrations, determine the impacts
of roadway proximity on ambient concentration of three hazardous air pollutants (1,3-butadiene,
benzene, and diesel PM), and to estimate the zone of influence around roadways. Concentrations
were modeled at several distances from major roadways (0-50, 5-200, 200-400, and > 400
meters). For benzene, the resulting average concentrations were 1.29, 0.64, 0.40, and 0.12
ug/m3, respectively, illustrating the steep concentration gradient along roadways. There was a
zone of influence between 200 and 400 meters, with concentrations falling to urban background
levels beyond this distance. The overall mean motor vehicle benzene concentration modeled in
Portland was about 0.21 ug/m3, with concentrations increasing to 1.29 ug/m3 at model receptor
sites within 50 meters of a road. The results indicate that in order to capture localized impacts of
hazardous air pollutants in a dispersion model, there is a need to include individual roadway
links.
3-81
-------
Figure 3.2-20. Model to Monitor Comparisons of Houston Benzene Concentrations
5
— f
^
1
^
^
6.0-j
5.0-
4.5-
4.0-
3.5-
3.0-
2.0-
1.5-
1.0-
0.5-
0.0-
ISCST3 gndded D ISCST3 link-based A Moni:or
"''•" r»"-*
!',,^L^!?.
A ' A
D
i
D
I I I I
4567
If 0! if fore
, ;, p. _,!
W1 1 : > 1
V. .^ - * !
o
A
i
A recent review of local-scale modeling studies concluded that:172
1) Significant variations in air toxic concentrations occurred across the cities, with
highest concentrations occurring near the highest emitting sources, illustrating the need
for modeling on a local scale.
2) Increasing the receptor density near high emission sources changes the location of
maximum concentrations, illustrating the concentration gradients that can occur near high
emission sources and the importance of receptor placement and density for model
performance.
3) Allocating on-road mobile emissions to road segments improved the agreement
between modeled concentrations when compared with the observations, and also resulted
in higher estimated concentrations in the urban center.
4) It is important to refine the national emissions inventory for input into local air quality
model applications.
In another US EPA study, researchers provide a comparison of "top down" and "bottom
up" approaches to developing a motor vehicle emissions inventory for one urban area,
Philadelphia, in calendar year 1999.173 Under the "top down" approach, emissions were
estimated at the county level, typically starting from more aggregated information. Data on
vehicle miles traveled (VMT) in the metropolitan statistical area were allocated to counties using
population information. Default national model inputs (e.g. fleet characteristics, vehicle speeds)
rather than local data were also used. The "bottom up" approach utilizes vehicle activity data
from a travel demand model (TDM), and this "bottom up" approach estimates emission rates
using more local input data to better estimate levels and spatial distribution of onroad motor
vehicle emissions. TDM data can include information on the spatial distribution of vehicle
activity, speeds along those roads (which can have a large impact on emissions), and the
distribution of the VMT among vehicle classes for different speed ranges. These data can be
3-82
-------
used to more accurately estimate the magnitude of toxic emissions at the local scale and where
they occur. Both the spatial distribution of emissions and the total county emissions in the
Philadelphia area differed significantly between the top-down and the bottom-up methodologies
as shown in Table 3.2-22.
Table 3.2-22. Comparison of Annual 1999 Benzene Emissions from Two Approaches in
Philadelphia Area Counties
County
Camden
Delaware
Gloucester
Montgomery
Philadelphia
Total
Local (TDM)
Based
165
162
110
333
255
1,025
National
(NEI)
210
160
104
209
467
1,150
Percent
Difference
-27%
1%
6%
59%
-45%
-12%
In the case of Philadelphia County, using local registration distribution data resulted in
significantly lower air toxics emission factors and resultant emissions, while Montgomery
County showed higher emissions. In the 1999 National Air Toxics Assessment, higher county-
level emissions were generally associated with higher county-level average concentrations, so it
is anticipated that county-level concentrations will follow similar trends. However, in
microscale settings near specific road links, these results may not apply.
Local-scale modeling could also be improved by using local data on nonroad equipment
activity for lawn and garden, recreational, construction and other sectors. EPA's county-level
inventories used in NATA and other modeling are developed using activity allocated from the
national or state level using surrogates.
The use of more spatially refined emission inventories, in conjunction with other refined
air quality modeling techniques, improve the performance of air quality models. They also
enable better characterization of the magnitude and distribution of air toxic emissions, exposure
and risk in urban areas, including risks associated with locations heavily impacted by mobile
sources.
In conclusion, local scale modeling studies indicated higher concentrations of air toxics
than predicted by National scale analysis, particularly in near-source microenvironments such as
near roads. Thus, National scale analyses such as 1999 NATA are likely underestimating high
end exposures and risks.
3.3 Ozone
In this section we review the health and welfare effects of ozone. We also describe the
air quality monitoring and modeling data which indicate that people in many areas across the
country continue to be exposed to high levels of ambient ozone and will continue to be into the
3-83
-------
future. Emissions of volatile organic compounds (VOCs) from the gas cans subject to this final
rule have been shown to contribute to these ozone concentrations. Information on air quality was
gathered from a variety of sources, including monitored ozone concentrations, air quality
modeling forecasts conducted for this rulemaking, and other state and local air quality
information.
3.3.1 Science of Ozone Formation
Ground-level ozone pollution is formed by the reaction of VOCs and nitrogen oxides
(NOX) in the atmosphere in the presence of heat and sunlight. These pollutants, often referred to
as ozone precursors, are emitted by many types of pollution sources such as highway and
nonroad motor vehicles, gas cans, power plants, chemical plants, refineries, makers of consumer
and commercial products, industrial facilities, and smaller area sources.
The science of ozone formation, transport, and accumulation is complex.174 Ground-
level ozone is produced and destroyed in a cyclical set of chemical reactions, many of which are
sensitive to temperature and sunlight. When ambient temperatures and sunlight levels remain
high for several days and the air is relatively stagnant, ozone and its precursors can build up and
result in more ozone than typically would occur on a single high-temperature day. Ozone also
can be transported into an area from pollution sources found hundreds of miles upwind, resulting
in elevated ozone levels even in areas with low VOC or NOX emissions.
The highest levels of ozone are produced when both VOC and NOX emissions are present
in significant quantities on clear summer days. Relatively small amounts of NOX enable ozone to
form rapidly when VOC levels are relatively high, but ozone production is quickly limited by
removal of the NOX. Under these conditions NOX reductions are highly effective in reducing
ozone while VOC reductions have little effect. Such conditions are called "NOx-limited".
Because the contribution of VOC emissions from biogenic (natural) sources to local ambient
ozone concentrations can be significant, even some areas where man-made VOC emissions are
relatively low can be NOX -limited.
When NOX levels are relatively high and VOC levels relatively low, NOX forms inorganic
nitrates (i.e., particles) but relatively little ozone. Such conditions are called "VOC-limited."
Under these conditions, VOC reductions are effective in reducing ozone, but NOX reductions can
actually increase local ozone under certain circumstances. Even in VOC-limited urban areas,
NOX reductions are not expected to increase ozone levels if the NOX reductions are sufficiently
large.
Rural areas are usually NOx-limited, due to the relatively large amounts of biogenic VOC
emissions in many rural areas. Urban areas can be either VOC- or NOX -limited, or a mixture of
both, in which ozone levels exhibit moderate sensitivity to changes in either pollutant.
Ozone concentrations in an area also can be lowered by the reaction of nitric oxide with
ozone, forming nitrogen dioxide (NO2); as the air moves downwind and the cycle continues, the
NO2 forms additional ozone. The importance of this reaction depends, in part, on the relative
concentrations of NOX, VOC, and ozone, all of which change with time and location.
3-84
-------
The Clean Air Act (CAA) requires EPA to set National Ambient Air Quality Standards
(NAAQS) for wide-spread pollutants from diverse sources considered harmful to public health
and the environment. The CAA established two types of NAAQS: primary standards to protect
public health, secondary standards to protect public welfare. The primary and secondary ozone
NAAQS are identical. The 8-hour ozone standard is met when the 3-year average of the annual
4th highest daily maximum 8-hour ozone concentration is less than or equal to 0.08 ppm. (62 FR
38855, July 18, 1997)
3.3.2 Health Effects of Ozone
Exposure to ambient ozone contributes to a wide range of adverse health effects.6 These
health effects are well documented and are critically assessed in the EPA ozone Air Quality
Criteria Document (ozone AQCD) and EPA staff paper.175'176 We are relying on the data and
conclusions in the ozone AQCD and staff paper, regarding the health effects associated with
ozone exposure.
Ozone-related health effects include lung function decrements, respiratory symptoms,
aggravation of asthma, increased hospital and emergency room visits, increased asthma
medication usage, inflammation of the lungs, and a variety of other respiratory effects and
cardiovascular effects. People who are more susceptible to effects associated with exposure to
ozone include children, asthmatics and the elderly. There is also suggestive evidence that certain
people may have greater genetic susceptibility. Those with greater exposures to ozone, for
instance due to time spent outdoors (e.g., outdoor workers), are also of concern.
Based on a large number of scientific studies, EPA has identified several key health
effects associated with exposure to levels of ozone found today in many areas of the country.
Short-term (1 to 3 hours) and prolonged exposures (6 to 8 hours) to higher ambient ozone
concentrations have been linked to lung function decrements, respiratory symptoms, increased
hospital admissions and emergency room visits for respiratory problems.177'178'179'180'181'182
Repeated exposure to ozone can increase susceptibility to respiratory infection and lung
inflammation and can aggravate preexisting respiratory diseases, such as asthma.183'184'185'186'187
Repeated exposure to sufficient concentrations of ozone can also cause inflammation of the lung,
impairment of lung defense mechanisms, and possibly irreversible changes in lung structure,
which over time could lead to premature aging of the lungs and/or chronic respiratory illnesses,
such as emphysema and chronic bronchitis.188'189'190'191
Children and adults who are outdoors and active during the summer months, such as
construction workers and other outdoor workers, are among those most at risk of elevated ozone
exposures.192 Children and outdoor workers tend to have higher ozone exposures because they
typically are active outside, working, playing and exercising, during times of day and seasons
(e.g., the summer) when ozone levels are highest.193 For example, summer camp studies in the
Eastern United States and Southeastern Canada have reported significant reductions in lung
e Human exposure to ozone varies over time due to changes in ambient ozone concentration and because people
move between locations which have notable different ozone concentrations. Also, the amount of ozone delivered to
the lung is not only influenced by the ambient concentration but also by the individuals breathing route and rate.
3-85
-------
function in children who are active outdoors.194'195'196'197'198'199'200'201 Further, children are
more at risk of experiencing health effects from ozone exposure than adults because their
respiratory systems are still developing. These individuals (as well as people with respiratory
illnesses such as asthma, especially asthmatic children) can experience reduced lung function
and increased respiratory symptoms, such as chest pain and cough, when exposed to relatively
low ozone levels during prolonged periods of moderate exertion.202'203'204'205
3.3.3 Current 8-Hour Ozone Levels
The gas can emission reductions will assist 8-hour ozone nonattainment areas in reaching
the standard by each area's respective attainment date and assist 8-hour ozone maintenance areas
in maintaining the 8-hour ozone standard in the future. In this section and the next section we
present information on current and model-projected future 8-hour ozone levels.
A nonattainment area is defined in the CAA as an area that is violating a NAAQS or is
contributing to a nearby area that is violating the NAAQS. EPA designated nonattainment areas
for the 8-hour ozone NAAQS in June 2004. The final rule on Air Quality Designations and
Classifications for the 8-hour Ozone NAAQS (69 FR 23858, April 30, 2004) lays out the factors
that EPA considered in making the 8-hour ozone nonattainment designations, including 2001-
2003 measured data, air quality in adjacent areas, and other factors/
As of October 26, 2006, approximately 157 million people live in the 116 areas that are
currently designated as nonattainment for either failing to meet the 8-hour ozone NAAQS or for
contributing to poor air quality in a nearby area. There are 461 full or partial counties that make
up the 116 8-hour ozone nonattainment areas. Figure 3.3-1 illustrates the widespread nature of
these problems. Shown in this figure are counties designated as nonattainment for the 8-hour
ozone NAAQS, also depicted are PM2.5 nonattainment areas and the mandatory class I federal
areas. The 8-hour ozone nonattainment areas, nonattainment counties and populations are listed
in Appendix 3B to this RIA.
f An ozone design value is the concentration that determines whether the ozone levels recorded at a monitoring site
meet the NAAQS for ozone. The level of a design value is determined based on three consecutive-year monitoring
periods. For example, an 8-hour design value is the fourth highest daily maximum 8-hour average ozone
concentration measured over a three-year period at a given monitor. Greater detail on how these values are
determined (including how to account for missing values and other complexities) is given in Appendices H and I of
40 CFR Part 50. Due to the precision with which the standards are expressed (0.08 ppm for the 8-hour NAAQS
value), a violation of the 8-hour standard is defined as any design value greater than or equal to 0.085 ppm, or 85
ppb. For any particular county, the design value is the highest design value from amongst all the monitors having
valid design values within that county. If there are no ozone monitors located in a particular county, that county is
not assigned a design value. However, readers should note that ozone design values represent air quality over a
broad area and the absence of a design value for a specific county does not imply that that county is in compliance
with the NAAQS for ozone. Therefore, our analysis may underestimate the number of counties with ozone levels,
i.e., design values, which are above the level of the ozone NAAQS.
3-86
-------
Figure 3.3.-1. 8-Hour Ozone and PM2.s Nonattainment Areas and Mandatory Class I
Federal Areas
^- -- * '
* _ y T
* -i
S _ k~ C?
'•*- * * "Cfe
u"^7- < r?
-i—-_ \
Legend
^^| PM and Ozone NonAttainment
^ Ozone NonAttainment
pm25 NonAttainment
^^1 Class I Areas
Counties designated as 8-hour ozone nonattainment were categorized, on the basis of
their one-hour ozone design value, as Subpart 1 or Subpart 2 (69 FR 23951, April 30, 2004).
Areas categorized as Subpart 2 were then further classified, on the basis of their 8-hour ozone
design value, as marginal, moderate, serious, severe or extreme. The maximum attainment date
assigned to an ozone nonattainment area is based on the area's classification.
Table 3B-1 presents the 8-hour ozone nonattainment areas, their 8-hour design values,
and their category or classification. States with 8-hour ozone nonattainment areas are required to
take action to bring those areas into compliance prior to the ozone season in the attainment year.
Based on the final rule designating and classifying 8-hour ozone nonattainment areas, most 8-
hour ozone nonattainment areas will be required to attain the 8-hour ozone NAAQS in the 2007
to 2013 time frame and then be required to maintain the 8-hour ozone NAAQS thereafter.8 The
gas can emission standards being finalized in this action will become effective in 2009. Thus,
g The Los Angeles South Coast Air Basin 8-hour ozone nonattainment area will have to attain before June 15, 2021.
3-87
-------
the expected ozone precursor emission inventory reductions from the standards finalized in this
action will be useful to States in attaining and/or maintaining the 8-hour ozone NAAQS.
EPA's review of the ozone NAAQS is currently underway and a proposed decision in
this review is scheduled for June 2007 with a final rule scheduled for March 2008. If the ozone
NAAQS is revised then new nonattainment areas could be designated. While EPA is not relying
on it for purposes of justifying this rule, the emission reductions from this rulemaking would also
be helpful to states if there is an ozone NAAQS revision.
3.3.4 Projected 8-Hour Ozone Levels
Recent air quality modeling predicts that without additional local, regional or national
controls there will continue to be a need for reductions in 8-hour ozone concentrations in some
areas in the future. In the following sections we describe recent ozone air quality modeling from
the CAIR analysis as well as results of the ozone response surface metamodel (RSM) analysis
we completed to assess the potential ozone impacts resulting from the VOC emissions controls
for gas cans.
3.3.4.1 CAIR Ozone Air Quality Modeling
Recently ozone air quality analyses were performed for the Clean Air Interstate Rule
(CAIR), which was promulgated by EPA in 2005. The Comprehensive Air Quality Model with
Extension (CAMx) was used as the tool for simulating base and future year concentrations of
ozone in support of the CAIR ozone air quality assessment. The CAIR analysis included all final
federal rules up to and including CAIR controls. Details on the air quality modeling are
provided in the Air Quality Modeling Technical Support Document for the Final Clean Air
Interstate Rule, included in the docket for this final rule.206
Air quality modeling performed for CAIR indicates that in the absence of additional
controls, counties with projected 8-hour ozone concentrations greater than or equal to 85 ppb are
likely to persist in the future. The CAIR analysis provided estimates of future ozone levels
across the country. For example, in 2010, in the absence of controls beyond those relied on for
the CAIR modeling, we project that 24 million people would live in 37 Eastern counties with 8-
hour ozone concentrations at and above 85 ppb, see Table 3.3-l.h Table 3.3-1 also lists the 148
Eastern counties, where 61 million people are projected to live, with 2010 projected design
values that do not violate the 8-hour ozone NAAQS but are within ten percent of it, in the
absence of emission reductions beyond those considered in the CAIR modeling. These are
counties that are not projected to violate the standard, but to be close to it. The rule may help
ensure that these counties continue to maintain their attainment status and the emission
reductions from this final rule will be included by the states in their baseline inventory modeling
for their ozone maintenance plans.
h Counties forecast to remain in nonattainment may need to adopt additional local or regional controls to attain the
standards by dates set pursuant to the Clean Air Act. The emissions reductions associated with this proposed rule
would help these areas attain the ozone standard by their statutory date.
3-88
-------
Table 3.3-1. Eastern Counties with 2010 projected 8-hour Ozone Concentrations
Above and within 10% of the 8-hour Ozone Standard
State
Arkansas
Connecticut
Connecticut
Connecticut
Connecticut
Connecticut
Connecticut
D.C.
Delaware
Delaware
Delaware
Georgia
Georgia
Georgia
Georgia
Georgia
Georgia
Georgia
Georgia
Georgia
Illinois
Illinois
Illinois
Illinois
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Indiana
Kentucky
Louisiana
Louisiana
Louisiana
Louisiana
Louisiana
Louisiana
Maine
2010 Projected
8-hour Ozone
County Concentration (ppb)a 2000 popb
Crittenden Co
Fairfield Co
Hartford Co
Middlesex Co
New Haven Co
New London Co
Tolland Co
Washington Co
Kent Co
New Castle Co
Sussex Co
Bibb Co
Cobb Co
Coweta Co
De Kalb Co
Douglas Co
Fayette Co
Fulton Co
Henry Co
Rockdale Co
Cook Co
Jersey Co
Lake Co
McHenry Co
Boone Co
Clark Co
Hamilton Co
Hancock Co
La Porte Co
Lake Co
Madison Co
Marion Co
Porter Co
Shelby Co
St Joseph Co
Campbell Co
Bossier Parish
East Baton Rouge Parish
Iberville Parish
Jefferson Parish
Livingston Parish
West Baton Rouge Parish
Hancock Co
80.8
92.2
80.1
90.6
91.3
83.4
82.7
85.0
78.7
84.7
80.3
80.0
79.4
76.6
81.9
78.7
76.7
85.1
80.3
80.4
81.8
77.0
76.8
76.6
78.1
78.4
81.7
80.4
81.8
82.8
78.6
79.6
81.1
81.6
77.8
81.5
77.0
80.6
79.4
78.6
77.8
78.8
80.5
50,866
882,567
857,183
155,071
824,008
259,088
136,364
572,058
126,697
500,264
156,638
153,887
607,750
89,215
665,864
92,174
91,263
816,005
119,341
70,111
5,376,739
21,668
644,356
260,077
46,107
96,472
182,740
55,391
110,106
484,563
133,358
860,453
146,798
43,445
265,559
88,616
98,310
412,852
33,320
455,466
91,814
21,601
51,791
2010popc
52,889
891,694
859,080
164,202
829,181
267,199
142,988
554,474
139,376
534,631
181,962
158,291
744,488
111,522
698,335
114,380
117,580
855,826
153,957
87,977
5,363,464
22,905
731,690
307,400
54,035
107,096
230,565
65,282
111,566
489,220
137,710
879,932
165,350
46,565
275,031
92,109
110,838
465,411
33,089
493,359
124,895
22,672
53,886
3-89
-------
State
Maine
Maryland
Maryland
Maryland
Maryland
Maryland
Maryland
Maryland
Maryland
Maryland
Maryland
Massachusetts
Massachusetts
Massachusetts
Massachusetts
Massachusetts
Massachusetts
Massachusetts
Michigan
Michigan
Michigan
Michigan
Michigan
Michigan
Michigan
Michigan
Michigan
Michigan
Michigan
Michigan
Michigan
Missouri
Missouri
Missouri
Missouri
Missouri
New Hampshire
New Jersey
New Jersey
New Jersey
New Jersey
New Jersey
New Jersey
New Jersey
New Jersey
New Jersey
2010 Projected
8-hour Ozone
County Concentration (ppb)a 2000 popb
York Co
Anne Arundel Co
Baltimore Co
Carroll Co
Cecil Co
Charles Co
Frederick Co
Harford Co
Kent Co
Montgomery Co
Prince Georges Co
Barnstable Co
Bristol Co
Essex Co
Hampden Co
Hampshire Co
Middlesex Co
Suffolk Co
Allegan Co
Benzie Co
Berrien Co
Cass Co
Genesee Co
Macomb Co
Mason Co
Muskegon Co
Oakland Co
Ottawa Co
St Clair Co
Washtenaw Co
Wayne Co
Clay Co
Jefferson Co
St Charles Co
St Louis City
St Louis Co
Hillsborough Co
Atlantic Co
Bergen Co
Camden Co
Cumberland Co
Gloucester Co
Hudson Co
Hunterdon Co
Mercer Co
Middlesex Co
80.2
88.6
83.7
80.0
89.5
78.7
78.1
92.8
85.8
79.3
84.2
83.6
83.0
81.7
80.2
78.0
79.1
78.1
82.1
77.9
78.1
78.2
76.7
85.4
78.9
82.0
80.7
76.6
80.6
81.0
84.7
76.5
76.7
80.5
79.4
80.5
76.6
80.4
86.0
91.6
84.4
91.3
84.3
88.6
95.2
92.1
186,742
489,656
754,292
150,897
85,951
120,546
195,277
218,590
19,197
873,341
801,515
222,230
534,678
723,419
456,228
152,251
1,465,396
689,807
105,665
15,998
162,453
51,104
436,141
788,149
28,274
170,200
1,194,155
238,314
164,235
322,895
2,061,161
184,006
198,099
283,883
348,188
1,016,315
380,841
252,552
884,118
508,932
146,438
254,673
608,975
121,989
350,761
750,162
2010 pop0
201,082
543,785
792,284
179,918
96,574
145,763
234,304
268,207
20,233
940,126
842,221
249,495
558,460
747,556
452,718
158,130
1,486,428
674,179
121,415
17,849
164,727
53,544
441,196
838,353
30,667
175,901
1,299,592
277,400
178,391
344,398
1,964,209
213,643
230,539
341,686
324,156
1,024,964
412,071
269,754
898,450
509,912
149,595
278,612
607,256
139,641
359,912
805,537
3-90
-------
State
New Jersey
New Jersey
New Jersey
New Jersey
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
New York
North Carolina
North Carolina
North Carolina
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Oklahoma
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
2010 Projected
8-hour Ozone
County Concentration (ppb)a 2000 popb
Monmouth Co
Morris Co
Ocean Co
Passaic Co
Bronx Co
Chautauqua Co
Dutchess Co
Erie Co
Essex Co
Jefferson Co
Monroe Co
Niagara Co
Orange Co
Putnam Co
Queens Co
Richmond Co
Suffolk Co
Westchester Co
Mecklenburg Co
Rowan Co
Wake Co
Allen Co
Ashtabula Co
Butler Co
Clermont Co
Clinton Co
Cuyahoga Co
Delaware Co
Franklin Co
Geauga Co
Hamilton Co
Knox Co
Lake Co
Lorain Co
Lucas Co
Medina Co
Portage Co
Summit Co
Trumbull Co
Warren Co
Wood Co
Tulsa Co
Allegheny Co
Armstrong Co
Beaver Co
Berks Co
86.4
85.5
100.3
79.7
79.7
81.8
81.0
86.9
77.6
80.5
76.9
82.3
77.1
82.3
78.3
87.1
90.8
84.7
81.4
80.1
77.2
76.8
83.5
78.0
78.0
81.4
77.3
77.3
81.9
86.6
78.6
76.5
82.2
78.5
80.0
76.5
79.8
82.4
79.7
80.0
77.4
79.2
81.9
79.7
79.6
81.7
615,301
470,212
510,916
489,049
1,332,649
139,750
280,150
950,265
38,851
111,738
735,343
219,846
341,367
95,745
2,229,379
443,728
1,419,369
923,459
695,453
130,340
627,846
108,473
102,728
332,806
177,977
40,543
1,393,977
109,989
1,068,977
90,895
845,302
54,500
227,511
284,664
455,053
151,095
152,061
542,898
225,116
158,383
121,065
563,299
1,281,665
72,392
181,412
373,637
2010 pop0
670,971
500,033
572,364
495,610
1,298,206
139,909
291,098
953,085
39,545
113,075
745,350
220,407
371,434
107,967
2,239,026
488,728
1,472,127
944,535
814,088
143,729
787,707
106,900
104,850
384,410
205,365
47,137
1,348,313
136,125
1,142,894
102,083
843,226
59,435
237,161
292,040
447,302
173,985
162,685
552,567
226,157
186,219
129,124
610,536
1,259,040
72,829
183,693
388,194
3-91
-------
State
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Rhode Island
Rhode Island
Rhode Island
South Carolina
Tennessee
Tennessee
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Texas
Virginia
Virginia
Virginia
Virginia
Virginia
Virginia
Virginia
Virginia
Virginia
Wisconsin
Wisconsin
2010 Projected
8-hour Ozone
County Concentration (ppb)a 2000 popb
Bucks Co
Cambria Co
Chester Co
Dauphin Co
Delaware Co
Erie Co
Franklin Co
Lancaster Co
Lehigh Co
Mercer Co
Montgomery Co
Northampton Co
Philadelphia Co
Washington Co
Westmoreland Co
York Co
Kent Co
Providence Co
Washington Co
Richland Co
Sevier Co
Shelby Co
Brazoria Co
Collin Co
Dallas Co
Denton Co
Galveston Co
Gregg Co
Harris Co
Jefferson Co
Johnson Co
Montgomery Co
Tarrant Co
Alexandria City
Arlington Co
Charles City Co
Fairfax Co
Hampton City
Hanover Co
Henrico Co
Loudoun Co
Suffolk City
Door Co
Kenosha Co
94.3
76.9
85.4
80.8
84.0
79.1
80.2
83.6
82.1
78.1
87.6
81.8
89.9
77.3
76.7
79.4
86.2
81.2
84.2
76.9
76.5
76.7
84.1
82.5
82.2
86.8
84.6
79.1
97.4
85.0
78.2
81.2
87.2
80.9
86.0
77.7
85.4
78.7
80.9
78.2
78.6
77.5
82.1
91.0
597,635
152,598
433,501
251,798
550,863
280,843
129,313
470,657
312,090
120,293
750,097
267,066
1,517,549
202,897
369,993
381,750
167,090
621,602
123,546
320,677
71,170
897,471
241,767
491,675
2,218,899
432,976
250,158
111,379
3,400,577
252,051
126,811
293,768
1,446,219
128,283
189,453
6,926
969,749
146,437
86,320
262,300
169,599
63,677
27,961
149,577
2010 pop0
648,796
146,811
478,460
265,019
543,169
284,835
135,088
513,684
323,215
122,546
772,849
279,797
1,420,803
205,153
372,941
404,807
174,126
621,355
137,756
349,826
96,097
958,501
281,960
677,868
2,382,657
554,033
283,963
121,241
3,770,129
260,847
157,545
413,048
1,710,920
130,422
193,370
7,382
1,085,483
153,246
98,586
294,174
214,469
69,003
30,508
166,359
3-92
-------
2010 Projected
8-hour Ozone
State County Concentration (ppb)a 2000 popb
Wisconsin Kewaunee Co
Wisconsin Manitowoc Co
Wisconsin Milwaukee Co
Wisconsin Ozaukee Co
Wisconsin Racine Co
Wisconsin Sheboygan Co
Number of Violating Counties
Population of Violating Counties
Number of Counties within 10%
Population of Counties within 10%
79.9
80.0
82.1
85.8
83.9
87.7
37
148
20,187
82,887
940,164
82,317
188,831
112,646
22,724,010
58,453,962
2010 pop0
20,538
83,516
922,943
95,549
199,178
118,866
24,264,574
61,409,062
a) Bolded concentrations indicate levels above the 8-hour ozone standard.
b) Populations are based on 2000 census data.
c) Populations are based on 2000 census projections.
3.3.4.2 Ozone Response Surface Metamodel Methodology
We performed ozone air quality modeling simulations for the Eastern United States using
the ozone RSM. The ozone RSM is a screening-level air quality modeling tool that allows users
to quickly assess the estimated air quality changes over the modeling domain. The ozone RSM
is a model of a full-scale air quality model and is based on statistical relationships between
model inputs and outputs obtained from the full-scale air quality model. In other words, the
ozone RSM uses statistical techniques to relate a response variable to a set of factors that are of
interest, e.g., emissions of precursor pollutants from particular sources and locations. The
following section describes the modeling methodology, including the development of the multi-
dimensional experimental design for control strategies and implementation and verification of
the RSM technique. Additional detail is available in the Air Quality Modeling Technical
Support Document (AQMTSD) for this rule.207
The foundation for the ozone response surface metamodeling analyses was the CAMx
modeling done in support of the final Clean Air Interstate Rule (CAIR). The CAIR modeling is
fully described in the CAIR Air Quality Modeling Technical Support Document, but a brief
description is provided below.208 The modeling procedures used in the CAIR analysis (e.g.,
domain, episodes, meteorology) have been used for several EPA rulemaking analyses over the
past five years and are well-established at this point.
The ozone RSM uses the 2015 controlled CAIR emissions inventory as its baseline.209
This inventory does not include the gas can emissions that are being controlled in this rule. The
uncontrolled and controlled gas can emissions have been incorporated into the base and control
runs of the ozone RSM (see Section 2.1 for more detail about the gas can emissions inventory).
The inventory also does not include the higher estimates of cold temperature emissions for
gasoline vehicles developed for this rule; however, these emissions are not likely to have a
significant impact on ozone formation. Finally, the inventory includes an error in mobile source
NOx for 13 Northeastern states. The impact of this error is minimized as the model is used in a
relative way. Because the base years of our air quality modeling analysis are 2020 and 2030, we
3-93
-------
extrapolate the model from 2015 to 2020 and 2030. Additional detail on how the model was
extrapolated to reflect gas can emissions and various projection years is included in the
AQMTSD for this final rule.210
The modeling simulations that comprised the metamodeling were conducted using
CAMx version 3.10. It should be noted that because the ozone RSM is built from CAMx air
quality model runs, it therefore has the same strengths and limitations of the underlying model
and its inputs. CAMx is a non-proprietary computer model that simulates the formation and fate
of photochemical oxidants including ozone for given input sets of meteorological conditions and
emissions. The gridded meteorological data for three historical episodes were developed using
the Regional Atmospheric Modeling System (RAMS), version 3b.211 In all, 30 episode days
were modeled using frequently-occurring, ozone-conducive, meteorological conditions from the
summer of 1995. Emissions estimates were developed for the evaluation year (1995) as well as a
future year (2015).
The CAMx model applications were performed for a domain covering all, or portions of,
37 States (and the District of Columbia) in the Eastern U.S., as shown in Figure 3.3-2. The
domain has nested horizontal grids of 36 km and 12 km. However, the output data from the
metamodeling is provided at a 12 km resolution (i.e., cells from the outer 36 km cells populate
the nine finer scale cells, as appropriate). Although the domain of the ozone RSM is the 37
Eastern states, the gas can controls are a nationwide program. Section 2.1.3 describes the
nationwide inventory reductions that could be achieved by the gas can controls. Section 2.1.1.2
also details the states that have their own gas can control programs and how the controls
finalized here impact states which already have gas can control programs.
3-94
-------
Figure 3.3-2. Map of the CAMx Domain used for MSAT Ozone Metamodeling
The ozone RSM used for assessing the impacts of gas can emission reductions was
developed broadly to look at various control strategies with respect to attaining the 8-hour ozone
NAAQS. The experimental design for the ozone RSM covered three key areas: type of precursor
emission (NOx or VOC), emission source type (i.e., onroad vehicles, nonroad vehicles, area
sources, electrical generating utility (EGU) sources, and non-utility point sources), and location
in or out of a 2015 model-projected residual ozone nonattainment area. This resulted in a set of
14 emissions factors. Since some of the spillage emissions associated with gas cans are currently
included in the NONROAD emissions model, for the purposes of the ozone RSM we have
included gas can emissions as part of the nonroad factor in our air quality modeling.
The 14 emission factors were randomly varied and used as inputs to CAMx. The
experimental design for these 14 factors was developed using a Maximin Latin Hypercube
method. Based on a rule of thumb of 10 runs per factor, we developed an overall design with
154 runs (a base case plus 139 control runs plus 10 evaluation runs plus 4 boundary condition
runs). The range of emissions reductions considered within the metamodel ranged from 0 to 120
percent of the 2015 CAIR emissions. This experimental design resulted in a set of CAMx
simulations that serve as the inputs to the ozone response surface metamodel. Because the
metamodeling was going to be used to assess the impacts of the gas can standards, the
3-95
-------
experimental design also included oversampling in the range of 0 to 10 percent control for the
nonroad VOC sector, as well as CAMx runs that only included VOC controls.
To develop a response surface approximation to CAMx, we used a multidimensional
kriging approach, implemented through the MIXED procedure in SAS. We modeled the
predicted changes in ozone in each CAMx grid cell as a function of the weighted average of the
modeled responses in the experimental design. A response-surface was then fit for the ozone
design value metric. Validation was performed and is summarized in the AQMTSD. The
validation exercises indicated that the ozone RSM replicates CAMx response to emissions
changes very well for most emissions combinations and in most locations.
The assessment of gas can controls conducted for this analysis involved adjusting the
nonroad mobile source VOC emissions both in and out of ozone nonattainment areas and looking
at the impact on the 8-hour ozone design value metric. We created an input or adjustment factor
for the nonroad mobile source VOC emission factor by adding future year gas can emission
estimates to the projected CAIR emission inventory and then relating the future year emissions
estimate to 2015. For this assessment the future years modeled are 2020 and 2030.
3.3.4.3 Ozone Response Surface Metamodel Results
This section summarizes the results of our modeling of ozone air quality impacts in the
future with and without the reductions in gas can emissions. Based upon our previous CAIR air
quality modeling, we anticipate that without emission reductions beyond those already required
under promulgated regulations and approved SIPs, ozone nonattainment will likely persist into
the future.
The inventories that underlie the ozone modeling conducted for this rulemaking included
emission reductions from all current or committed federal, state, and local controls, including the
recent CAIR. There was no attempt to examine the prospects of areas attaining or maintaining
the 8-hour ozone standard with possible additional future controls (i.e., controls beyond current
or committed federal, State, and local controls).
According to the ozone response surface metamodel (RSM), the gas can controls are
projected to result in a very small population-weighted net improvement in future ozone. The
net improvement is generally so small as to be rendered insignificant when presenting design
values. The model changes are smaller than the precision with which the ozone standard is
expressed (0.08 parts per million (ppm)) and to which 8-hour ozone data is reported.1
Nonetheless, there are some areas where the ozone improvement is more significant. These
areas include Chicago, Milwaukee, Detroit and New York City. It is also important to note that
the ozone RSM results indicate that the counties which are projected to experience the greatest
improvement in ozone design values are generally also those that are projected to have the
highest ozone design values. Those counties that are projected to experience an extremely small
increase in ozone design values generally have design values that are lower, below 70 ppb. The
results from the metamodeling projections indicate a net overall improvement in future 8-hour
1 Appendix I of 40 CFR Part 50.
3-96
-------
ozone design values due to the gas can controls, when weighted by population. The AQMTSD,
contained in the docket for this final rule, includes additional detail on the ozone RSM results.
3.3.5 Environmental Effects of Ozone Pollution
There are a number of public welfare effects associated with the presence of ozone in the
ambient air.212 In this section we discuss the impact of ozone on plants, including trees,
agronomic crops and urban ornamentals.
3.3.5.1 Impacts on Vegetation
The ozone AQCD notes that "ozone affects vegetation throughout the United States,
impairing crops, native vegetation, and ecosystems more than any other air pollutant."213 Like
carbon dioxide (CO2) and other gaseous substances, ozone enters plant tissues primarily through
apertures (stomata) in leaves in a process called "uptake." To a lesser extent, ozone can also
diffuse directly through surface layers to the plant's interior.214 Once sufficient levels of ozone, a
highly reactive substance, (or its reaction products) reaches the interior of plant cells, it can
inhibit or damage essential cellular components and functions, including enzyme activities,
lipids, and cellular membranes, disrupting the plant's osmotic (i.e., water) balance and energy
utilization patterns.215'216 This damage is commonly manifested as visible foliar injury such as
chlorotic or necrotic spots, increased leaf senescence (accelerated leaf aging) and/or reduced
photosynthesis. All these effects reduce a plant's capacity to form carbohydrates, which are the
primary form of energy used by plants.217 With fewer resources available, the plant reallocates
existing resources away from root growth and storage, above ground growth or yield, and
reproductive processes, toward leaf repair and maintenance. Studies have shown that plants
stressed in these ways may exhibit a general loss of vigor, which can lead to secondary impacts
that modify plants' responses to other environmental factors. Specifically, plants may become
more sensitive to other air pollutants, more susceptible to disease, insect attack, harsh weather
(e.g., drought, frost) and other environmental stresses. Furthermore, there is some evidence that
ozone can interfere with the formation of mycorrhiza, essential symbiotic fungi associated with
the roots of most terrestrial plants, by reducing the amount of carbon available for transfer from
the host to the symbiont.218
Ozone can produce both acute and chronic injury in sensitive species depending on the
concentration level and the duration of the exposure. Ozone effects also tend to accumulate over
the growing season of the plant, so that even lower concentrations experienced for a longer
duration have the potential to create chronic stress on sensitive vegetation. Not all plants,
however, are equally sensitive to ozone. Much of the variation in sensitivity between individual
plants or whole species is related to the plant's ability to regulate the extent of gas exchange via
leaf stomata (e.g., avoidance of Os uptake through closure of stomata).219'220'221 Other resistance
mechanisms may involve the intercellular production of detoxifying substances. Several
biochemical substances capable of detoxifying ozone have been reported to occur in plants
including the antioxidants ascorbate and glutathione. After injuries have occurred, plants may be
capable of repairing the damage to a limited extent.222 Because of the differing sensitivities
among plants to ozone, ozone pollution can also exert a selective pressure that leads to changes
in plant community composition. Given the range of plant sensitivities and the fact that
3-97
-------
numerous other environmental factors modify plant uptake and response to ozone, it is not
possible to identify threshold values above which ozone is consistently toxic for all plants. The
next few paragraphs present additional information on ozone damage to trees, ecosystems,
agronomic crops and urban ornamentals.
Ozone also has been shown conclusively to cause discernible injury to forest trees.223'224
In terms of forest productivity and ecosystem diversity, ozone may be the pollutant with the
greatest potential for regional-scale forest impacts.225 Studies have demonstrated repeatedly that
ozone concentrations commonly observed in polluted areas can have substantial impacts on plant
function.226'227
Because plants are at the center of the food web in many ecosystems, changes to the plant
community can affect associated organisms and ecosystems (including the suitability of habitats
that support threatened or endangered species and below ground organisms living in the root
zone). Ozone impacts at the community and ecosystem level vary widely depending upon
numerous factors, including concentration and temporal variation of tropospheric ozone, species
composition, soil properties and climatic factors.228 In most instances, responses to chronic or
recurrent exposure in forested ecosystems are subtle and not observable for many years. These
injuries can cause stand-level forest decline in sensitive ecosystems.229'230'231 It is not yet
possible to predict ecosystem responses to ozone with much certainty; however, considerable
knowledge of potential ecosystem responses has been acquired through long-term observations
in highly damaged forests in the United States.
Laboratory and field experiments have also shown reductions in yields for agronomic
crops exposed to ozone, including vegetables (e.g., lettuce) and field crops (e.g., cotton and
wheat). The most extensive field experiments, conducted under the National Crop Loss
Assessment Network (NCLAN) examined 15 species and numerous cultivars. The NCLAN
results show that "several economically important crop species are sensitive to ozone levels
typical of those found in the Unites States."232 In addition, economic studies have shown
reduced economic benefits as a result of predicted reductions in crop yields associated with
observed ozone levels.233'234'235
Urban ornamentals represent an additional vegetation category likely to experience some
degree of negative effects associated with exposure to ambient ozone levels and likely to impact
large economic sectors. It is estimated that more than $20 billion (1990 dollars) are spent
annually on landscaping using ornamentals, both by private property owners/tenants and by
governmental units responsible for public areas.236 This is therefore a potentially costly
environmental effect. However, in the absence of adequate exposure-response functions and
economic damage functions for the potential range of effects relevant to these types of
vegetation, no direct quantitative analysis has been conducted. Methods are not available to
allow for plausible estimates of the percentage of these expenditures that may be related to
impacts associated with ozone exposure.
3-98
-------
3.4 Particulate Matter
In this section we review the health and welfare effects of paniculate matter (PM). We
also describe air quality monitoring and modeling data that indicate many areas across the
country continue to be exposed to levels of ambient PM above the NAAQS. Emissions of PM
and VOC from the vehicles subject to this rule contribute to these PM concentrations.
Information on air quality was gathered from a variety of sources, including monitored PM
concentrations, air quality modeling done for recent EPA rulemakings and other state and local
air quality information.
3.4.1 Science of PM Formation
Particulate matter (PM) represents a broad class of chemically and physically diverse
substances. It can be principally characterized as discrete particles that exist in the condensed
(liquid or solid) phase spanning several orders of magnitude in size. PM is further described by
breaking it down into size fractions. PMio refers to particles generally less than or equal to 10
micrometers (|im) in diameter. PM2 5 refers to fine particles, those particles generally less than
or equal to 2.5 jim in diameter. Inhalable (or "thoracic") coarse particles refer to those particles
generally greater than 2.5 jim but less than or equal to 10 jim in diameter. Ultrafine PM refers to
particles with diameters generally less than 100 nanometers (0.1 jim). Larger particles (>10 |im)
tend to be removed by the respiratory clearance mechanisms, whereas smaller particles are
deposited deeper in the lungs.
Fine particles are produced primarily by combustion processes and by transformations of
gaseous emissions (e.g., SOx, NOx and VOCs) in the atmosphere. The chemical and physical
properties of PM2.5 may vary greatly with time, region, meteorology and source category. Thus,
PM2.s, may include a complex mixture of different pollutants including sulfates, nitrates, organic
compounds, elemental carbon and metal compounds. These particles can remain in the
atmosphere for days to weeks and travel through the atmosphere hundreds to thousands of
kilometers.
The vehicles that will be covered by the standards contribute to ambient PM levels
through primary (direct) and secondary (indirect) PM. Primary PM is directly emitted into the
air, and secondary PM forms in the atmosphere from gases emitted by fuel combustion and other
sources. Along with primary PM, the vehicles controlled in this action emit VOC, which react in
the atmosphere to form secondary PM2.5, namely organic carbonaceous PM2.5. The gas cans that
will be covered by the standards also emit VOC which contribute to secondary PM2.s. Both
types of directly and indirectly formed particles from vehicles and gas cans are found principally
in the fine fraction.
EPA has recently amended the PM NAAQS (71 FR 61144, October 17, 2006). The final
rule, signed on September 21, 2006 and published on October 17, 2006, addressed revisions to
the primary and secondary NAAQS for PM to provide increased protection of public health and
welfare, respectively. The primary PM2.5 NAAQS include a short-term (24-hour) and a long-
term (annual) standard. The level of the 24-hour PM2 5 NAAQS has been revised from 65ug/m3
to 35 ug/m3 to provide increased protection against health effects associated with short-term
3-99
-------
exposures to fine particles. The current form of the 24-hour PM2.5 standard was retained (e.g.,
based on the 98th percentile concentration averaged over three years). The level of the annual
PM2.5 NAAQS was retained at 15ug/m3' continuing protection against health effects associated
with long-term exposures. The current form of the annual PM2.5 standard was retained as an
annual arithmetic mean averaged over three years, however, the following two aspects of the
spatial averaging criteria were narrowed: (1) the annual mean concentration at each site shall be
within 10 percent of the spatially averaged annual mean, and (2) the daily values for each
monitoring site pair shall yield a correlation coefficient of at least 0.9 for each calendar quarter.
With regard to the primary PMio standards, the 24-hour PMio NAAQS was retained at a level of
150 ug/m3 not to be exceeded more than once per year on average over a three-year period.
Given that the available evidence does not suggest an association between long-term exposure to
coarse particles at current ambient levels and health effects, EPA has revoked the annual PMio
standard.
With regard to the secondary PM standards, EPA has revised these standards to be
identical in all respects to the revised primary standards. Specifically, EPA has revised the
current 24-hour PM2.5 secondary standard by making it identical to the revised 24-hour PM2.5
primary standard, retained the annual PM2 5 and 24-hour PMio secondary standards, and revoked
the annual PMio secondary standards. This suite of secondary PM standards is intended to
provide protection against PM-related public welfare effects, including visibility impairment,
effects on vegetation and ecosystems, and material damage and soiling.
3.4.2 Health Effects of Particulate Matter
As stated in the EPA Particulate Matter Air Quality Criteria Document (PMAQCD),
available scientific findings "demonstrate well that human health outcomes are associated with
ambient PM."J We are relying primarily on the data and conclusions in the PM AQCD and PM
staff paper, which reflects EPA's analysis of policy-relevant science from the PM AQCD,
OT7 01Ł
regarding the health effects associated with particulate matter. ' We also present additional
recent studiesk published after the cut-off date for the PM AQCD.239 Taken together this
information supports the conclusion that PM-related emissions such as those controlled in this
action are associated with adverse health effects.
3.4.2.1 Short-Term Exposure Mortality and Morbidity Studies
As discussed in the PM AQCD, short-term exposure to PM2.5 is associated with
premature mortality from cardiopulmonary diseases (PM AQCD, p. 8-305), hospitalization and
J Personal exposure includes contributions from many different types of particles, from many sources, and in many
different environments. Total personal exposure to PM includes both ambient and nonambient components; and
both components may contribute to adverse health effects.
k These additional studies are included in the 2006 Provisional Assessment of Recent Studies on Health Effects of
Particulate Matter Exposure. The provisional assessment did not and could not (given a very short timeframe)
undergo the extensive critical review by EPA, CASAC, and the public, as did the PM AQCD. The provisional
assessment found that the "new" studies expand the scientific information and provide important insights on the
relationship between PM exposure and health effects of PM. The provisional assessment also found that the "new"
studies generally strengthen the evidence that acute and chronic exposure to fine particles and acute exposure to
thoracic coarse particles are associated with health effects.
3-100
-------
emergency department visits for cardiopulmonary diseases (PMAQCD, p. 9-93), increased
respiratory symptoms (PM AQCD, p. 9-46), decreased lung function (PM AQCD Table 8-34)
and physiological changes or biomarkers for cardiac changes (PM AQCD, Section 8.3.1.3.4). In
addition, the PM AQCD describes a limited body of new evidence from epidemiologic studies
for potential relationships between short-term exposure to PM and health endpoints such as low
birth weight, preterm birth, and neonatal and infant mortality (PM AQCD, Section 8.3.4).
Among the studies of effects from short-term exposure to PM2 5, several studies
specifically address the contribution of mobile sources to short-term PM2.5 effects on daily
mortality. These studies indicate that there are statistically significant associations between
mortality and PM related to mobile source emissions (PM AQCD, p.8-85). The analyses
incorporate source apportionment tools into daily mortality studies and are briefly mentioned
here. Analyses incorporating source apportionment by factor analysis with daily time-series
studies of daily death indicated a relationship between mobile source PM25 and mortality.240'241
Another recent study in 14 U.S. cities examined the effect of PMio exposures on daily hospital
admissions for cardiovascular disease. They found that the effect of PMi0 was significantly
greater in areas with a larger proportion of PMio coming from motor vehicles, indicating that
PMio from these sources may have a greater effect on the toxicity of ambient PMio when
compared with other sources.242 These studies provide evidence that PM-related emissions,
specifically from mobile sources, are associated with adverse health effects.
3.4.2.2 Long-Term Exposure Mortality and Morbidity Studies
Long-term exposure to elevated ambient PM2 5 is associated with mortality from
cardiopulmonary diseases and lung cancer (PM AQCD, p. 8-307), and effects on the respiratory
system such as decreased lung function or the development of chronic respiratory disease (PM
AQCD, pp. 8-313, 8-314). Of specific importance to this rule, the PM AQCD also notes that the
PM components of gasoline and diesel engine exhaust represent one class of hypothesized likely
important contributors to observed ambient PM-related increases in lung cancer incidence and
mortality (PM AQCD, p. 8-318).
The PM AQCD and PM Staff Paper emphasize the results of two long-term studies, the
Six Cities and American Cancer Society (ACS) prospective cohort studies, based on several
factors - the inclusion of measured PM data, the fact that the study populations were similar to
the general population, and the fact that these studies have undergone extensive reanalysis (PM
AQCD, p. 8-306, Staff Paper, p.3-18).243'244'245 These studies indicate that there are significant
associations for all-cause, cardiopulmonary, and lung cancer mortality with long-term exposure
to PM2.s. A variety of studies have been published since the completion of the AQCD. One such
study, which was summarized in EPA's provisional assessment, was an analysis of a subset of
the ACS cohort data, which was published after the PM AQCD was finalized but in time for the
2006 Provisional Assessment, found a larger association than had previously been reported
between long-term PM2 5 exposure and mortality in the Los Angeles area using a new exposure
estimation method that accounted for variations in concentration within the city.246 EPA is
assessing the significance of this study within the context of the broader literature.
As discussed in the PM AQCD, the morbidity studies that combine the features of cross-
3-101
-------
sectional and cohort studies provide the best evidence for chronic exposure effects. Long-term
studies evaluating the effect of ambient PM on children's development have shown some
evidence indicating effects of PM2.5 and/or PMio on reduced lung function growth (PM AQCD,
Section 8.3.3.2.3). A variety of studies have been published since the completion of the AQCD.
One such study, which was summarized in EPA's provisional assessment, reported the results of
a cross-sectional study of outdoor PM2 5 and measures of atherosclerosis in the Los Angeles
basin.247 The study found significant associations between ambient residential PM2.5 and carotid
intima-media thickness (CEVIT), an indicator of subclinical atherosclerosis, an underlying factor
in cardiovascular disease. EPA is assessing the significance of this study within the context of
the broader literature.
3.4.2.3 Roadway-Related Pollution Exposure
A recent body of studies reinforces the findings of these PM morbidity and mortality
effects by looking at traffic-related exposures, PM measured along roadways, or time spent in
traffic and adverse health effects. While many of these studies did not measure PM specifically,
they include potential exhaust exposures which include mobile source PM because they employ
indices such as roadway proximity or traffic volumes. One study with specific relevance to
PM2.5 health effects is a study that was done in North Carolina looking at concentrations of PM2.5
inside police cars and corresponding physiological changes in the police personnel driving the
cars. The authors report significant elevations in markers of cardiac risk associated with
9zlS
concentrations of PM2 5 inside police cars on North Carolina state highways. A number of
studies of traffic-related pollution have shown associations between fine particles and adverse
respiratory outcomes in children who live near major roadways.249'250'251 Additional information
on near-roadway health effects is included in Section 3.5 of this RIA.
3.4.3 Current and Projected PM Levels
The emission reductions from this rule will assist PM nonattainment areas in reaching the
standard by each area's respective attainment date and assist PM maintenance areas in
maintaining the PM standards in the future. In this section we present information on current
and future attainment of the PM standards.
3.4.3.1 Current PM2.s Levels
A nonattainment area is defined in the Clean Air Act (CAA) as an area that is violating
an ambient standard or is contributing to a nearby area that is violating the standard. In 2005,
EPA designated 39 nonattainment areas for the 1997 PM2.5 NAAQS based on air quality design
values (using 2001-2003 or 2002-2004 measurements) and a number of other factors.1 (70 FR
943, January 5, 2005; 70 FR 19844, April 14, 2005). These areas are comprised of 208 full or
partial counties with a total population exceeding 88 million. The 1997 PM2.5 nonattainment
areas and populations, as of October 2006, are listed in Appendix 3C to this RIA. As mentioned
in Section 3.4.1, the 1997 PM2.5 NAAQS was recently revised and the 2006 PM2.5 NAAQS
became effective on December 18, 2006. Nonattainment areas will be designated with respect to
the 2006 PM2.s NAAQS in early 2010. Table 3.4-1 presents the number of counties in areas
1 The full details involved in calculating a PM2 5 design value are given in Appendix N of 40 CFR Part 50.
3-102
-------
currently designated as nonattainment for the 1997 PM2.5 NAAQS as well as the number of
additional counties which have monitored data that is violating the 2006 PM2 5 NAAQS.
Table 3.4-1. PM2.s Standards: Current Nonattainment Areas and Other Violating Counties
1997PM2.5 Standards:
2006 PM2.5 Standards:
39 areas currently designated
Counties with violating monitors2
Total
Number of
Counties
208
49
257
Population1
88,394,000
18,198,676
106,592,676
1) Population numbers are from 2000 census data.
2) This table provides an estimate of the counties violating the 2006 PM25 NAAQS based on 2003-05 air quality
data. The areas designated as nonattainment for the 2006 PM2 5 NAAQS will be based on 3 years of air quality data
from later years. Also, the county numbers in the summary table includes only the counties with monitors violating
the 2006 PM2 5 NAAQS. The monitored county violations may be an underestimate of the number of counties and
populations that will eventually be included in areas with multiple counties designated nonattainment.
States with PM2.5 nonattainment areas will be required to take action to bring those areas
into compliance in the future. Most PM2 5 nonattainment areas will be required to attain the 1997
PM2.5 NAAQS in the 2010 to 2015 time frame and then be required to maintain the 1997 PM2.5
NAAQS thereafter.™ The attainment dates associated with the potential nonattainment areas
based on the 2006 PM2.5 NAAQS would likely be in the 2015 to 2020 timeframe. The emission
standards being finalized in this action will become effective between 2009 and 2015. The
expected PM2 5 and PM2 5 precursor inventory reductions from the standards finalized in this
action will be useful to states in attaining or maintaining the PM2.5 NAAQS.
3.4.3.2 Current PMio Levels
EPA designated PMio nonattainment areas in 1990." As of October 2006, approximately
28 million people live in the 46 areas that are designated as PMio nonattainment, for either
failing to meet the PMio NAAQS or for contributing to poor air quality in a nearby area. There
are 46 full or partial counties that make up the PMio nonattainment areas. The PMio
nonattainment areas and populations are listed in Appendix 3C to this RIA.
As mentioned in Section 3.4.1, the 1997 PM NAAQS was recently revised and the 2006
PM NAAQS became effective on December 18, 2006. The annual PMio NAAQS was revoked
and the 24 hour PMio NAAQS was not changed. The projected reductions in emissions from the
controls finalized in this action will be useful to states to maintain the PMio NAAQS.
m The EPA finalized PM2 5 attainment and nonattainment areas in April 2005. The EPA proposed the PM
Implementation rule in November 2005 (70 FR 65984).
n A PM10 design value is the concentration that determines whether a monitoring site meets the NAAQS for PM10.
The full details involved in calculating a PM10 design value are given in Appendices H and I of 40 CFR Part 50.
3-103
-------
3.4.3.3 Projected PM2.5 Levels
Recent air quality modeling predicts that without additional controls there will continue
to be a need for reductions in PM concentrations in the future. In the following sections we
describe the recent PM air quality modeling and results of the modeling.
3.4.3.3.1 PM Modeling Methodology
Recently PM air quality analyses were performed for the PM NAAQS final rule, which
was promulgated by EPA in 2006. The Community Multiscale Air Quality (CMAQ) model was
used as the tool for simulating base and future year concentrations of PM, visibility and
deposition in support of the PM NAAQS air quality assessment. The PM NAAQS analysis
included all final federal rules up to and including Clean Air Interstate Rule (CAIR) and all final
mobile source rule controls as of October 2006. Details on the air quality modeling are provided
in the Regulatory Impact Analysis (RIA) for the Final PM NAAQS Rule, included in the docket
for this final rule.252
3.4.3.3.2 Areas at Risk of Future PM2.5 Violations
Air quality modeling performed for the final PM NAAQS indicates that in the absence of
additional local, regional or national controls, there will likely continue to be counties that will
not attain some combination of the annual 2006 PM2.5 standard (15 |ig/m3) and the daily 2006
PM2.s standard (35 |ig/m3). The PM NAAQS analysis provides estimates of future PM2.s levels
across the country. For example, in 2015 based on emission controls currently adopted or
expected to be in place0, we project that 53 million people will live in 52 counties with projected
PM2.5 design values at and above the 2006 standard, see Table 3.4-2.p The rule will assist these
counties in attaining the PM2.5NAAQS. Table 3.4-2 also lists the 54 counties, where 27 million
people are projected to live, with 2015 projected design values that do not violate the PM2.5
NAAQS but are within ten percent of it. The rule may help ensure that these counties continue
to maintain their attainment status.
Table 3.4-2. Counties with 2015 Projected Annual and Daily PM2.s Design Values
Above and within 10% of the 2006 PM2.5 Standard"
2015
Projected
Annual
PM2.5
Design
Value
State County (M9/m3)
2015
Projected
Daily
PM2.5
Design
Value 2015
(ug/m3) Population15
0 Counties forecast to remain in nonattainment may need to adopt additional local or regional controls to attain the
standards by dates set pursuant to the Clean Air Act. The emissions reductions associated with this rule will help
these areas attain the PM standards by their statutory date.
p Note that this analysis identifies only counties projected to have a violating monitor; the number of counties to be
designated and the associated population would likely exceed these estimates.
3-104
-------
Alabama
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
California
Connecticut
Georgia
Georgia
Georgia
Georgia
Georgia
Georgia
Georgia
Idaho
Idaho
Idaho
Idaho
Idaho
Illinois
Illinois
Illinois
Illinois
Indiana
Indiana
Jefferson Co
Alameda Co
Butte Co
Colusa Co
Contra Costa Co
Fresno Co
Imperial Co
Inyo Co
Kern Co
Kings Co
Los Angeles Co
Merced Co
Orange Co
Placer Co
Riverside Co
Sacramento Co
San Bernardino
Co
San Diego Co
San Francisco Co
San Joaquin Co
San Luis Obispo
Co
San Mateo Co
Santa Clara Co
Solano Co
Sonoma Co
Stanislaus Co
Sutter Co
Tulare Co
Ventura Co
Yolo Co
Fairfield Co
Bibb Co
Clayton Co
DeKalb Co
Floyd Co
Fulton Co
Muscogee Co
Wilkinson Co
Ada Co
Bannock Co
Canyon Co
Power Co
Shoshone Co
Cook Co
Madison Co
St. Clair Co
Will Co
Clark Co
Lake Co
15.9
13.3
13.4
9.5
12.6
20.1
14.8
6.1
21.3
17.2
23.7
15.8
20.0
11.4
27.8
12.2
24.6
15.8
11.3
15.4
9.4
10.5
10.7
11.7
10.0
16.6
11.2
21.2
14.1
10.2
11.0
13.7
13.9
13.6
14.0
15.5
13.4
13.6
8.9
9.1
9.2
10.5
12.4
15.5
15.2
14.6
13.2
13.6
13.4
36.9
59.4
50.7
33.5
61.3
73.0
45.7
38.1
81.4
70.6
62.2
54.4
41.1
38.1
73.5
49.8
65.7
40.7
52.5
51.1
35.8
41.9
48.5
57.7
38.9
61.9
39.3
77.2
38.8
33.0
31.6
27.0
28.7
31.5
30.9
32.2
34.2
29.3
32.2
40.2
32.6
36.6
36.2
37.1
35.5
30.4
32.0
31.1
40.8
669,850
1,628,698
242,166
23,066
1,155,323
960,934
173,482
19,349
804,940
161,607
9,910,805
250,152
3,467,120
403,624
2,015,955
1,488,456
2,157,926
3,489,368
765,846
675,362
304,079
785,949
1,899,727
529,784
569,486
547,041
99,716
441,185
923,205
206,388
893,629
160,468
280,476
715,947
97,674
877,365
197,634
11,259
397,456
88,033
154,137
8,932
15,646
5,362,931
271,854
251,612
634,068
112,523
490,795
3-105
-------
Indiana
Kentucky
Maryland
Maryland
Maryland
Massachusetts
Michigan
Michigan
Michigan
Michigan
Michigan
Montana
Montana
New Jersey
New Jersey
New Jersey
New York
New York
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Ohio
Oregon
Oregon
Oregon
Oregon
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Pennsylvania
Tennessee
Utah
Utah
Utah
Utah
Utah
Washington
Washington
Washington
Washington
Washington
Marion Co
Jefferson Co
Anne Arundel Co
Baltimore city
Baltimore Co
Hampden Co
Kalamazoo Co
Kent Co
Oakland Co
St. Clair Co
Wayne Co
Lincoln Co
Missoula Co
Camden Co
Hudson Co
Union Co
Bronx Co
New York Co
Cuyahoga Co
Franklin Co
Hamilton Co
Jefferson Co
Lucas Co
Scioto Co
Trumbull Co
Jackson Co
Klamath Co
Lane Co
Washington Co
Allegheny Co
Beaver Co
Berks Co
Dauphin Co
Lancaster Co
Lehigh Co
Mercer Co
Northampton Co
Philadelphia Co
York Co
Knox Co
Box Elder Co
Cache Co
Salt Lake Co
Utah Co
Weber Co
Clark Co
King Co
Pierce Co
Snohomish Co
Thurston Co
13.5
13.8
11.1
13.0
11.3
11.6
12.8
12.0
13.0
12.5
17.4
15.0
10.6
11.1
12.0
12.2
12.8
14.0
15.4
13.7
14.3
14.2
12.5
15.6
12.1
10.9
10.1
12.9
9.0
16.5
12.1
12.0
11.0
12.2
10.5
11.0
10.9
13.3
12.3
13.6
8.6
12.5
12.6
9.3
9.1
9.2
10.8
11.1
11.3
8.9
33.1
33.4
33.2
35.5
32.6
32.9
32.7
31.9
33.2
32.5
39.0
42.4
32.1
32.1
32.8
32.8
33.2
33.2
40.0
33.5
34.2
34.2
32.2
34.3
34.2
37.6
39.1
53.6
32.0
53.4
33.2
35.5
33.3
33.7
34.7
31.6
35.0
35.2
35.9
29.6
39.0
51.9
49.3
36.7
36.2
34.3
34.0
43.0
40.1
34.9
889,645
710,231
574,322
596,076
810,172
452,055
257,817
654,449
1,355,670
185,970
1,921,253
19,875
118,303
512,135
604,036
525,096
1,283,316
1,551,641
1,325,507
1,181,578
841,858
68,909
443,230
81,013
227,546
250,169
69,423
387,237
639,839
1,245,917
184,648
396,410
272,748
535,622
328,523
123,577
286,838
1,372,037
417,408
448,931
49,878
114,729
1,133,410
508,106
229,807
479,002
2,013,808
879,363
782,319
264,364
3-106
-------
Washington Yakima Co
West Virginia Berkeley Co
West Virginia Hancock Co
West Virginia Kanawha Co
Wisconsin Milwaukee Co
Wisconsin Waukesha Co
Wyoming Sheridan Co
Number of Violating Counties
Population of Violating Counties
Number of Counties within 10%
Population of Counties within 10%
9.6
12.0
13.4
13.9
12.1
11.8
10.5
34.9
32.7
32.7
28.9
32.1
32.4
31.8
52
54
261,452
99,349
30,857
196,498
908,336
441,482
28,623
53,468,515
26,896,926
a) Bolded concentrations indicate levels above the PM2 5 standard.
b) Populations are based on 2000 census projections.
3.4.4 Environmental Effects of PM Pollution
In this section we discuss public welfare effects of PM and its precursors including
visibility impairment, atmospheric deposition, and materials damage and soiling.
3.4.4.1 Visibility Impairment
Visibility can be defined as the degree to which the atmosphere is transparent to visible
light.253 Visibility impairment manifests in two principal ways: as local visibility impairment
and as regional haze.q Local visibility impairment may take the form of a localized plume, a
band or layer of discoloration appearing well above the terrain as a result from complex local
meteorological conditions. Alternatively, local visibility impairment may manifest as an urban
haze, sometimes referred to as a "brown cloud." This urban haze is largely caused by emissions
from multiple sources in the urban areas and is not typically attributable to only one nearby
source or to long-range transport. The second type of visibility impairment, regional haze,
usually results from multiple pollution sources spread over a large geographic region. Regional
haze can impair visibility over large regions and across states.
Visibility is important because it has direct significance to people's enjoyment of daily
activities in all parts of the country. Individuals value good visibility for the well-being it
provides them directly, where they live and work, and in places where they enjoy recreational
opportunities. Visibility is also highly valued in significant natural areas such as national parks
and wilderness areas, and special emphasis is given to protecting visibility in these areas. For
more information on visibility see the 2004 PMAQCD as well as the 2005 PM Staff Paper.254'255
Fine particles are the major cause of reduced visibility in parts of the United States. To
address the welfare effects of PM on visibility, EPA set secondary PM2.5 standards which would
q See discussion in U.S. EPA, National Ambient Air Quality Standards for Paniculate Matter; Proposed Rule;
January 17, 2006, Vol71 p 2676. This information is available electronically at http://epa.gov/fedrgstr/EP A-
AIR/2006/Januarv/DaY-17/al77.pdf.
3-107
-------
act in conjunction with the establishment of a regional haze program. In setting this secondary
standard, EPA concluded that PM2 5 causes adverse effects on visibility in various locations,
depending on PM concentrations and factors such as chemical composition and average relative
humidity. The secondary (welfare-based) PM2.5 NAAQS was established as equal to the suite of
primary (health-based) NAAQS. Furthermore, Section 169A of the Act provides additional
authority to address existing visibility impairment and prevent future visibility impairment in the
156 national parks, forests and wilderness areas categorized as mandatory class I federal areas
(62 FR 38680-81, July 18, 1997).r In July 1999 the regional haze rule (64 FR 35714) was put in
place to protect the visibility in mandatory class I federal areas. Visibility can be said to be
impaired in both PM2 5 nonattainment areas and mandatory class I federal areas.
Data showing PM2.5 nonattainment areas and visibility levels above background at the
Mandatory Class I Federal Areas demonstrate that visibility impairment is experienced
throughout the U.S., in multi-state regions, urban areas, and remote mandatory Federal class I
areas. The PM and PM precursor emissions from the vehicles and gas cans subject to this
proposed rule contribute to these visibility effects.
3.4.4.1.1 Current Visibility Impairment
The need for reductions in the levels of PM2 5 is widespread. Currently, high ambient
PM2.5 levels are measured throughout the country. Fine particles may remain suspended for days
or weeks and travel hundreds to thousands of kilometers, and thus fine particles emitted or
created in one county may contribute to ambient concentrations in a neighboring region.256
As mentioned above the secondary PM2.5 standards were set as equal to the suite of
primary PM2 5 standards. Recently designated PM2 5 nonattainment areas indicate that almost 90
million people live in 208 counties that are in nonattainment for the 1997 PM2 5 NAAQS, see
Appendix 3C. Thus, at least these populations (plus others who travel to these areas) would
likely be experiencing visibility impairment.
3.4.4.1.2 Current Visibility Impairment at Mandatory Class I Federal Areas
Detailed information about current and historical visibility conditions in mandatory class
I federal areas is summarized in the EPA Report to Congress and the 2002 EPA Trends
Report.257'258 The conclusions draw upon the Interagency Monitoring of Protected Visual
Environments (IMPROVE) network data. One of the objectives of the IMPROVE monitoring
network program is to provide regional haze monitoring representing all mandatory class I
federal areas where practical. The National Park Service report also describes the state of
national park visibility conditions and discusses the need for improvement.259
The regional haze rule requires states to establish goals for each affected mandatory class
I federal area to improve visibility on the haziest days (20% most impaired days) and ensure no
degradation occurs on the cleanest days (20% least impaired days). Although there have been
r These areas are defined in Section 162 of the Act as those national parks exceeding 6,000 acres, wilderness
areas and memorial parks exceeding 5,000 acres, and all international parks which were in existence on August 7,
1977.
3-108
-------
general trends toward improved visibility, progress is still needed on the haziest days.
Specifically, as discussed in the 2002 EPA Trends Report, without the effects of pollution a
natural visual range in the United States is approximately 75 to 150 km in the East and 200 to
300 km in the West. In 2001, the mean visual range for the worst days was 29 km in the East
and 98 km in the West. 26°
3.4.4.1.3 Future Visibility Impairment
Recent modeling for the final PM NAAQS rule was used to project PM2.5 levels in the
U.S. in 2015. The results suggest that PM2.5 levels above the 2006 NAAQS will persist in the
future. We predicted that in 2015, there will be 52 counties with a population of 53 million
where annual PM25 levels will exceed the 2006 PM25 NAAQS, see Table 3.4-1. Thus, in the
future, a percentage of the population may continue to experience visibility impairment in areas
where they live, work and recreate.
The PM and PM precursor emissions from the vehicles and gas cans subject to the
proposed controls contribute to visibility impairment. These emissions occur in and around areas
with PM2.s levels above the annual 1997 PM2 5 NAAQS. Thus, the emissions from these sources
contribute to the current and anticipated visibility impairment and the emission reductions
finalized here may help improve future visibility impairment.
3.4.4.1.4 Future Visibility Impairment at Mandatory Class I Federal Areas
Achieving the PM2 5 NAAQS will help improve visibility across the country, but it will
not be sufficient to meet the statutory goal of no manmade impairment in the mandatory class I
federal areas (64 FR 35714, July 1, 1999 and 62 FR 38652, July 18, 1997). In setting the
NAAQS, EPA discussed how the NAAQS in combination with the regional haze program, is
deemed to improve visibility consistent with the goals of the Act. In the East, there are and will
continue to be areas with PM2.5 concentrations above the PM2.5 NAAQS and where light
extinction is significantly above natural background. Thus, large areas of the Eastern United
States have air pollution that is causing and will continue to cause visibility impairment. In the
West, scenic vistas are especially important to public welfare. Although the PM2.5 NAAQS is
met in most areas outside of California, virtually the entire West is in close proximity to a scenic
mandatory class I federal area protected by 169A and 169B of the CAA.
Recent modeling for CAIR was also used to project visibility conditions in mandatory
class I federal areas across the country in 2015. The results for the mandatory class I federal
areas suggest that these areas are predicted to continue to have visibility impairment above
background on the 20% worst days in the future.
The overall goal of the regional haze program is to prevent future visibility impairment
and remedy existing visibility impairment in mandatory class I federal areas. As shown by the
future visibility estimates in Appendix 3D it is projected that there will continue to be mandatory
class I federal areas with visibility levels above background in 2015.261 Additional emission
reductions will be needed from the broad set of sources that contribute, including the vehicles
and gas cans subject to this rule. The reductions being finalized in this action are a part of the
3-109
-------
overall strategy to achieve the visibility goals of the Act and the regional haze program.
3.4.4.2 Atmospheric Deposition
Wet and dry deposition of ambient particulate matter delivers a complex mixture of
metals (e.g., mercury, zinc, lead, nickel, aluminum, cadmium), organic compounds (e.g., POM,
dioxins, furans) and inorganic compounds (e.g., nitrate, sulfate) to terrestrial and aquatic
ecosystems. The chemical form of the compounds deposited is impacted by a variety of factors
including ambient conditions (e.g., temperature, humidity, oxidant levels) and the sources of the
material. Chemical and physical transformations of the parti culate compounds occur in the
atmosphere as well as the media onto which they deposit. These transformations in turn
influence the fate, bioavailability and potential toxicity of these compounds. Atmospheric
deposition has been identified as a key component of the environmental and human health hazard
posed by several pollutants including mercury, dioxin and PCBs.262
Adverse impacts on water quality can occur when atmospheric contaminants deposit to
the water surface or when material deposited on the land enters a waterbody through runoff.
Potential impacts of atmospheric deposition to waterbodies include those related to both nutrient
and toxic inputs. Adverse effects to human health and welfare can occur from the addition of
excess particulate nitrate nutrient enrichment which contributes to toxic algae blooms and zones
of depleted oxygen, which can lead to fish kills, frequently in coastal waters. Particles
contaminated with heavy metals or other toxins may lead to the ingestion of contaminated fish,
ingestion of contaminated water, damage to the marine ecology, and limited recreational uses.
Several studies have been conducted in U.S. coastal waters and in the Great Lakes Region in
which the role of ambient PM deposition and runoff is investigated.263'264'265'266'267
Adverse impacts on soil chemistry and plant life have been observed for areas heavily
impacted by atmospheric deposition of nutrients, metals and acid species, resulting in species
shifts, loss of biodiversity, forest decline and damage to forest productivity. Potential impacts
also include adverse effects to human health through ingestion of contaminated vegetation or
livestock (as in the case for dioxin deposition), reduction in crop yield, and limited use of land
due to contamination.
In the following subsections, atmospheric deposition of heavy metals and particulate
organic material is discussed.
3-110
-------
3.4.4.2.1 Heavy Metals
Heavy metals, including cadmium, copper, lead, chromium, mercury, nickel and zinc,
have the greatest potential for influencing forest growth (PM AQCD, p. 4-87).268 Investigation
of trace metals near roadways and industrial facilities indicate that a substantial burden of heavy
metals can accumulate on vegetative surfaces. Copper, zinc, and nickel have been documented
to cause direct toxicity to vegetation under field conditions (PM AQCD, p. 4-75). Little research
has been conducted on the effects associated with mixtures of contaminants found in ambient
PM. While metals typically exhibit low solubility, limiting their bioavailability and direct
toxicity, chemical transformations of metal compounds occur in the environment, particularly in
the presence of acidic or other oxidizing species. These chemical changes influence the mobility
and toxicity of metals in the environment. Once taken up into plant tissue, a metal compound can
undergo chemical changes, accumulate and be passed along to herbivores or can re-enter the soil
and further cycle in the environment.
Although there has been no direct evidence of a physiological association between tree
injury and heavy metal exposures, heavy metals have been implicated because of similarities
between metal deposition patterns and forest decline (PM AQCD, p. 4-76).269 Contamination of
plant leaves by heavy metals can lead to elevated soil levels. Some trace metals absorbed into
the plant and can bind to the leaf tissue (PM AQCD, p. 4-75). When these leaves fall and
decompose, the heavy metals are transferred into the soil.270'271
The environmental sources and cycling of mercury are currently of particular concern due
to the bioaccumulation and biomagnification of this metal in aquatic ecosystems and the potent
toxic nature of mercury in the forms in which is it ingested by people and other animals.
Mercury is unusual compared with other metals in that it largely partitions into the gas phase (in
elemental form), and therefore has a longer residence time in the atmosphere than a metal found
predominantly in the particle phase. This property enables a portion of emitted mercury to travel
far from the primary source before being deposited and accumulating in the aquatic ecosystem.
Localized or regional impacts are also observed for mercury emitted from combustion sources.
The major source of mercury in the Great Lakes is from atmospheric deposition, accounting for
r)Hr) T71
approximately eighty percent of the mercury in Lake Michigan. ' Over fifty percent of the
mercury in the Chesapeake Bay has been attributed to atmospheric deposition.274 Overall, the
National Science and Technology Council (NSTC, 1999) identifies atmospheric deposition as the
primary source of mercury to aquatic systems. Forty-four states have issued health advisories for
the consumption offish contaminated by mercury; however, most of these advisories are issued
in areas without a mercury point source.
Elevated levels of zinc and lead have been identified in streambed sediments, and these
elevated levels have been correlated with population density and motor vehicle use.275'276 Zinc
and nickel have also been identified in urban water and soils. In addition, platinum, palladium,
and rhodium, metals found in the catalysts of modern motor vehicles, have been measured at
elevated levels along roadsides.277 Plant uptake of platinum has been observed at these
locations.
3-111
-------
3.4.4.2.2 Poly cyclic Organic Matter
Poly cyclic organic matter (POM) is a byproduct of incomplete combustion and consists
of organic compounds with more than one benzene ring and a boiling point greater than or equal
to 100 degrees centigrade.278 Polycyclic aromatic hydrocarbons (PAHs) are a class of POM that
contains compounds which are known or suspected carcinogens.
Major sources of PAHs include mobile sources. PAHs in the environment may be
present as a gas or adsorbed onto airborne parti culate matter. Since the majority of PAHs are
adsorbed onto particles less than 1.0 jam in diameter, long range transport is possible. However,
studies have shown that PAH compounds adsorbed onto diesel exhaust particulate and exposed
to ozone have half lives of 0.5 to 1.0 hours.279
Since PAHs are insoluble, the compounds generally are particle reactive and accumulate
in sediments. Atmospheric deposition of particles is believed to be the major source of PAHs to
the sediments of Lake Michigan.280'281 Analyses of PAH deposition to Chesapeake and
Galveston Bay indicate that dry deposition and gas exchange from the atmosphere to the surface
water predominate.282'283 Sediment concentrations of PAHs are high enough in some segments
of Tampa Bay to pose an environmental health threat. EPA funded a study to better characterize
the sources and loading rates for PAHs into Tampa Bay.284 PAHs that enter a waterbody
through gas exchange likely partition into organic rich particles and be biologically recycled,
while dry deposition of aerosols containing PAHs tends to be more resistant to biological
recycling.285 Thus, dry deposition is likely the main pathway for PAH concentrations in
sediments while gas/water exchange at the surface may lead to PAH distribution into the food
web, leading to increased health risk concerns.
Trends in PAH deposition levels are difficult to discern because of highly variable
ambient air concentrations, lack of consistency in monitoring methods, and the significant
influence of local sources on deposition levels.286 Van Metre et al. (2000) noted PAH
concentrations in urban reservoir sediments have increased by 200-300% over the last forty years
and correlates with increases in automobile use.287
Cousins et al. (1999) estimates that greater than ninety percent of semi-volatile organic
compound (SVOC) emissions in the United Kingdom deposit on soil.288 An analysis of
polycyclic aromatic hydrocarbon (PAH) concentrations near a Czechoslovakian roadway
indicated that concentrations were thirty times greater than background.289
3.4.4.3 Materials Damage and Soiling
The deposition of airborne particles can also reduce the aesthetic appeal of buildings and
culturally important articles through soiling, and can contribute directly (or in conjunction with
other pollutants) to structural damage by means of corrosion or erosion.290 Particles affect
materials principally by promoting and accelerating the corrosion of metals, by degrading paints,
and by deteriorating building materials such as concrete and limestone. Particles contribute to
these effects because of their electrolytic, hygroscopic, and acidic properties, and their ability to
3-112
-------
sorb corrosive gases (principally sulfur dioxide). The rate of metal corrosion depends on a
number of factors, including the deposition rate and nature of the pollutant; the influence of the
metal protective corrosion film; the amount of moisture present; variability in the
electrochemical reactions; the presence and concentration of other surface electrolytes; and the
orientation of the metal surface.
3.5 Health and Welfare Impacts of Near-Roadway Exposure
Over the years there have been a large number of studies that have examined associations
between living near major roads and different adverse health endpoints. These studies generally
examine people living near heavily-trafficked roadways, typically within several hundred meters,
where fresh emissions from motor vehicles are not yet fully diluted with background air.
As discussed in Chapter 3.1.3, many studies have measured elevated concentrations of
pollutants emitted directly by motor vehicles near large roadways, as compared to overall urban
background levels. These elevated concentrations generally occur within approximately 200
meters of the road, although the distance may vary depending on traffic and environmental
conditions. Pollutants measured with elevated concentrations include benzene, polycyclic
aromatic hydrocarbons, carbon monoxide, nitrogen dioxide, black carbon, and coarse, fine, and
ultrafme particles. In addition, resuspended road dust, and wear particles from tire and brake use
also show concentration increases in proximity of major roadways.
As noted in section 3.2, HAPEM6 estimates the changes in time-weighted exposures
associated with proximity to roadways for individual pollutants. The studies discussed in this
section address exposures and health effects that are at least partially captured by our modeling,
but there may be additional exposures and health effects associated with pollutants, singly or in
combination, that are not explicitly quantified. However, because the studies discussed in this
section often employ exposure estimation metrics associated with multiple pollutants, exposure-
response information from these studies may not be suitable for risk assessment geared around
one or several chemicals.
At this point, there exists no exposure metric specific to "traffic," although as noted
above, a wide variety of gaseous, particulate, and semi-volatile species are elevated near
roadways. As a result, the exposure metrics employed generally indicate the presence and/or
intensity of a mixture of air pollutants for exposure assessment. Many of the health studies
discussed below employ non-specific exposure metrics, including traffic on roads nearest home
or school, distance to the nearest road, measured outdoor nitrogen dioxide concentrations, air
quality dispersion modeling of specific traffic-generated chemicals, and exposure assignment
based on land use. These exposure metrics represent the mixture of traffic-generated pollutants,
rather than individual pollutants. Accordingly, such results are not directly comparable with
community epidemiology studies that employ ambient measurements of particulate matter or
ozone over a fixed time period, or to toxicological studies employing a single pollutant to
evaluate responses in humans or animals.
A wide range of health effects are reported in the literature related to near roadway and
in-vehicle exposures. This is not unexpected, given the chemical and physical complexity of the
3-113
-------
mixture to which people are exposed in this environment. These effects overlap with those
identified in our discussion of the effects of PM and ozone. The discussion below addresses the
studies in detail. However, in general terms, the near-roadway health studies provide stronger
evidence for some health endpoints than others. Epidemiologic evidence of adverse responses to
traffic-related pollution is strongest for non-allergic respiratory symptoms, and several well-
conducted epidemiologic studies have shown associations with cardiovascular effects, premature
adult mortality, and adverse birth outcomes, including low birth weight and size. Traffic-related
pollutants have been repeatedly associated with increased prevalence of asthma-related
respiratory symptoms in children, although epidemiologic evidence remains inconclusive for a
hypothesized link between traffic and the development of allergies and new onset asthma.
For childhood cancer, in particular childhood leukemia, epidemiologic studies have
shown less ability to detect the risks predicted from toxicological studies. Several small studies
report positive associations, though such effects have not been observed in two larger studies. As
described above in Chapter 1.3, benzene and 1,3-butadiene are both known human leukemogens
in adults from occupational exposures. As previously mentioned, epidemiologic studies have
shown an increased risk of leukemia among children whose parents have been occupationally
exposed to benzene. While epidemiologic studies of near-roadway exposures have not always
shown a statistically significant association with childhood leukemias, the results are consistent
with the risks predicted from the studies at higher exposure levels. As a whole the toxicology
and epidemiology are consistent with a potentially serious children's health concern and
additional research is needed.
Significant scientific uncertainties remain in research on health effects near roads,
including the exposures of greatest concern, the importance of chronic versus acute exposures,
the role of fuel type (e.g. diesel or gasoline) and composition (e.g., percent aromatics), and
relevant traffic patterns. Furthermore, in these studies, it is often difficult to understand the role
of co-stressors including noise and socioeconomic status (e.g., access to health care, nutritional
status), and the role of differential susceptibility.
3.5.1 Mortality
The quantifiable effects of this rule on premature mortality associated with exposure to
PM2.5 are assessed as part of the benefits estimates for this rule. In addition to studies that have
documented the relationship between ambient PM and premature mortality, a few recent studies
have investigated the relationship between premature mortality and broader indicators of
transportation emissions, such as residence near traffic. The extent to which these studies are
detecting any additional effects not accounted for in the ambient PM-premature mortality
relationship is unclear.
Living near major roads has been investigated in both long-term and short-term mortality
studies. Long-term studies track subjects over time and investigate the mortality rates among
groups with different levels of exposure to ambient pollutants. Short term studies employ daily
variation in ambient concentrations to estimate the daily deaths attributable to air pollution.
3-114
-------
A total of three cohort studies have examined premature mortality in relation to residence
near traffic, another examined county-level traffic density, while one other has examined stroke
mortality. In addition, one study accounted for the effect of residence along a major road on
associations with daily deaths in a time-series study. These studies constitute all of the studies
examining mortality with reference to proximity to traffic.
Premature mortality in adults in association with living near high-traffic roadways has
been studied in three recent cohort studies for all-cause and cardiopulmonary mortality from the
Netherlands, Ontario, Canada, and most recently, Germany.291'292'293 Canadian vehicles and
emission standards largely mirror the U.S. vehicle fleet. Both studies defined living near a major
road as having a residence within 100 meters of a highway or within 50 meters of a major urban
roadway. In the first study, involving approximately 5,000 people over 55 years old living
throughout the Netherlands, residence near major roadways was associated with a 41% increase
in the mortality rate from all causes and a 95% increase in the cardiopulmonary mortality rate.294
The second study involved over 5,200 subjects aged 40 years or more, all living in the
Hamilton, Ontario area. This study examined total mortality, finding a statistically significant
18% increase associated with living near a major roadway. No difference in response was found
among those with pre-existing respiratory illness. The study also calculated "rate advancement
periods," which describe the effect of an exposure in terms of the time period by which exposed
persons reach prematurely the same disease risk as unexposed persons reach later on. The rate
advancement period for total mortality was 2.5 years. The rate advancement periods were also
calculated for other risk factors for mortality, including chronic pulmonary disease excluding
asthma (3.4 years), chronic ischemic heart disease (3.1 years), and diabetes mellitus (4.4 years).
A subsequent follow-up study found elevated mortality rates from circulatory causes in the
Canadian study population.
Most recently, German investigators followed up a series of cross-sectional studies on
women age 50-59 living in the North Rhine-Westphalia region during the late 1980's and
1990's, tracking vital status and migration to the years 2002-2003.295 In total, the cohort
consisted of approximately 4800 women. Exposures were categorized using ambient NO2 and
PMio (estimated from TSP), and an indicator of residence within 50 m of a "major road", defined
at > 10,000 cars/day. Overall, living within 50 meters of a major road was associated with a
significant 70% increase in the rate of cardiopulmonary mortality. Nearest-monitor NO2 and
PMio were also associated with a 57% and 34% increase in the rate of cardiopulmonary
mortality. Exposure to NO2 was also associated with a 17% increase in all-cause mortality.
Despite differences in the vehicle fleets of Europe and Canada, whose emission standards
largely mirror those of the U.S., the results of these studies are similar.
In another study evaluating a cohort of older, hypertensive male U.S. veterans, county-
level traffic index and pollution estimates were employed in estimating exposure to traffic
activity and other air pollutants.296 Area-based traffic density was significantly associated with
increased mortality rates, as were constituents of motor vehicle exhaust, such as elemental
carbon.
3-115
-------
One cohort study conducted in the United Kingdom examined cardiocerebral (stroke)
mortality in relation to living near traffic.297 Those living in census areas near roadways had
significantly higher stroke mortality rates. In a study involving nearly 190,000 stroke deaths in
1990-1992, Maheswaran and Elliott (2002) examined stroke mortality rates in census districts
throughout England and Wales. Census districts closest to major roads showed significant
increases in stroke mortality rates for men and women. Compared to those living in census
districts whose center was greater than 1000 m from a main road, men and women living in
census regions with centers less than 200 m away had stroke mortality rates 7% and 4% higher,
respectively.
One study from the Netherlands used time-series analysis to evaluate the change in the
magnitude of the association between daily concentrations of black smoke, an air metric related
to black carbon, and daily deaths, for populations living along roads with at least 10,000 vehicles
per day.298 Compared with the population living elsewhere, the traffic-exposed population had
significantly higher associations between black smoke and daily mortality.
Although the studies of mortality have employed different study designs and metrics of
exposure, they provide evidence for increased mortality rates in proximity of heavy traffic. In
evaluating the generalizability of these study results, questions remain regarding differences in
housing stock, residential ventilation, vehicle type and fuel differences, personal activity
patterns, and the appropriate exposure metric. Furthermore, in the cohort studies, although
controls for income level were incorporated based on postal code or census area, it is possible
that other unmeasured covariates explain the associations with traffic.
3.5.2 Non-Allergic Respiratory Symptoms
Our analysis of the benefits associated with reduced exposure to PM2.5 includes chronic
bronchitis, hospital admissions for respiratory causes, emergency room visits for asthma, acute
bronchitis, upper and lower respiratory symptoms and exacerbation of asthma. In addition,
studies in Europe, Asia and North America have found increased risk of respiratory symptoms
such as wheeze, cough, chronic phlegm production, and dyspnea (shortness of breath) in children
and adults with increased proximity to roadways and/or associated with local traffic density.
Most of these studies were cross-sectional and relied solely on questionnaire assessments of
health outcomes, in combination with simple exposure indicators. There are a large number of
studies available, but for the sake of brevity, only studies conducted in the United States are
discussed here. European studies reach similar conclusions, as summarized in a recent review of
the European literature.299 The discussion below covers all studies conducted in the United
States. EPA has not formally evaluated the extent to which these studies may be documenting
health effects that are already included in the benefits analysis associated with PM.
Most recently, a study from Cincinnati, OH examined the prevalence of wheezing in a
group of infants less than one year of age.300 Infants with at least one atopic parent qualified for
enrollment. The study compared infants living near stop-and-go truck traffic with others living
near smoothly-flowing truck traffic, and others further from traffic. Infants with wheeze were
significantly more likely to live near stop-and-go traffic than either those living near smoothly-
flowing traffic or those living away from traffic. Truck volume was not associated with wheeze.
3-116
-------
A respiratory health study in the east San Francisco Bay area looked at a series of
community schools upwind and downwind of major roads along a major transportation corridor,
where ambient air quality was monitored.301 Over 1,100 children in grades three through five
attending the schools were assessed for respiratory symptoms and physician's diagnosis of
asthma. Overall, concentrations of traffic-related air pollutants measured at each school were
associated with increased prevalence of bronchitis symptoms and physician confirmed asthma,
both within the last 12 months.
A case-control study in Erie County, NY compared home proximity to traffic among
children admitted into local hospitals for asthma with those admitted for non-respiratory
conditions.302 Overall, children hospitalized for asthma were more likely to live within 200
meters of roads above the 90th percentile of daily vehicle miles traveled, and to have trucks and
trailers passing within 200 meters of their residences. However, hospitalization for asthma was
not associated with residential distance from major state routes.
A study in San Diego County, C A compared the residential location of asthmatic children
with children having a non-respiratory diagnosis within the state Medicaid system.303 Traffic
volumes on streets nearby the home were not associated with the prevalence of asthma.
However, among asthmatic children, high street volumes on the nearest street were associated
with an increased annual frequency of medical visits for asthma.
In the only U.S. study examining adult respiratory symptoms, Massachusetts veterans
were evaluated for traffic-health relationships.304 In the study, living within 50 m of a major
roadway was associated with increased reporting of persistent wheeze. This trend held only for
roads with at least 10,000 vehicles per day. Patients experiencing chronic phlegm were also
more likely to live within 50 meters of roads with at least 10,000 vehicles per day. However,
chronic cough was not associated with living near traffic.
The studies described above employ different exposure metrics and health endpoints,
making evaluation difficult. However, numerous other studies from around the world also
provide evidence for increased prevalence of respiratory symptoms among people living near
major roads. For a detailed listing, refer to the docket of this rule. Taken together, these studies
provide evidence that respiratory symptoms may be associated with living near major roadways,
particularly in children, upon whom the preponderance of studies have focused.
3.5.3 Development of Allergic Disease and Asthma
A significant number of studies have examined evidence of a role of traffic-generated
pollution in the development (e.g. new onset) of atopic illnesses (i.e., hypersensitivity to
allergens), such as asthma, allergic rhinitis, and dermatitis. A critical review of evidence,
primarily generated in European studies, was recently published.305 Overall, the review
concluded that there is some limited evidence of an association between traffic-generated
pollutants and asthma incidence. More recent studies have also found significant associations
between prevalent asthma and living near major roads.306 Toxicological evidence provides some
evidence that particles from diesel engine exhaust may serve as adjuvants to IgE-mediated
3-117
-------
immune responses. EPA's Health Assessment Document for Diesel Engine Exhaust addresses
many of the toxicological studies on diesel exhaust. However, in community epidemiology
studies, the evidence remains tentative. The potential for these effects is not taken into account
in the benefits analysis for PM because EPA's various scientific advisors have argued that the
literature is not strong enough to support a causal association.
3.5.4 Cardiovascular Effects
Cardiovascular effects are currently seen as a potentially important set of mechanisms
whereby PM2 5 may be leading to premature mortality. In Chapter 12, we estimate the
quantifiable benefits of PM-related non-fatal acute myocardial infarction and cardiovascular
hospital admissions. The studies described in Section 3.5.1 found higher relative risks for
cardiopulmonary causes of death.
In addition to cardiopulmonary mortality, some studies have looked at morbidity. A
recent study from Germany also found significant increased odds of coronary heart disease
(CHD) in a cohort of approximately 3400 participants.307 Residents living within 150 meters of
major roads were compared to those living further ways. Overall, controlling for background air
pollution and individual risk factors, the adjusted odds ratio for CHD prevalence was
significantly elevated (1.85). Subgroup analyses indicated stronger effects in men, in
participants under 60 year of age, and in never-smokers.
Several additional studies have provided suggestive evidence that exposure to fresh
emissions from traffic predispose people to adverse cardiovascular events. Studies have focused
on both short-term variations in exposure, as well as long-term residential history. As discussed
in the summary section below, there are stressors in the roadway environment in addition to
ambient air pollutants (e.g., noise, anxiety) that also have an impact on cardiovascular activity.
The potential role of these co-stressors has not been adequately investigated.
A study from Augsburg, Germany interviewed survivors of myocardial infarction (MI)
shortly after they had recovered to examine ambient pollution and activities that might
predispose someone to having a heart attack.308 Survivors of MI were nearly three times as
likely to be in a car, in transit, or on a bicycle in the hour prior to the event as they were to be in
traffic at other times. Ambient air pollutants measured in the hour prior to MI at a central site in
the city were not associated with the risk of MI.
A study of healthy young North Carolina state patrolmen conducted by EPA's Office of
Research and Development monitored in-vehicle concentrations of PM2.5, VOCs, and metals.309
In-vehicle PM2.5 concentrations were associated with altered heart rate variability, an indicator of
cardiac stress. In-vehicle concentrations were also associated with increased concentrations of
factors in the blood associated with long-term cardiac risk, such as C-reactive protein, an
indicator of inflammation. This study provides information on possible mechanisms by which
cardiac stress could be induced by exposures to traffic-generated air pollution.
310
Heart rate variability has also been measured in a study of elderly residents of the Boston
area. In the study, ambient PM2.5 was associated with changes consistent with reduced
3-118
-------
autonomic control of the heart. Black carbon, often a more reliable index of traffic-related
pollution, was also associated with these changes. In a related study, ST-segment depression, a
cardiographic indicator of cardiac ischemia or inflammation, was associated with black carbon
levels as well.311 These studies further document a hypothesized mechanism associated with
motor vehicle emissions, but do not necessarily suggest effects independent of those identified in
our discussion of PM health effects.
3.5.5 Birth Outcomes
A few studies examining birth outcomes in populations living near major traffic sources
have found evidence of low birth weight, preterm birth, reduced head circumference and heart
defects among children of mothers living in close proximity to heavy traffic. Our discussion of
PM health effects also quantitatively accounts for premature mortality effects in infants and
qualitatively accounts for low birth weight.
One measure of exposure to traffic-generated pollution is "distance-weighted traffic
density," where traffic volume is treated as a measure that "disperses" along a Gaussian bell-
shaped curve evenly on both sides of a roadway. This approach captures some of the patterns of
dispersion from line sources, but does not account for micrometeorology. One study from Los
Angeles County, California employed this metric in a study of birth outcomes for births from
1994 to 1996. The study showed associations between distance-weighted traffic volume near
women's residences during pregnancy and premature birth and low birth weight in their
babies.312 The elevated risks occurred primarily for mothers whose third trimesters fell during
fall or winter months.
The same researchers had conducted an earlier study of births occurring between 1989
and 1993. In that study, consisting of over 125,000 births, exposures to ambient carbon
monoxide (CO), an indicator of traffic pollution, during the third trimester were significantly
associated with increased risk of low birth weight.313 In another study, preterm birth was
associated with ambient PMio and CO.314 These authors have also reported in a separate study
on the increase in cardiac ventricular septal defects with increasing CO exposure during the
second month of pregnancy.315 The role of socioeconomic status and factors associated with it
should be investigated in future study design.
Although the exposure metrics employed in these studies are based on surrogate
approaches to exposure estimation, other researchers have shown associations between New
York mothers' measured personal exposure to polycyclic aromatic hydrocarbons (PAHs) during
pregnancy and an increased risk of low birth weight and size.316 Subsequent follow-up of the
same birth cohort to age three found evidence of neurodevelopmental deficits associated with
maternal exposure to PAHs during pregnancy, particularly in cognitive development.317
Overall, although the number of studies examining perinatal exposures is small, there is
some evidence that exposure to traffic-related pollutants may be associated with adverse birth
outcomes, including low birth weight and preterm birth. However, given the variety of exposure
metrics employed and the relatively limited geographic extent of studies, the generalization of
3-119
-------
the conclusions requires a better understanding of relevant sources, pollutants, susceptibility, and
local factors.
3.5.6 Childhood Cancer
Several MSATs are associated with cancer in adult populations. However, children have
physical and biochemical differences that may affect their susceptibility to and metabolism of
MSATs. Particularly in the first year or two after birth, infants' liver enzyme profiles undergo
rapid change. As such, children may respond to MSATs in different ways from adults. Some
evidence exists that children may face different cancer risks from adults as a result of exposure to
certain MSATs and other components of motor vehicle exhaust. EPA recently recommended
default adjustments to cancer risk estimates for compounds with a mutagenic mode of action to
account for early life exposures in the Supplemental Guidance for Assessing Susceptibility from
Early-Life Exposure to Carcinogens.318
Evidence from human and animal studies suggests that increases in childhood leukemia
may be associated with in utero exposures to benzene and maternal and paternal exposure prior
to conception. Furthermore, there is some evidence that key changes related to the development
of childhood leukemia occur in the developing fetus.319
In the last 15 years, several studies have evaluated the association between maternal or
childhood residence near busy roads and the risk of cancer in children. Most studies to date have
been ecological in nature, with several employing individual-level exposure estimates within
cohort designs. The studies employed widely varying exposure metrics, including modeled air
quality, proximity to sources, and distance-weighted traffic volumes. Positive studies tend to
have used small population sizes, although one recent positive study used a large population.
Due to differences in ages studied, study design, exposure metrics, and study location (e.g.
Europe vs. U.S.), a systematic comparison between studies is difficult. A description of several
key studies from this literature follows.
One early study from Colorado showed significant elevated risk of childhood leukemia in
children under age 15 associated with living near roads with higher traffic volumes. The
strongest associations were with roads with at least 10,000 vehicles per day.320 The study was
reanalyzed using an approach to combine traffic volume with residential distance from major
roads to assess "distance-weighted traffic volume."321 The study found that the significant,
monotonically increasing risks associated with increased distance-weighted traffic volume.
NC>2 has been used as an indicator of traffic emissions in some studies; however, it is
important to note that NC>2 is not implicated as causing cancer. For instance, a study used a
dispersion model of NC>2 from traffic to conduct a case-control study of childhood cancer in
IT)
Sweden. The study found that in the highest-exposed group, risk of any cancer was
significantly elevated. Risks in the most-exposed group were also elevated for leukemia and
central nervous system tumors, but were not statistically significant.
These earlier studies were based on relatively small populations of children with cancer.
In response, subsequent studies focused on either replicating the earlier studies or studying larger
3-120
-------
groups of children. A study in Los Angeles, California applied the same distance-weighted
traffic volume approach as the earlier Colorado study, but found no elevation in risk in a larger
group of children.323 A large study of nearly 2,000 Danish children with cancer found no
association between modeled concentrations of benzene and NO2 at home and the risk of
leukemia, central nervous system tumors, or total cancers.324 However, the study did find a
dose-dependent relationship between Hodgkin's disease and modeled air pollution from traffic.
Several large studies were conducted in California using a statewide registry of cancer.
These studies employed study sizes of several thousand subjects. In one cross-sectional study,
the potency-weighted sum of concentrations of 25 air toxics modeled using EPA's ASPEN
model was not associated with mobile source emissions, but increased rates of childhood
leukemia were found when accounting for all sources of air toxics together, and for point sources
separately.325 Another study from the same researchers found that roadway density and traffic
density within 500 meters of children's homes was not associated with risk of cancer.326
Most recently, a novel approach to assessing childhood leukemia in relation to early life
exposures was employed in the United Kingdom. The study examined all children dying of
cancer between 1955 and 1980, consisting of over 22,000 cases. Birth and death addresses of
children with cancer who moved before death were compared with regard to proximity to nearby
sources and emissions of specific chemicals.327 An excess of births near sources, relative to
deaths, was used to indicate sources in early life associated with greatest cancer. Greater risks
were associated with birth addresses within 300 meters of high emissions of benzene, 1,3-
butadiene, NOX, PMio, dioxins, and benzo[a]pyrene. In addition, births within 1.0 km of bus
stations, hospitals, freight terminals, railways, and oil installations were associated with elevated
risk. Overall, locations with the highest emissions of 1,3-butadiene and carbon monoxide
showed the greatest risk.
In summary, the lack of consistency in results between large studies and the multiplicity
of study designs makes it difficult to draw firm conclusions. Epidemiologic methods for
detection of childhood cancer risks may lack sufficient power to detect risks with precision.
However, given the well-established carcinogenicity of benzene and 1,3-butadiene in the
toxicological and occupational epidemiologic literature, and data suggesting exposure to benzene
prior to conception and in utero can lead to increased risk of childhood leukemia, the potential
for public health concern is present. The standards proposed in this rule will reduce such
exposures.
3.5.7 Summary of Near-Roadway Health Studies
Taken together, the available studies of health effects in residents near major roadways
suggest a possible public health concern. These studies' exposure metrics are reflective of a
complex mixture from traffic, and the standards will reduce a broad range of pollutants present in
higher concentrations near roadways. It is unclear to what extent these health effects are
attributable to PM versus other components of the complex mixture. Note that the benefits
associated with the direct PM reductions from the cold temperature vehicle standards are
presented in Chapter 12 of this RIA.
3-121
-------
3.5.8 Size and Characteristics of Populations Living near Major Roads
In assessing the public health implications of near-roadway health concerns, some
understanding of the population living near major roads is required. Those living near major
roadways are a subpopulation of the total population included in quantitative analysis, and to the
extent that there may be additional exposures and health effects not captured in analyses for the
total population, we enumerate the size and characteristics of the subpopulation. A study of the
populations nationally using geographic information systems indicated that more than half of the
population lives within 200 meters of a major road (see file USbytract.txt in the docket for this
rule).8 It should be noted that this analysis relied on the Census Bureau definition of a major
road, which is not based on traffic volume. Thus, some of the roads designated as
"major" may carry a low volume of traffic. Detailed analyses of data were conducted in three
states, Colorado, Georgia, and New York. In Colorado, 22% live within 75 meters of a major
road, while an additional 33% live between 75 and 200 meters of major roads. In Georgia, the
respective percentages are 17% living within 75 meters and an additional 24% living between 75
and 200 meters. In New York, the percentages are 31% and 36%.328
To date, the only source of national data on populations living in close proximity to major
transportation sources is the American Housing Survey, conducted by the U.S. Census
Bureau.329 This study characterizes the properties and neighborhood characteristics of housing
units throughout the U.S. According to the Census Bureau's summaries of the 2003 survey,
among approximately 120,777,000 housing units in the nation, 15,182,000 were within 300 feet
of a "4-or-more-lane highway, railroad, or airport." This constitutes 12.6% of total U.S. housing
units. A simple assumption that the U.S. population is uniformly distributed among all types of
housing leads to the conclusion that approximately 37.4 million people live in what might be
considered a "mobile source hot spot."
According to the American Housing Survey's summary tables, occupied housing units in
central cities are 35% more likely to be close to major transportation sources than housing units
in suburban areas.330 Furthermore, nationally, housing units that are renter-occupied are 2.3
times more likely to be close to major transportation sources, compared to housing units that are
owner-occupied. In the 2003 American Housing Survey, median household income for owner-
occupied units was $52,803, while only $26,983 for renter-occupied units. These statistics imply
that those houses sited near major transportation sources are likely to be lower in income than
houses not located near major transportation sources.
A few population-based epidemiology studies have also examined whether discrete
groups of people live close to major roadways. In one study of veterans living in southeastern
Massachusetts, 23% lived within 50 meters of a "major road," 33% lived within 100 meters, and
51% within 200 meters.331 In examining traffic volumes, 13% lived within 50 meters of a road
with annual average daily traffic of 10,000 vehicles or more, while other distances were not
analyzed.
In another study using 150 meters as a definition of "near" a road, 2.3% of California
s Major roads are defined as those roads defined by the U.S. Census as one of the following: "limited
access highway," "highway," "major road (primary, secondary and connecting roads)," or "ramp."
3-122
-------
public schools were found to be near a road with more than 50,000 vehicles per day, while 7.2%
were near roads with between 25,000 and 49,999 vehicles per day.332 This corresponded to 2.6%
and 9.8% of total enrollment, respectively. In that study, traffic exposure increased, the fractions
of school populations comprised of black and Hispanic students also increased, as did the
fraction of children in government-subsidized meal programs.
Another study in California defined the issue differently, examining the child population
living in census block groups and traffic density.333 The study found that approximately 3% of
the state child population resided in the highest traffic density census tracts. Furthermore, block
groups with lower income were more likely to have high traffic density. Children of color were
more likely than white children to live in high traffic density areas.
In summary, a substantial fraction of the U.S. population lives within approximately 200
meters of major roads.
3-123
-------
Appendix 3A: Influence of Emissions in Attached Garages on
Indoor Air Benzene Concentrations and Human Exposure
Introduction
Measurement studies provide strong evidence that VOC sources in attached garages can
significantly increase VOC concentrations inside homes.334 Preliminary analyses of data from a
pilot study for the National Human Exposure Assessment Survey (NHEXAS) in Arizona also
found indoor concentrations of mobile source-related VOC compounds significantly higher in
homes with attached garages than in homes without them.335 This population-based exposure
study included measurements from 187 homes. A study in 50 Alaska residences found that in
homes with attached garages, indoor benzene levels averaged 70.8 ng/m3, while in homes
without attached garages, concentrations averaged 8.6 |j,g/m3.336 Multiple factors, including
house architecture, ventilation design, garage configuration, and climate can all play roles as
well.
National-scale air toxics modeling efforts, such as those discussed in RIA Section 3.2.1.2,
employ Gaussian dispersion models in combination with human exposure models to calculate the
concentrations of air toxics in various microenvironments. Exposure models calculate an
average exposure resulting from the movement of a simulated population through a time-activity
pattern that brings them into contact with air in the various microenvironments.
At this point, the NATA and the analyses performed for this rulemaking have only
included exposures from outdoor sources. Although the HAPEM6 exposure model is capable of
addressing indoor sources, more thorough analyses of the prevalence and use of emission sources
within attached garages are required to develop quantitative estimates of model parameters to
address attached garage contributions across the U.S. population.
This appendix addresses the potential impact of all benzene sources within an attached
garage on residential indoor air quality.
Methods
Calculation ofWithin-garage Source Emission Rate
Emission rates for indoor sources of VOCs can be derived by several methods. Most
accurately, the actual emission rates of an indoor VOC source can be measured through the use
of a Sealed Housing for Evaporative Determination (SHED). However, test conditions must be
representative of real world applications. Short of SHED-based measurement, several surrogate
approaches may be employed. For evaporative losses from a sealed container, the change in
weight of a container over time may be used to calculate a total mass loss rate, which can be
assumed to be in the form of VOC. Alternatively, if the air concentrations and ventilation
conditions of a defined indoor space are known, mass balance equations can be employed to
derive a "virtual" emission rate for all sources within the space.
This appendix employs the latter approach in calculating source emission factors. The
general approach of a mass balance equation is to calculate the change in mass over a given time,
accounting for the mass of a pollutant transported into a space, the mass of pollutant transported
out of a space, the emission rate of a source within the space, and the decay of any pollutants
3-124
-------
within the space, which can be treated as a first-order decay. A simple space like a garage can be
treated as a single zone. The differential equation representing this mass balance is as follows:
^^ = ckd_L+^_cjv_
dt ° dt dt ' dt
Here, dMt/dt represents the rate of change of total indoor mass, Q is the indoor concentration,
C0 is the outdoor concentration, dV/dt is the volumetric air flow through the space, k is the
penetration fraction indicating the proportion of mass that passes through the wall of the
compartment, and dM/dt represents the mass emission rate inside the space. Note that all air
entering the garage is assumed to enter from outdoors.
Assuming steady-state conditions, dMt;i/dt assumes the value of zero, meaning that the
concentration in the garage does not change over time. Algebraically, this allows the equation
above to be represented as:
(2) ffk-C.*)^
\dt j dt
In other words, the indoor source terms can be calculated if the volumetric flow through the
space and concentrations indoor and outdoor are known. Any gradient in concentration between
indoor and outdoor concentrations is explained by indoor sources and the fraction of mass that
does not penetrate from indoors to outdoors.
The volumetric flow can be calculated by multiplying the volume of the space by the
number of times per hour that the air within the space is turned over. As such:
(3) ^--aV
dt
Here, a is the "air exchange rate," expressed in air changes per hour (ACH). Combining
equations (2) and (3), the mass emission rate is represented as:
(4) av(C,-C0k)-^-
A recent study in Ann Arbor, MI measured the air exchange rates and the in-garage and
outdoor concentrations of VOCs needed to perform these calculations.337 The homes in the
study were based on a convenience sample, and so may not be generally representative of the
local or national housing stock. All garages but one adjoined a house. All attached garages had
between one and three walls adjoining a residence. The distributions of garage benzene
concentration and ACH are shown in Figure 3 A-l. The distributions of each were not
significantly different from lognormal, judging by the Kolmogorov-Smirnov Z statistic.
Values of k, the penetration factor, are dependent on the physical pathways through
which air passes into a garage, as well as the presence and chemical composition of any
insulating material through which air passes. In the case of garages, the infrequency of insulated
garages and the low reactivity of benzene justifies the assumption that k=l.338
These data from the Ann Arbor, MI study were used to solve equation (2) to derive a
distribution of benzene mass emission rates in each garage in the study, based on variability in
measurements of outdoor concentrations. Equation 4 was implemented using a Microsoft Excel
spreadsheet with the @Risk probabilistic simulation add-in (version 4.5).339 Monte Carlo
sampling was used for all terms in deriving the emission rates.
As described below, this distribution can be used to evaluate the effect of various fuel
control measures on indoor benzene concentrations. A single lognormal distribution was used to
3-125
-------
represent C0 in equation 4, based on other studies of ambient air, which have found that many
pollutants' concentrations are lognormally distributed.
Calculation of Garage Contributions to Indoor Air
In the same way that a mass balance calculation can be used to calculate emission rates
for sources within garages, a mass balance equation can be used to estimate the additional
concentration in a home that will occur as a result of elevated concentrations in the garage.
However, unlike the garage case, it is not valid to assume that all air entering the home comes
directly from outdoors.
Recent studies have provided indications that over multiple sequential days, variability in
within-home benzene concentration is relatively small. A recent study from Ann Arbor, MI
found a coefficient of variation (COV) of 4.6% for benzene.340 Furthermore, recent data
obtained by EPA through the Environmental and Occupational Health Sciences Institute
(EOHSI) on homes in the Elizabeth, NJ area indicates no significant differences in within-home
concentrations at a 95% confidence level.1'341 These data are preliminary, and analyses are still
in progress.
Given the fraction of air entering the home through the home-garage interface, the
appropriate mass balance equation for a single-compartment (e.g. well-mixed) home can be
represented as such:
r.dV dV dV
Here, Q is the in-house concentration, C0 is the outdoor concentration, Cg is the concentration in
the garage, dV/dt is the volumetric air flow through the house, and^is the fraction of air entering
the home from the garage. One assumption made here is that the penetration factor for the air
moving through the house-garage interface is the same as air moving through the house-outdoors
interface. Reactive decay is assumed to be zero. Such mass balance equations are standard
approaches in environmental science and engineering, and are frequently found in textbooks on
these subjects.342
Again assuming steady-state conditions, dM^/dt = 0, the equation above simplifies to:
(6) Cl=kC0(l-fg) + kCgfg
Or more simply, the indoor concentration under steady state conditions is proportional to the
fraction of air entering the house through the garage.
Figure 3A-2 is a contour plot illustrating the range of average indoor air concentrations
that could plausibly arise given a range of values of Cg and^g, with a background concentration
of zero. However, Figure 3A-2 does not answer the question of what the likely indoor air values
are in a sample of real homes.
The text below describes procedures and results of a small-scale modeling study.
Modeling Approach
' In that study, one air sample was obtained in the room adjacent to an attached garage in each home and another was
obtained in another location. The New Jersey Department of Environmental Protection and EPA provided joint
funding for the study. A two-sided paired t-test was applied to data obtained from 36 homes over approximately 24
hours.
3-126
-------
All modeling analyses employed Equation 6 in a Microsoft Excel spreadsheet with the
@Risk probabilistic modeling add-in was the software employed in all modeling analyses.
Where appropriate, each of the terms in Equation 6 was treated as a random variable represented
as either a parametric distribution or as an empirical distribution based on measured data.
Often, in employing data obtained from more than one study, combining data into a
single distribution was not justified on a priori grounds. In ventilation studies, ambient
conditions such as temperature and geography can substantially affect air flow patterns and
building constructions. For instance, residential air exchange rates differ significantly between
regions with substantially different climates.343 Furthermore, based on the limited number of
studies available, combining data from multiple studies into a single data set had the potential to
apply de facto weights to data, potentially shifting the fitted model parameters away from truly
"representative" distributions.
Another consideration is the potential for independence of the^ and Cg variables. There
is no a priori reason why the "leakiness" of the house-garage envelope should be related to the
concentration of benzene in the garage.
Because of these considerations, data onfg or Cg from studies in different areas were not
formally combined. Rather, distributions fit separately to data from each study were used to
develop several model "scenarios." As described below, four different studies provided data for
Cg and three different studies provided data forfg. As such, a minimum of 12 (3 x 4) scenarios
were needed to represent the totality of available data.
For each scenario modeled, @Risk sampled from each distribution 20,000 times using a
proprietary Latin Hypercube sampling framework. The large number of samples and Latin
Hypercube strategy were employed to ensure that modeled concentration distributions achieved
stability.
Lastly, for comparison to the current approaches for exposure modeling, the following
equation was used, paralleling the approach taken by HAPEM5 with no garage emissions:
(7) Ct = kC0
Data for Populating Model Parameters
Fraction of Air Entering Home through the Garage (fg)
Several studies have examined the fraction of air entering the home from the garage.
Except for one, all of these studies took place in northern states and Canada, where homes are
built with more insulation. A recent study of a set of homes in Ontario, Canada found that
approximately 13% of the air entering the home came from the garage.344 One study from
Minnesota found that in newer homes, houses built in the year 1994 had an average of 17.4% of
total air leakage coming through their garages, houses build in 1998 had an average leakage
fraction of 10.5%, and houses built in 2000 had an average leakage fraction of 9.4%.345 Two
recent studies have employed perfluorocarbon tracer (PFT) gases to estimate air transport
between different "zones" of houses with attached garages. A recent study by Isbell et al. (2005)
based in Fairbanks, Alaska found that in a modern air-tight Alaskan home ventilated with an air-
to-air heat exchanger, 12.2% of the air entering a home entered through the garage, while 47.4%
of the air entering an older home ventilated passively by structural defects came through the
attached garage.346 Another study of a home in Ann Arbor, Michigan built in 1962 found that
3-127
-------
16% of the air entering the home originated in the garage.347 In a more recent study from Ann
Arbor, investigators deployed PFT tracers in 15 homes and calculated the fraction of air entering
each home through an attached garage, with an average of 6.5±5.3% of the air entering through
the garage.348 From these studies, it is apparent that across homes, the fraction of air entering
through the garage is highly variable, making it necessary to acknowledge significant
uncertainties in characterizing "typical" infiltration patterns.
Benzene Concentrations in Garage Air (Cg)
Four sources of in-garage concentration data are available in the format relevant for
steady-state modeling over extended periods of time. First, there is the study by Batterman et al.
(2005), in which average garage concentrations of benzene were measured over a period of four
days in each of 15 homes using passive sampling badges. The average garage concentration
reported was 36.6 |J,g/m3, with a standard deviation of 38.5 |J,g/m3.
Second, a study in Alaska by George et al. (2002) measured benzene concentrations in 28
Alaska homes and 48 garages with passive diffusion badges.349 One disadvantage of this study
is the relatively high detection limit for benzene, 7 ppb (22 ng/m3). As a result, many of the data
available are based on a reported value of 50% of the detection limit. In the Alaska study, in-
garage benzene concentrations averaged 103 ng/rn3, and the standard deviation was 135 ng/rn3.
The study included concurrent in-home measurement of benzene in homes with attached
garages, allowing evaluation of the modeled indoor concentrations. However, it is not apparent
that this study underwent scientific peer review.
A third study in one New Jersey home also evaluated garage and indoor benzene, as part
of an investigation into in-garage emissions of vehicles fueled with methanol blends.350 Only
one home was sampled, but it was sampled multiple times inside the garage and at multiple
locations inside the residence. A fourth study from Fairbanks, Alaska conducted measurements
in 12-hour periods on four separate days in two houses in two seasons, summer and winter.351
The study obtained two daily measurements of benzene concentration within each garage over a
12-hour sampling period. One home was a modern, well-insulated home with an air-to-air heat
exchanger for ventilation. The other was an older home ventilated passively by structural defects
in the building envelope. Because of the large differences in concentrations between homes and
seasons, data from each home-season combination was treated as a separate distribution within
the indoor air model (Equation 6 in Excel/@Risk). Treating these data as separate distributions
increased the number of modeled "scenarios" to 21 (3fgx7 Cg).
3-128
-------
Penetration Factor (k)
The values of k in this case were obtained from the HAPEM5 user's manual, using the
PEN-1 factor, representing the fraction of benzene from outdoor air penetrating indoors. The
values in HAPEM5 are presented as a distribution that assigns a 2/3 weight to the value 0.8 and a
1/3 weight to the value 1. These estimates are based on a comprehensive review of indoor and
outdoor air quality studies.
Outdoor Ambient Concentration (C0)
The 1999 National Air Toxics Assessment (NATA) provided ambient concentration
estimates for every census tract in the U.S. For this modeling exercise, a lognormal distribution
was fit to these data.
Results
Within-Garage Emission Rates
Equation 4 was used with Monte Carlo sampling to calculate a distribution of emission
factors for each home, based on the variability in outdoor concentrations reported in Batterman
et al. (2005). As shown in Figure 3-A3, the within-garage variation was a very small component
of overall variability compared to between-garage variation. This finding implies that the factors
in individual garages, such as storage of vehicles, nonroad equipment, and fuels, have a major
effect on in-garage concentrations.
In aggregate, the mean emission rate for all garages sampled fell along a lognormal
distribution (p > 0.05). The mean emission rate was 3049 ng/hr (73 mg/day), with a standard
deviation of 4220 ng/hr (101 mg/day).
To evaluate the plausibility of these steady-state emission factors, known emission
factors for other emission sources were evaluated. The California Air Resources Board (CARB)
conducted a study of emissions from portable fuel containers, finding that volume-specific
emissions rates for total VOC due to evaporation and permeation was 0.37 g/gal-day. Assuming
an average fuel container volume of two gallons, the average emission factor per can would be
0.74 g VOC/day.
To evaluate the derived emission rates relative to CARB's measurements, a benzene fuel
vapor pressure fraction of 0.5-1% was assumed, based on MOBILE6.2 evaporative emission
factors. Given that assumption, the average benzene emission rate from CARB's study is 3.7-7.4
mg/day. This value is in the lower range of emission rates shown in Figure 3A-3. This
comparison suggests that emissions due to permeation and evaporation from portable fuel
containers may be a relatively small fraction of overall garage benzene.
Subsequently, one additional study used perfluorocarbon tracers (PFT) and VOC
measurement in two Fairbanks, Alaska homes to estimate two garages' "source strengths" for
benzene.352 For a new, energy efficient "tight" home with an air-to-air heat exchanger, median
garage emission estimates for benzene were 21 mg/h in summer and 14 mg/h in winter. In an
older home with passive ventilation due to structural defects, median benzene source strengths
were calculated at 40 and 22 mg/h in summer and winter, respectively. These values are
substantially higher than those calculated based on Batterman et al. (2005). However, the
3-129
-------
difference may be attributable to higher fuel benzene in Fairbanks than in Michigan. Study
design may also play a key role. In the Fairbanks study, the measurement periods were 12 hours
each in duration. In the Michigan study, measurement periods lasted four days each. The
Michigan study's longer duration may have allowed for a broader range of emissions activities
than the Fairbanks study.
Garage Contributions to Benzene in Indoor Air
Figures 3A-4 to 3A-8 display the results of @Risk simulations of indoor air. Each figure
represents the modeled outputs as cumulative probability distributions. In the legend of each
figure, the label of each distribution describes its^ and Cg sources. For instance, "George et al.
(2002) / Fugler FG Ci" indicates a distribution using garage concentration data from George et
al. (2002) andfg data from Fugler et al.
Figure 3A-4 presents the output of Equation 6, a daily average indoor benzene
concentration including contributions from outdoor air and from attached garages. As noted in
the "Methods" section of this appendix, it was necessary to run a large number of scenarios to
account for different combinations offg and Cg data sources. The figure depicts results using
studies that contain Cg data from multiple homes as bold solid lines, while the model simulations
based on studies that employ Cg data from only one home are shown in dashed lines. As
indicated in the figure, there is no major difference in the Ct distributions predicted by using Cg
data from multiple homes or by using Cg measured from a single home. The average modeled
indoor benzene concentrations ranged from 2.9 to 16.4 |j,g/m3.
For comparison, Figure 3A-5 presents cumulative distributions of the observed results
from several studies that measured indoor air concentrations in homes with attached garages.
Schlapia and Morris (1998) measured integrated 24-hour benzene concentrations inside 91
homes with attached garages in Anchorage, Alaska between 1994 and 1996.353 George et al.
(2002) reported average benzene concentrations in 36 homes in Anchorage, Alaska, but no
distributional data. Mentioned above, Isbell et al. (2005) also measured integrated 12-hour
benzene in two seasons in one modern air-tight home ("Home V" in Figures 3 A-4 to 3 A-8) and
one older passively-ventilated home ("Home NV" in Figures 3A-4 to 3A-8).354 Both homes
were located in Fairbanks, Alaska. Batterman et al. (2006) measured indoor air benzene
concentrations in 15 homes in southeastern Michigan over four-day sampling periods throughout
spring and summer of 2005.355 Lastly, Weisel (2006) conducted a study of indoor air in 21
homes in Union County, NJ between April 2005 and January 2006. One monitor in each home
was sited in the room adjacent to the garage, while another was located in another part of the
house.356
Comparing Figures 3 A-4 and 3A-5, it is apparent that the distributions of modeled indoor
air concentrations of benzene are very similar to those observed in monitoring studies. Both
figures indicate that there is substantial variability in concentrations between homes and between
studies.
Figure 3 A-6 presents the mean concentrations from modeling scenarios and from
monitoring studies. In general, the range of mean concentrations is close to the values monitored
in the indoor air studies. Notable exceptions are the indoor air values by George et al. (2002),
the winter data from the passively-ventilated "NV" home from Isbell et al. (2005), and by
Schlapia and Morris (1998). All of these studies took place in Alaska, which may have uniquely
high benzene fuel levels or housing architectures that create higher garage air infiltration indoors.
3-130
-------
Of particular note, all of these studies included substantial numbers of homes with "tuck-under"
garages where one or more rooms of a house are situated above a garage. Schlapia and Morris
(1998) reported a very high average value that was not matched by the "average" conditions of
any other run. It is notable that this high value is the average across 91 homes with attached
garages.
Another consistent trend shown in Figure 3 A-6 is that scenarios employing^ data from
Batterman et al. (2006) produced consistently lower average benzene concentrations than
scenarios employing other sources. This trend is attributable to the lower average fg reported in
Batterman et al. (2006), 6.5%, as compared to values found in Sheltersource (11.7%) and Fugler
etal. (13.6%).
It is unclear whether the studies measuring Cg,fg, and Ct constitute a representative
sample of homes. In general Alaskan studies report higher concentrations, but not consistently.
The relatively greater prevalence of homes with "tuck under" garages in some Alaskan studies
may explain this discrepancy.
In comparison to the values reported in Figures 3 A-4 and 3 A-5, indoor air concentrations
calculated with the default d = kC0 approach, similar to that employed in the national-scale
modeling for this rule, averaged 1.2 ng/m3.
Overall, modeled concentrations presented here appear to provide a credible estimate of
indoor benzene concentration in homes with attached garages. However, it is unclear whether
the homes included in the studies employed herein may be considered "representative."
Implications
Effect on Exposures Nationwide
In calculating the hypothetical effect of attached garage on national estimates of chronic,
time-weighted average (TWA) human exposure, precise estimates are not possible. As noted
previously, the extent to which available studies of indoor air of homes with attached garages is
representative of the entire population of such homes is unclear. Furthermore, the distribution of
housing stock by climate and meteorology is not well understood. However, despite these
limitations, a bounding exercise is still feasible.
One simple approach for such a bounding exercise is determined by the following
equation:
(8) ES = C1S*P*TS
\ / & l'& & &
Here, Eg represents the national average exposure to benzene in air attributable to
attached garages. Ciig represents the average indoor concentration attributable to an attached
garage, Pg represents the fraction of the population living in a home with an attached garage, and
Tg represents the time spent in a home with an attached garage.
C!;gis derived from Equation 6, and can be derived by setting the outdoor concentration
term (C0) to zero. An estimate of the attached garage contribution to indoor air can be made for
studies with only indoor measurements as well. This can be accomplished by substituting
ASPEN concentration estimates for the county in which each study took place. For Equation 6,
C0 estimates from NATA for each census tract in the relevant county were assembled into a
lognormal distribution. With this data and the other assumptions of Equation 6, an estimate of
d:g could be derived from the measurement studies.
3-131
-------
To estimate Pg, an estimate of the national fraction of homes with attached garages is
required. The Residential Energy Consumption Survey (RECS), run by the U.S. Energy
Information Administration, provides an estimate of the fraction of homes with attached
garages.357 RECS estimates a total of 107.0 million housing units nationally, 37.1 million
(34.7%) of which are homes with attached garages. Assuming that the population is uniformly
distributed across housing units allows this figure to serve as an estimate ofPg.
Information on the fraction of time spent in a residence (Tg) is required to determine how
the microenvironmental concentration in homes with attached garages affects overall time-
weighted exposure concentrations. As cited in EPA's Exposure Factors Handbook, the average
person studied by the National Human Activity Pattern Survey (NHAPS) spent 1001.39 minutes
(16.68 hours) per day indoors within any room of a residence.358
Results of model simulations using Equation 7 are shown in Figure 3A-7. As before, the
results of each combination of Cg andfg data source are shown. For each study, the legend lists
the source for both Cg andfg data. As described above, the estimates C!;g derived from indoor air
measurements are also presented in Figure 3A-7. In the legend of Figure 3A-6, these studies are
denoted by the term "Direct Ci Measure." As shown, there tends to be a greater degree of
agreement between modeling scenarios for lower concentration estimates, but less agreement for
higher concentration estimates.
As described above, it is unclear to what the extent to which the homes studied for
benzene related to attached garages are representative of homes nationally. As such, in
summarizing the scenarios, several different approaches to "averaging" across scenarios are
presented here. Figure 3A-8 shows the results of these different averaging scenarios. In the "All
Data" distribution shown in the figure, all scenarios are averaged together. In the "Weighted
Average" distribution, weights are equal to the number of homes included in each study. In the
"Model Only" distribution, only scenarios involving modeling C, are shown. In the "Measure
Only" distribution, only those studies in which C, was measured directly are shown. In the "AK
Only" distribution, only scenarios employing Alaskan Cg orfg studies are shown. In the "Non-
AK Only" distribution, only scenarios excluding Alaskan Cg orfg data are shown. These
scenarios are intended to span a range of estimates for the national estimate.
The average concentrations from these "summary scenarios" are shown in Table 3A-1.
As shown in Table 3A-1 and in Figure 3A-8, scenarios employing only measured indoor data
resulted in higher predicted benzene TWA exposure concentrations than the studies employing
only modeling. Scenarios employing Alaskan data result in higher benzene concentrations than
scenarios excluding Alaskan data. Also weighting scenarios by the number of homes resulted in
higher benzene concentrations.
Accordingly, the national average TWA exposure concentration attributable to attached
garages is estimated to be 1.2 - 6.6 ng/m3. This range is intended to span possible values of
average TWA exposure from attached garages, given currently available information. The actual
average TWA exposure concentration due to attached garages could be outside of this range.
Because of limited information on the representativeness of the homes studied, a more precise
"central estimate" is not appropriate at this time. The width of the range, with the upper end
being 5.5 times the lower end, is an indicator of the magnitude of uncertainty in the estimate. It
is not a confidence interval in the traditional sense. As more data become available, more
precise estimates will hopefully emerge.
In comparison, the national average exposure concentration of census tract median
exposure concentrations in this rule is estimated at 1.4 |J,g/m3 for calendar year 1999.
3-132
-------
Accordingly, if the attached garage exposure contribution is considered, the estimate of national
average exposure to benzene rises to 2.6 - 8.0 |J,g/m3, corresponding to an increase of 85-471%.
Effects of Emission Standards
Several limitations prevent precise estimation of the effect of the standards in this rule on
garage-related exposures. First, cold temperature vehicle ignition and evaporative vehicle,
engine, and fuel container emissions can occur either in a garage or outdoors. Second, detailed
tracking of the time during which people are inside a house during cold vehicle starting or hot
vehicle soaking, when a majority of benzene emissions are likely to occur, is limited. However,
a bounding exercise can provide some estimates as to the effect of the standards in this rule.
First, assuming full mixing and steady-state conditions, concentrations within a garage is
estimable359 as:
(9) Cg = (dMr/df) I aV
Here, the terms are similar to Equations 1-7.
Given a change in the mass benzene emission rates from vehicle cold temperature
ignition, fuel evaporation from vehicles, engines, and fuel containers, an estimate of a change in
Cg is feasible. Table 2.2-52 of the RIA displays the emission reductions attributable to each
program. By splitting the emission reductions into evaporative and exhaust emissions and
applying several simple assumptions about where emissions occur (in garage vs. outdoors), the
fraction of emission reductions occurring within attached garages can be estimated. This
estimate is calculated by assuming ranges of values for the fraction of evaporative and exhaust
emissions from each program that occur within an attached garage." As such, while the total
benzene mobile source and PFC emission reductions occurring as a result of the rule in 2030 are
37% less than the projected emissions without controls (Table 2.2-52 of the RIA), emissions
inside attached garages are reduced by an estimated 43-44%.
Applying this fraction to Equation 8 and Equation 7, for the "average" scenarios modeled
presented in Table 1, this amounts to a national average exposure reduction of approximately 0.5
-2.6
Limitations
As apparent in the wide range of "scenario" averages, there remains considerable
uncertainty in ascertaining the true magnitude of attached garage exposure contributions
nationally. There are a number of limitations in the approaches undertaken here. First, although
comparison with measured indoor data shows reasonable performance for the modeling approach
employed here, the selection of simple one-compartment mass balance models for both garage
and home modeling may substantially understate the variation in concentrations within these
microenvironments. All estimates here assumed steady-state conditions, and this may not be
u The assumed fraction of evaporative and exhaust emission reductions from each source occurring within an
attached garage are as follows. Ranges are represented as [min, max]. For LDGV, about 90% of emission
reductions are exhaust-related, of which Pg*[25%,75%] occur within attached garages; the fraction of evaporative
reductions occurring within attached garages are Pg *[25%,50%]. For small nonroad gasoline equipment, about
72% of emissions are from exhaust, of which Pg *[0%,2%] occur in attached garages; 24% are evaporative, of which
Pg *[90%,100%] occur in attached garages, 4% are refilling-related, of whichPg *[25%,75%] occur in attached
garages. For portable fuel containers, Pg *[25%,75%] of emissions are assumed to occur in attached garages.
3-133
-------
appropriate for a source like a garage, where door opening, car entry and ignition, and other
major sources of benzene are likely to produce short-term spikes in exposure not accounted for
with steady-state assumptions.
Second, the preponderance of these data were collected in locations with cold climates,
so the results may not be applicable to warmer locations where houses are not built with the
same degree of weather-tightness. Furthermore, studies suggest that indoor concentrations
arising from attached garages vary considerably in response to emission-related activities in a
garage such as cold vehicle ignition and parking a hot vehicle.360 Ambient temperatures may
affect the magnitude of emissions from these activities.
Lastly, the extent to which the houses studied in the publications cited here are
"representative" of the national housing stock is unknown.
Conclusions
Modeled indoor benzene concentrations indicate that indoor air concentrations in homes
with attached garages may be substantially higher than in homes without attached garages.
Garage concentrations of benzene appear to be a major source of indoor benzene in
homes with attached garages. According to the modeling conducted here, this source could
explain the majority of exposures experienced by typical residents of such homes. Given this
finding, interventions that result in a reduction in emissions within the garage would be a
relatively efficient means of reducing overall personal exposure, particularly in areas
geographically similar to the areas of the studies upon which this analysis relies. Given the
proximity of this source to homes, one major set of beneficiaries of the rule's emission controls
is likely to be people with homes with attached garages, particularly in areas with high fuel
benzene levels. Emissions from vehicles and fuel containers also may have greater relative
impacts on those with attached garages. An elementary calculation of the intake fraction (iF) of
emissions occurring within attached garages with very basic assumptions indicates that for
benzene emitted in a garage, approximately 3-18 parts per thousand are inhaled by a person in an
attached garage. This estimate is far in excess of estimated iF from ambient sources, and similar
to estimated iF estimates for indoor sources.361
3-134
-------
Table 3 A-l. Summary of National Average Exposure Estimates Attributable to Attached
Garages. Different "averaging" assumptions shown.
"Averaging"
Scenario
All Data
Weighted Average
Measure Only
Model Only
AK Only
Non-AK Only
Benzene TWA
(ug/m3)
4.3
6.6
6.1
3.4
5.5
1.2
3-135
-------
Figure 3A-la. Density of Garage Benzene Concentrations from Batterman et al. (2005)
Density of Garage Concentrations
Q
CD
O
CN
Q
O
p
o
o
o
o
0 50 100 150
Benzene Concentration (ug/m3)
Figure 3A-lb. Density of Air Exchange Rates (ACH)
Density of Garage ACH
oq
o
-------
Figure 3A-2. Additional Indoor Air Concentrations from Garage as a Function of Cg andfg
Indoor Concentration as a Function of Garage Concentration (Cg) and %lntake Air
from Garage (fg)
Cg=110
1=100
• 40-45
D 35-40
• 30-35
D 25-30
• 20-25
D15-20
D10-15
• 5-10
DO-5
3-137
-------
Figure 3A-3. Distributions of Individual Garage Emission Factors
Emission Rate (ug/hr)
O
O
O
O
O
O
CD
O
O
O
00
O
O
O
O
O
O
O
O
O
O
O
O
O
CD
O
O
O
00
O
O
O
p
CO
35
H
3Ł
35
3E
35
1 5 i
p
CO
p
bo
p
CD
IH
Q)Q)Q)Q)Q)Q)Q)Q)Q)Q)Q)Q)Q)Q)Q)
Q)Q)
CQCQ
CDCD
Q)Q)Q)Q)Q)Q)Q)Q)Q)Q)Q)Q)Q)
CQCQCQCQCQCQCQCQCQCQCQCQCQ
CDCDCDCDCDCDCDCDCDCDCDCDCD
3-138
-------
Figure 3A-4. Cumulative Distribution of Modeled Indoor Benzene Concentrations
Benzene Concentration (ug/m3)
p
o
8
rt-
0"
a
p
o
-n N>
(D O
= OJ
fl> o
"O cn
c o
1 *
8 §
< °
aT gs
O"
= oo
Zr O
CD
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
o
p
o
o_
OL
(D
0)
S
c_
^
-a_
n>
I
o
(D
W
(D
(/>
s
O
Q.
2.
ffi
Q.
Q.
O
O
O
^ s-
w
n> =• n
-o » is!
fl> g (/)
I s =
Q. 0) CT
ll O
o
Q.
D
fi>
0)
O
*• »
CD CD
ra ra
00)/FuglerFC
00) / Shelterso
Ł= u'
S 0
CD
Tl
O
O
CD
-g
O
?
0)
S
o
(fl
S
Si
o
Home NV, W
nter / Batt
3
=3
-n
O
o
i
C/)
CD
Si
O
g
ter / Batte
g
O
(fl
CD
Si
o
Home NV, Si
g
g
CD
CD
1
g
ZJ
-n
(,)
O
C/)
CD
Ł
§
HomeV, Sun
nmer/ Bat
CD
g
0
C/)
S
Si
O
Home NV, W
oT
~n
CQ
CD
~n
O
O
C/)
CD
Si
O
^
of
Tl
Ł=
(Q
CD
o
o
1
C/)
S
Si
O
Home NV, Si
3
3
CD
Tl
(Q
~n
n
f
^>
§
CO
Ł=
CD
Tl
Ł=
CQ
CD
o
o
• S-
CD
Ł
§
Home NV, W
nter /She
sr
S
CD
-n
G)
i
01
o- G
= f
Si
O
Home V, Win
ter/Shelte
t/5
o
^
CD
T|
O
o
I
^&^
T E
Si
o
Home NV, Si
mmer/Sh
CD
crT
°
a
CD
i
'I
Si
O
CO
Ł=
inner/She
oT
S
o
-1
CD
C)
}
O
CD
O
To
2) / Sheltersoi
rce FG Ci
It
O O
CD CD
O O
To To
S 1
O -n
O
O
1
1
=
Q)
ff
source FG
o
II
Q) Q)
1 1
= =
Q) Q)
CD" fD
an FG Ci
rGCi
ffl_
(D
I
O
O
o
3
JD
3-139
-------
Figure 3 A-5. Cumulative Distributions of Observed Benzene Levels in Homes with Attached
Garages
Benzene Concentration (ug/m3)
TJ
O
I I
a. 3
i~ =:
3' o
8 D
«> =•
o -•
3 2.
(D -»
(D
U
(/)
Q.
O
O
^
DO
m
S. 3
X N
O O
q o
O O O
DUO
CD
§ J
as
measured ii
Maxmimum
NVsummei
r: __ zj
N) g- jn
^ 1 a
eg 9L^-
3 S _„
"° i
C/) ,—
It
CD. CO
9L
o
3
CD
i; ^®
CD
CD
Q_
5'
JQ.
g_
^
^
laximum i
Q_
o
o
9L
CD
artithmetic i
=i
CD
Q)
Q)
Q_
CQ
CD
O
3
a
o
3
CD
Q)
Q)
cn
Q) ~Z_
CD g.
ffS
cn >,
II
Q) §
o" -13
tq co
— i f\
0 §-
3 of
Q) 15.
— Q)
%'^
3". Q.
&. o
o g
-^ cn
— h
3 Q-
3- a
cr
c±
o
1
S
CD
z
D
m
~n
'/EPA (2006
CD
0)
m
0
O
CO
O
ST
•o.
^
o
^
w'
CD
Q)
Q_
o
DO
CD
i
Q)
CD_
CD
Q)
Q.
Q
CD
g
CD"
^
CD
0)
W
C
i
o
*
CD
Q.
o
O
-------
Figure 3A-6. Comparison of Modeled and Observed Indoor Benzene Concentrations
Batterman et al.
(2005/6)
Benzene Concentration (ug/m3)
CO
O
cn
o
CD
o
oo
o
o
ID
D
a
fi>
V)
O
George et al. (2002)
Isbelletal. (2005),
Home NV, Summer
Isbelletal. (2005),
Home NV, Wnter
Isbelletal. (2005),
Home V, Summer
Isbelletal. (2005),
Home V, Wnter
Schlapia & Morris
(1998)
I
|
I
|
|
,
I
I
I
I
,
I
|
I
Tsai & Weisel (2000)
Weisel - NJDEP/EPA
(2006)
D D D D
2 co ~n CD
0 =T Ł= Q)
oJ CD. (Q. ft
Ł CD * 5
3 I a 1
I " r
3-141
-------
Figure 3A-7. Multiple Scenario Output of Predicted National Average Benzene Exposure
Attributable to Attached Garages
Benzene TWA Concentration (ug/m3)
CO
o
Ol
O
CD
O
CO
O
o
o
TJ
cn
o
CD
o
oo
o
CD
o
o
o
\
1 §•
$ -g
i Q)'
Ł <"
0 3
m o
ffl %
5 "
398) / Direct
k (2006) / Direct
tiiti
(/)(/)(/)(/)(/)(/)(/)(/)(/)
o-o-o-o-o-o-o-cro-
000000000
Q)Q)Q)Q)Q)Q)Q)Q)Q)
NJNJNJNJNJNJNJNJNJ
OOOOOOOOO
OOOOOOOOO
00 3
0 0 0 0 ^^,^,^3
^^^^DCD~nw^
5g>?»S-S<§?^
(Ss(Q0a?t)stS
1 PS it
= 1 8
0
Ht
t/)t/)t/)t/)t/)t/)t/)^^
S"S"S"S"S"S"S"e. <8
S-S-S-0,-0,-0,-0,-^^
Q)Q)Q)Q)Q)Q)Q)55
U) U)
NJNJNJNJNJNJNJ0 0
ooooooo — —
ooooooo-T^-r^
o^o^o^cncncncn^^
o o
IIIIIIIOO
3333333m~n
0000000QjC
| | | < < < < s a
- - - w w w w 3 -1
cccllll™
3333333 =
0 00 1^ ^ ^ ^
-^ -^ -^ D CD ~n w
g> 2 » s S <§ J
^ CQ 0 0 0 Q r-4-
1 ?! If
fl) 0 c
1 8
0
Ht tl
-lOOOrorororo
U)000Q)Q)^)Q)
w. Soo^SSS
Poc3c3c3^^^^
-000^333
5000^)^)^)'u
_^-j^-j^-j^fl) 0) 0) 0)
Orvjrvj|vjoooo
5^?»xxxx
ogtSQrojm
ff3-'!"S3
-------
Figure 3A-8. Average "Summarized" Benzene Exposure Distributions
Benzene TWA Concentration (ug/m3)
en o en
p
k>
p
CO
O
c
3
c_
~
o>
p
en
o
0>
D
^
TrT P*
0>
p
bo
p
CD
-
O
3_
^<
Q)
A O CD m" —
^ Q- Q) ,~ n
O CD. « CO D
•< o s
CQ
CD
0)
2.
0)
I-K
o'
3
5'
D
O 51
D) C
8 3
CD -"
0) z
(/) Q)
(D S-
tt 5'
O =
3 OJ.
I
: O
5- m
= 1
> (/>
w c
g S
3
o
3
S-
D)
CT
(D
I-K
O
I
D)
O
(D
Q.
3-143
-------
Appendix 3B: 8-Hour Ozone Nonattainment
Table 3B-1. 8-Hour Ozone Nonattainment Areas, Counties and Populations (Data is
Current through October 2006 and Population Numbers are from 2000 Census Data)
8-hour Ozone Nonattainment
Area
Albany-Schenectady-Troy Area
Albany-Schenectady-Troy Area
Albany-Schenectady-Troy Area
Albany-Schenectady-Troy Area
Albany-Schenectady-Troy Area
Albany-Schenectady-Troy Area
Albany-Schenectady-Troy Area
Allegan County Area
Allentown-Bethlehem-Easton Area
Allentown-Bethlehem-Easton Area
Allentown-Bethlehem-Easton Area
Altoona Area
Amadorand Calaveras Counties
(Central Mountain Counties) Area
Amadorand Calaveras Counties
(Central Mountain Counties) Area
Atlanta Area
Atlanta Area
Atlanta Area
Atlanta Area
Atlanta Area
Atlanta Area
Atlanta Area
Atlanta Area
Atlanta Area
Atlanta Area
Atlanta Area
Atlanta Area
Atlanta Area
Atlanta Area
State
NY
NY
NY
NY
NY
NY
NY
Ml
PA
PA
PA
PA
CA
CA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
GA
Classification3'15
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
County Name
Albany Co
Greene Co
Montgomery Co
RensselaerCo
Saratoga Co
Schenectady Co
Schoharie Co
Allegan Co
Carbon Co
Lehigh Co
Northampton Co
Blair Co
AmadorCo
Calaveras Co
Barrow Co
Bartow Co
Carroll Co
Cherokee Co
Clayton Co
CobbCo
Coweta Co
De Kalb Co
Douglas Co
Fayette Co
Forsyth Co
Fulton Co
Gwinnett Co
Hall Co
Whole
/Part
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
2000 Cty
Pop
294,565
48,195
49,708
152,538
200,635
146,555
31,582
105,665
58,802
312,090
267,066
129,144
35,100
40,554
46,144
76,019
87,268
141,903
236,517
607,751
89,215
665,865
92,174
91,263
98,407
816,006
588,448
139,277
3-144
-------
Atlanta Area
Atlanta Area
Atlanta Area
Atlanta Area
Atlanta Area
Atlanta Area
Baltimore Area
Baltimore Area
Baltimore Area
Baltimore Area
Baltimore Area
Baltimore Area
Baton Rouge Area
Baton Rouge Area
Baton Rouge Area
Baton Rouge Area
Baton Rouge Area
Beaumont-Port Arthur Area
Beaumont-Port Arthur Area
Beaumont-Port Arthur Area
Benton Harbor Area
Benzie County Area
Berkeley and Jefferson Counties
Area
Berkeley and Jefferson Counties
Area
Boston-Lawrence-Worcester (E.
Mass) Area
Boston-Lawrence-Worcester (E.
Mass) Area
Boston-Lawrence-Worcester (E.
Mass) Area
Boston-Lawrence-Worcester (E.
Mass) Area
GA
GA
GA
GA
GA
GA
MD
MD
MD
MD
MD
MD
LA
LA
LA
LA
LA
TX
TX
TX
Ml
Ml
WV
WV
MA
MA
MA
MA
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart 1
Subpart 1
Subpart 1 - EAC
Subpart 1 - EAC
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Henry Co
Newton Co
Paulding Co
Rockdale Co
Spalding Co
Walton Co
Anne Arundel
Co
Baltimore (City)
Baltimore Co
Carroll Co
Harford Co
Howard Co
Ascension Par
East Baton
Rouge Par
Iberville Par
Livingston Par
West Baton
Rouge Par
Hardin Co
Jefferson Co
Orange Co
Berrien Co
Benzie Co
Berkeley Co
Jefferson Co
Barnstable Co
Bristol Co
Dukes Co
Essex Co
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
119,341
62,001
81,678
70,111
58,417
60,687
489,656
651,154
754,292
150,897
218,590
247,842
76,627
412,852
33,320
91,814
21,601
48,073
252,051
84,966
162,453
15,998
75,905
42,190
222,230
534,678
14,987
723,419
3-145
-------
Boston-Lawrence-Worcester (E.
Mass) Area
Boston-Lawrence-Worcester (E.
Mass) Area
Boston-Lawrence-Worcester (E.
Mass) Area
Boston-Lawrence-Worcester (E.
Mass) Area
Boston-Lawrence-Worcester (E.
Mass) Area
Boston-Lawrence-Worcester (E.
Mass) Area
Boston-Manchester-Portsmouth
(SE) Area
Boston-Manchester-Portsmouth
(SE) Area
Boston-Manchester-Portsmouth
(SE) Area
Boston-Manchester-Portsmouth
(SE) Area
Buffalo-Niagara Falls Area
Buffalo-Niagara Falls Area
Canton-Massillon Area
Cass County Area
Charlotte-Gastonia-Rock Hill Area
Charlotte-Gastonia-Rock Hill Area
Charlotte-Gastonia-Rock Hill Area
Charlotte-Gastonia-Rock Hill Area
Charlotte-Gastonia-Rock Hill Area
Charlotte-Gastonia-Rock Hill Area
Charlotte-Gastonia-Rock Hill Area
Charlotte-Gastonia-Rock Hill Area
Chattanooga Area
Chattanooga Area
Chattanooga Area
Chicago-Gary-Lake County Area
Chicago-Gary-Lake County Area
Chicago-Gary-Lake County Area
Chicago-Gary-Lake County Area
Chicago-Gary-Lake County Area
MA
MA
MA
MA
MA
MA
NH
NH
NH
NH
NY
NY
OH
Ml
NC
NC
NC
NC
NC
NC
NC
SC
GA
TN
TN
IL
IL
IL
IL
IL
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart 1
Subpart 1
Subpart 1
Subpart
2/Marginal
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1 - EAC
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Middlesex Co
Nantucket Co
Norfolk Co
Plymouth Co
Suffolk Co
Worcester Co
Hillsborough Co
Merrimack Co
Rockingham Co
Stratford Co
Erie Co
Niagara Co
Stark Co
Cass Co
Cabarrus Co
Gaston Co
Iredell Co
Lincoln Co
Mecklenburg Co
Rowan Co
Union Co
York Co
Catoosa Co
Hamilton Co
Meigs Co
Cook Co
Du Page Co
Grundy Co
Kane Co
Kendall Co
W
W
W
W
W
W
P
P
P
P
W
W
W
W
W
W
P
W
W
W
W
P
W
W
W
W
W
P
W
P
1,465,396
9,520
650,308
472,822
689,807
750,963
336,518
11,721
266,340
82,134
950,265
219,846
378,098
51,104
131,063
190,365
39,885
63,780
695,454
130,340
123,677
102,000
53,282
307,896
11,086
5,376,741
904,161
6,309
404,119
28,417
3-146
-------
Chicago-Gary-Lake County Area
Chicago-Gary-Lake County Area
Chicago-Gary-Lake County Area
Chicago-Gary-Lake County Area
Chicago-Gary-Lake County Area
Chico Area
Cincinnati-Hamilton Area
Cincinnati-Hamilton Area
Cincinnati-Hamilton Area
Cincinnati-Hamilton Area
Cincinnati-Hamilton Area
Cincinnati-Hamilton Area
Cincinnati-Hamilton Area
Cincinnati-Hamilton Area
Cincinnati-Hamilton Area
Clearfield and Indiana Counties
Area
Clearfield and Indiana Counties
Area
Cleveland-Akron-Lorain Area
Cleveland-Akron-Lorain Area
Cleveland-Akron-Lorain Area
Cleveland-Akron-Lorain Area
Cleveland-Akron-Lorain Area
Cleveland-Akron-Lorain Area
Cleveland-Akron-Lorain Area
Cleveland-Akron-Lorain Area
Columbia Area
Columbia Area
Columbus Area
Columbus Area
Columbus Area
Columbus Area
Columbus Area
Columbus Area
Dallas-Fort Worth Area
Dallas-Fort Worth Area
IL
IL
IL
IN
IN
CA
IN
KY
KY
KY
OH
OH
OH
OH
OH
PA
PA
OH
OH
OH
OH
OH
OH
OH
OH
SC
SC
OH
OH
OH
OH
OH
OH
TX
TX
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart
2/M ode rate
Subpart
2/M ode rate
Lake Co
Me Henry Co
Will Co
Lake Co
Porter Co
Butte Co
Dearborn Co
Boone Co
Campbell Co
Kenton Co
Butler Co
Clermont Co
Clinton Co
Hamilton Co
Warren Co
Clearfield Co
Indiana Co
Ashtabula Co
Cuyahoga Co
Geauga Co
Lake Co
Lorain Co
Medina Co
Portage Co
Summit Co
Lexington Co
Richland Co
Delaware Co
Fairfield Co
Franklin Co
Knox Co
Licking Co
Madison Co
Collin Co
Dallas Co
W
W
W
W
W
W
P
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
P
P
W
W
W
W
W
W
W
W
644,356
260,077
502,266
484,564
146,798
203,171
10,434
85,991
88,616
151,464
332,807
177,977
40,543
845,303
158,383
83,382
89,605
102,728
1,393,978
90,895
227,511
284,664
151,095
152,061
542,899
181,265
313,253
109,989
122,759
1,068,978
54,500
145,491
40,213
491,675
2,218,899
3-147
-------
Dallas-Fort Worth Area
Dallas-Fort Worth Area
Dallas-Fort Worth Area
Dallas-Fort Worth Area
Dallas-Fort Worth Area
Dallas-Fort Worth Area
Dallas-Fort Worth Area
Dayton-Springfield Area
Dayton-Springfield Area
Dayton-Springfield Area
Dayton-Springfield Area
Denver-Boulder-Greeley-Ft. Collins-
Love. Area
Denver-Boulder-Greeley-Ft. Collins-
Love. Area
Denver-Boulder-Greeley-Ft. Collins-
Love. Area
Denver-Boulder-Greeley-Ft. Collins-
Love. Area
Denver-Boulder-Greeley-Ft. Collins-
Love. Area
Denver-Boulder-Greeley-Ft. Collins-
Love. Area
Denver-Boulder-Greeley-Ft. Collins-
Love. Area
Denver-Boulder-Greeley-Ft. Collins-
Love. Area
Denver-Boulder-Greeley-Ft. Collins-
Love. Area
Detroit-Ann Arbor Area
Detroit-Ann Arbor Area
Detroit-Ann Arbor Area
Detroit-Ann Arbor Area
Detroit-Ann Arbor Area
Detroit-Ann Arbor Area
Detroit-Ann Arbor Area
Detroit-Ann Arbor Area
Door County Area
Erie Area
TX
TX
TX
TX
TX
TX
TX
OH
OH
OH
OH
CO
CO
CO
CO
CO
CO
CO
CO
CO
Ml
Ml
Ml
Ml
Ml
Ml
Ml
Ml
Wl
PA
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1 - EAC
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart 1
Subpart 1
Denton Co
Ellis Co
Johnson Co
Kaufman Co
Parker Co
Rockwall Co
Tarrant Co
Clark Co
Greene Co
Miami Co
Montgomery Co
Adams Co
Arapahoe Co
Boulder Co
Broomfield Co
Denver Co
Douglas Co
Jefferson Co
Larimer Co
Weld Co
Lenawee Co
Livingston Co
Macomb Co
Monroe Co
Oakland Co
St Clair Co
Washtenaw Co
Wayne Co
Door Co
Erie Co
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
P
P
W
W
W
W
W
W
W
W
W
W
432,976
111,360
126,811
71,313
88,495
43,080
1,446,219
144,742
147,886
98,868
559,062
348,618
487,967
269,814
38,272
554,636
175,766
525,507
239,000
172,000
98,890
156,951
788,149
145,945
1,194,156
164,235
322,895
2,061,162
27,961
280,843
3-148
-------
Essex County (Whiteface Mtn.)
Area
Fayetteville Area
Flint Area
Flint Area
Fort Wayne Area
Franklin County Area
Frederick County Area
Frederick County Area
Grand Rapids Area
Grand Rapids Area
Greater Connecticut Area
Greater Connecticut Area
Greater Connecticut Area
Greater Connecticut Area
Greater Connecticut Area
Greene County Area
Greensboro-Winston-Salem-High
Point Area
Greensboro-Winston-Salem-High
Point Area
Greensboro-Winston-Salem-High
Point Area
Greensboro-Winston-Salem-High
Point Area
Greensboro-Winston-Salem-High
Point Area
Greensboro-Winston-Salem-High
Point Area
Greensboro-Winston-Salem-High
Point Area
Greensboro-Winston-Salem-High
Point Area
Greenville-Spartanburg-Anderson
Area
Greenville-Spartanburg-Anderson
Area
Greenville-Spartanburg-Anderson
Area
Hancock, Knox, Lincoln and Waldo
NY
NC
Ml
Ml
IN
PA
VA
VA
Ml
Ml
CT
CT
CT
CT
CT
PA
NC
NC
NC
NC
NC
NC
NC
NC
SC
SC
SC
ME
Subpart 1
Subpart 1 - EAC
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1
Subpart 1
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart 1
Subpart
2/Marginal -
EAC
Subpart
2/Marginal -
EAC
Subpart
2/Marginal -
EAC
Subpart
2/Marginal -
EAC
Subpart
2/Marginal -
EAC
Subpart
2/Marginal -
EAC
Subpart
2/Marginal -
EAC
Subpart
2/Marginal -
EAC
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1
Essex Co
Cumberland Co
Genesee Co
Lapeer Co
Allen Co
Franklin Co
Frederick Co
Winchester
Kent Co
Ottawa Co
Hartford Co
Litchfield Co
New London Co
Tolland Co
Windham Co
Greene Co
Alamance Co
Caswell Co
Davidson Co
Davie Co
Forsyth Co
Guilford Co
Randolph Co
Rockingham Co
Anderson Co
Greenville Co
Spartanburg Co
Hancock Co
P
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
P
1,000
302,963
436,141
87,904
331,849
129,313
59,209
23,585
574,335
238,314
857,183
182,193
259,088
136,364
109,091
40,672
130,800
23,501
147,246
34,835
306,067
421,048
130,454
91,928
165,740
379,616
253,791
29,805
3-149
-------
Counties (Central Maine Coast)
Area
Hancock, Knox, Lincoln and Waldo
Counties (Central Maine Coast)
Area
Hancock, Knox, Lincoln and Waldo
Counties (Central Maine Coast)
Area
Hancock, Knox, Lincoln and Waldo
Counties (Central Maine Coast)
Area
Harrisburg-Lebanon-Carlisle Area
Harrisburg-Lebanon-Carlisle Area
Harrisburg-Lebanon-Carlisle Area
Harrisburg-Lebanon-Carlisle Area
Haywood and Swain Counties
(Great Smoky NP) Area
Haywood and Swain Counties
(Great Smoky NP) Area
Hickory-Morganton-Lenoir Area
Hickory-Morganton-Lenoir Area
Hickory-Morganton-Lenoir Area
Hickory-Morganton-Lenoir Area
Houston-Galveston-Brazoria Area
Houston-Galveston-Brazoria Area
Houston-Galveston-Brazoria Area
Houston-Galveston-Brazoria Area
Houston-Galveston-Brazoria Area
Houston-Galveston-Brazoria Area
Houston-Galveston-Brazoria Area
Houston-Galveston-Brazoria Area
Huntington-Ashland Area
Huron County Area
Imperial County Area
Indianapolis Area
Indianapolis Area
Indianapolis Area
Indianapolis Area
Indianapolis Area
Indianapolis Area
Indianapolis Area
Indianapolis Area
Indianapolis Area
ME
ME
ME
PA
PA
PA
PA
NC
NC
NC
NC
NC
NC
TX
TX
TX
TX
TX
TX
TX
TX
KY
Ml
CA
IN
IN
IN
IN
IN
IN
IN
IN
IN
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1 - EAC
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart 1
Subpart 1
Subpart
2/Marginal
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Knox Co
Lincoln Co
Waldo Co
Cumberland Co
Dauphin Co
Lebanon Co
Perry Co
Haywood Co
Swain Co
Alexander Co
Burke Co
Caldwell Co
Catawba Co
Brazoria Co
Chambers Co
Fort Bend Co
Galveston Co
Harris Co
Liberty Co
Montgomery Co
Waller Co
Boyd Co
Huron Co
Imperial Co
Boone Co
Hamilton Co
Hancock Co
Hendricks Co
Johnson Co
Madison Co
Marion Co
Morgan Co
Shelby Co
P
P
P
W
W
W
W
P
P
W
P
P
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
33,563
28,504
604
213,674
251,798
120,327
43,602
28
260
33,603
69,970
64,254
141,685
241,767
26,031
354,452
250,158
3,400,578
70,154
293,768
32,663
49,752
36,079
142,361
46,107
182,740
55,391
104,093
115,209
133,358
860,454
66,689
43,445
3-150
-------
Jamestown Area
Jefferson County Area
Johnson City-Kingsport-Bristol Area
Johnson City-Kingsport-Bristol Area
Johnstown Area
Kalamazoo-Battle Creek Area
Kalamazoo-Battle Creek Area
Kalamazoo-Battle Creek Area
Kent and Queen Anne's Counties
Area
Kent and Queen Anne's Counties
Area
Kern County (Eastern Kern) Area
Kewaunee County Area
Knoxville Area
Knoxville Area
Knoxville Area
Knoxville Area
Knoxville Area
Knoxville Area
Knoxville Area
La Porte County Area
Lancaster Area
Lansing-East Lansing Area
Lansing-East Lansing Area
Lansing-East Lansing Area
Las Vegas Area
Lima Area
Los Angeles and San Bernardino
Counties (W Mojave Desert) Area
Los Angeles and San Bernardino
Counties (W Mojave Desert) Area
Los Angeles-South Coast Air Basin
Area
Los Angeles-South Coast Air Basin
Area
Los Angeles-South Coast Air Basin
Area
Los Angeles-South Coast Air Basin
Area
Louisville Area
Louisville Area
Louisville Area
Louisville Area
Louisville Area
Macon Area
Macon Area
Manitowoc County Area
NY
NY
TN
TN
PA
Ml
Ml
Ml
MD
MD
CA
Wl
TN
TN
TN
TN
TN
TN
TN
IN
PA
Ml
Ml
Ml
NV
OH
CA
CA
CA
CA
CA
CA
IN
IN
KY
KY
KY
GA
GA
Wl
Subpart 1
Subpart
2/M ode rate
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart
2/Marginal
Subpart
2/Marginal
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart
2/Marginal
Subpart
2/Marginal
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/Severe 1 7
Subpart
2/Severe 1 7
Subpart
2/Severe 1 7
Subpart
2/Severe 1 7
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Chautauqua Co
Jefferson Co
Hawkins Co
Sullivan Co
Cambria Co
Calhoun Co
Kalamazoo Co
Van Buren Co
Kent Co
Queen Annes
Co
Kern Co
Kewaunee Co
Anderson Co
Blount Co
Cocke Co
Jefferson Co
Knox Co
Loudon Co
Sevier Co
La Porte Co
Lancaster Co
Clinton Co
Eaton Co
Ingham Co
Clark Co
Allen Co
Los Angeles Co
San Bernardino
Co
Los Angeles Co
Orange Co
Riverside Co
San Bernardino
Co
Clark Co
Floyd Co
Bullitt Co
Jefferson Co
Oldham Co
Bibb Co
Monroe Co
Manitowoc Co
W
W
W
W
W
W
W
W
W
W
P
W
W
W
P
W
W
W
W
W
W
W
W
W
P
W
P
P
P
W
P
P
W
W
W
W
W
W
P
W
139,750
111,738
53,563
153,048
152,598
137,985
238,603
76,263
19,197
40,563
99,251
20,187
71,330
105,823
20
44,294
382,032
39,086
71,170
110,106
470,658
64,753
103,655
279,320
1,348,864
108,473
297,058
359,350
9,222,280
2,846,289
1,194,859
1,330,159
96,472
70,823
61,236
693,604
46,178
153,887
50
82,887
3-151
-------
Mariposa and Tuolumne Counties
(Southern Mountain Counties) Area
Mariposa and Tuolumne Counties
(Southern Mountain Counties) Area
Mason County Area
Memphis Area
Memphis Area
Milwaukee-Racine Area
Milwaukee-Racine Area
Milwaukee-Racine Area
Milwaukee-Racine Area
Milwaukee-Racine Area
Milwaukee-Racine Area
Murray County (Chattahoochee Nat
Forest) Area
Muskegon Area
Nashville Area
Nashville Area
Nashville Area
Nashville Area
Nashville Area
Nevada County (Western part) Area
NewYork-N. New Jersey-Long
Island Area
NewYork-N. New Jersey-Long
Island Area
NewYork-N. New Jersey-Long
Island Area
NewYork-N. New Jersey-Long
Island Area
NewYork-N. New Jersey-Long
Island Area
NewYork-N. New Jersey-Long
Island Area
NewYork-N. New Jersey-Long
Island Area
NewYork-N. New Jersey-Long
Island Area
NewYork-N. New Jersey-Long
Island Area
NewYork-N. New Jersey-Long
Island Area
NewYork-N. New Jersey-Long
Island Area
CA
CA
Ml
AR
TN
Wl
Wl
Wl
Wl
Wl
Wl
GA
Ml
TN
TN
TN
TN
TN
CA
CT
CT
CT
NJ
NJ
NJ
NJ
NJ
NJ
NJ
NJ
Subpart 1
Subpart 1
Subpart 1
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart 1
Subpart
2/Marginal
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Mariposa Co
Tuolumne Co
Mason Co
Crittenden Co
Shelby Co
Kenosha Co
Milwaukee Co
Ozaukee Co
Racine Co
Washington Co
Waukesha Co
Murray Co
Muskegon Co
Davidson Co
Rutherford Co
SumnerCo
Williamson Co
Wilson Co
Nevada Co
Fairfield Co
Middlesex Co
New Haven Co
Bergen Co
Essex Co
Hudson Co
Hunterdon Co
Middlesex Co
Monmouth Co
Morris Co
Passaic Co
W
W
W
W
W
W
W
W
W
W
W
P
W
W
W
W
W
W
P
W
W
W
W
W
W
W
W
W
W
W
17,130
54,501
28,274
50,866
897,472
149,577
940,164
82,317
188,831
117,493
360,767
1,000
170,200
569,891
182,023
130,449
126,638
88,809
77,735
882,567
155,071
824,008
884,118
793,633
608,975
121,989
750,162
615,301
470,212
489,049
3-152
-------
NewYork-N. New Jersey-Long
Island Area
NewYork-N. New Jersey-Long
Island Area
NewYork-N. New Jersey-Long
Island Area
NewYork-N. New Jersey-Long
Island Area
NewYork-N. New Jersey-Long
Island Area
NewYork-N. New Jersey-Long
Island Area
NewYork-N. New Jersey-Long
Island Area
NewYork-N. New Jersey-Long
Island Area
NewYork-N. New Jersey-Long
Island Area
NewYork-N. New Jersey-Long
Island Area
NewYork-N. New Jersey-Long
Island Area
NewYork-N. New Jersey-Long
Island Area
NewYork-N. New Jersey-Long
Island Area
Norfolk-Virginia Beach-Newport
News (Hampton Roads) Area
Norfolk-Virginia Beach-Newport
News (Hampton Roads) Area
Norfolk-Virginia Beach-Newport
News (Hampton Roads) Area
Norfolk-Virginia Beach-Newport
News (Hampton Roads) Area
Norfolk-Virginia Beach-Newport
News (Hampton Roads) Area
Norfolk-Virginia Beach-Newport
News (Hampton Roads) Area
Norfolk-Virginia Beach-Newport
News (Hampton Roads) Area
Norfolk-Virginia Beach-Newport
News (Hampton Roads) Area
Norfolk-Virginia Beach-Newport
News (Hampton Roads) Area
Norfolk-Virginia Beach-Newport
News (Hampton Roads) Area
Norfolk-Virginia Beach-Newport
News (Hampton Roads) Area
Norfolk-Virginia Beach-Newport
News (Hampton Roads) Area
Norfolk-Virginia Beach-Newport
News (Hampton Roads) Area
Parkersburg-Marietta Area
Parkersburg-Marietta Area
NJ
NJ
NJ
NJ
NY
NY
NY
NY
NY
NY
NY
NY
NY
VA
VA
VA
VA
VA
VA
VA
VA
VA
VA
VA
VA
VA
OH
WV
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart 1
Subpart 1
Somerset Co
Sussex Co
Union Co
Warren Co
Bronx Co
Kings Co
Nassau Co
New York Co
Queens Co
Richmond Co
Rockland Co
Suffolk Co
Westchester Co
Chesapeake
Gloucester Co
Hampton
Isle Of Wight Co
James City Co
Newport News
Norfolk
Poquoson
Portsmouth
Suffolk
Virginia Beach
Williamsburg
York Co
Washington Co
Wood Co
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
297,490
144,166
522,541
102,437
1,332,650
2,465,326
1,334,544
1,537,195
2,229,379
443,728
286,753
1,419,369
923,459
199,184
34,780
146,437
29,728
48,102
180,150
234,403
1 1 ,566
100,565
63,677
425,257
11,998
56,297
63,251
87,986
3-153
-------
Philadelphia-Wilmington-Atlantic
City Area
Philadelphia-Wilmington-Atlantic
City Area
Philadelphia-Wilmington-Atlantic
City Area
Philadelphia-Wilmington-Atlantic
City Area
Philadelphia-Wilmington-Atlantic
City Area
Philadelphia-Wilmington-Atlantic
City Area
Philadelphia-Wilmington-Atlantic
City Area
Philadelphia-Wilmington-Atlantic
City Area
Philadelphia-Wilmington-Atlantic
City Area
Philadelphia-Wilmington-Atlantic
City Area
Philadelphia-Wilmington-Atlantic
City Area
Philadelphia-Wilmington-Atlantic
City Area
Philadelphia-Wilmington-Atlantic
City Area
Philadelphia-Wilmington-Atlantic
City Area
Philadelphia-Wilmington-Atlantic
City Area
Philadelphia-Wilmington-Atlantic
City Area
Philadelphia-Wilmington-Atlantic
City Area
Philadelphia-Wilmington-Atlantic
City Area
Phoenix-Mesa Area
Phoenix-Mesa Area
Pittsburgh-Beaver Valley Area
Pittsburgh-Beaver Valley Area
Pittsburgh-Beaver Valley Area
Pittsburgh-Beaver Valley Area
Pittsburgh-Beaver Valley Area
Pittsburgh-Beaver Valley Area
Pittsburgh-Beaver Valley Area
Portland Area
Portland Area
Portland Area
Portland Area
DE
DE
DE
MD
NJ
NJ
NJ
NJ
NJ
NJ
NJ
NJ
NJ
PA
PA
PA
PA
PA
AZ
AZ
PA
PA
PA
PA
PA
PA
PA
ME
ME
ME
ME
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
Kent Co
New Castle Co
Sussex Co
Cecil Co
Atlantic Co
Burlington Co
Camden Co
Cape May Co
Cumberland Co
Gloucester Co
Mercer Co
Ocean Co
Salem Co
Bucks Co
Chester Co
Delaware Co
Montgomery Co
Philadelphia Co
Maricopa Co
Pinal Co
Allegheny Co
Armstrong Co
Beaver Co
Butler Co
Fayette Co
Washington Co
Westmoreland
Co
Androscoggin
Co
Cumberland Co
Sagadahoc Co
York Co
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
P
P
W
W
W
W
W
W
W
P
P
W
P
126,697
500,265
156,638
85,951
252,552
423,394
508,932
102,326
146,438
254,673
350,761
510,916
64,285
597,635
433,501
550,864
750,097
1,517,550
3,054,504
31,541
1,281,666
72,392
181,412
174,083
148,644
202,897
369,993
3,390
252,907
35,214
164,997
3-154
-------
Poughkeepsie Area
Poughkeepsie Area
Poughkeepsie Area
Providence (all of Rl) Area
Providence (all of Rl) Area
Providence (all of Rl) Area
Providence (all of Rl) Area
Providence (all of Rl) Area
Raleigh-Durham-Chapel Hill Area
Raleigh-Durham-Chapel Hill Area
Raleigh-Durham-Chapel Hill Area
Raleigh-Durham-Chapel Hill Area
Raleigh-Durham-Chapel Hill Area
Raleigh-Durham-Chapel Hill Area
Raleigh-Durham-Chapel Hill Area
Raleigh-Durham-Chapel Hill Area
Reading Area
Richmond-Petersburg Area
Richmond-Petersburg Area
Richmond-Petersburg Area
Richmond-Petersburg Area
Richmond-Petersburg Area
Richmond-Petersburg Area
Richmond-Petersburg Area
Richmond-Petersburg Area
Richmond-Petersburg Area
Riverside County (Coachella
Valley) Area
Roanoke Area
Roanoke Area
Roanoke Area
Roanoke Area
Rochester Area
Rochester Area
NY
NY
NY
Rl
Rl
Rl
Rl
Rl
NC
NC
NC
NC
NC
NC
NC
NC
PA
VA
VA
VA
VA
VA
VA
VA
VA
VA
CA
VA
VA
VA
VA
NY
NY
2/Marginal
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Serious
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1
Subpart 1
Dutch ess Co
Orange Co
Putnam Co
Bristol Co
Kent Co
Newport Co
Providence Co
Washington Co
Chatham Co
Durham Co
Franklin Co
Granville Co
Johnston Co
Orange Co
Person Co
Wake Co
Berks Co
Charles City Co
Chesterfield Co
Colonial Heights
Hanover Co
Henrico Co
Hopewell
Petersburg
Prince George
Co
Richmond
Riverside Co
Botetourt Co
Roanoke
Roanoke Co
Salem
Genesee Co
Livingston Co
W
W
W
W
W
W
W
W
P
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
P
W
W
W
W
W
W
280,150
341,367
95,745
50,648
167,090
85,433
621,602
123,546
21,320
223,314
47,260
48,498
121,965
118,227
35,623
627,846
373,638
6,926
259,903
16,897
86,320
262,300
22,354
33,740
33,047
197,790
324,750
30,496
94,911
85,778
24,747
60,370
64,328
3-155
-------
Rochester Area
Rochester Area
Rochester Area
Rochester Area
Rocky Mount Area
Rocky Mount Area
Sacramento Metro Area
Sacramento Metro Area
Sacramento Metro Area
Sacramento Metro Area
Sacramento Metro Area
Sacramento Metro Area
San Antonio Area
San Antonio Area
San Antonio Area
San Diego Area
San Francisco Bay Area
San Francisco Bay Area
San Francisco Bay Area
San Francisco Bay Area
San Francisco Bay Area
San Francisco Bay Area
San Francisco Bay Area
San Francisco Bay Area
San Francisco Bay Area
San Joaquin Valley Area
San Joaquin Valley Area
San Joaquin Valley Area
San Joaquin Valley Area
San Joaquin Valley Area
San Joaquin Valley Area
San Joaquin Valley Area
NY
NY
NY
NY
NC
NC
CA
CA
CA
CA
CA
CA
TX
TX
TX
CA
CA
CA
CA
CA
CA
CA
CA
CA
CA
CA
CA
CA
CA
CA
CA
CA
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart
2/Serious
Subpart
2/Serious
Subpart
2/Serious
Subpart
2/Serious
Subpart
2/Serious
Subpart
2/Serious
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1 - EAC
Subpart 1
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Marginal
Subpart
2/Serious
Subpart
2/Serious
Subpart
2/Serious
Subpart
2/Serious
Subpart
2/Serious
Subpart
2/Serious
Subpart
Monroe Co
Ontario Co
Orleans Co
Wayne Co
Edgecombe Co
Nash Co
El Dorado Co
Placer Co
Sacramento Co
Solano Co
Sutter Co
Yolo Co
Bexar Co
Comal Co
Guadalupe Co
San Diego Co
Alameda Co
Contra Costa Co
Marin Co
Napa Co
San Francisco
Co
San Mateo Co
Santa Clara Co
Solano Co
Sonoma Co
Fresno Co
Kern Co
Kings Co
Madera Co
Merced Co
San Joaquin Co
Stanislaus Co
W
W
W
W
W
W
P
P
W
P
P
W
W
W
W
P
W
W
W
W
W
W
W
P
P
W
P
W
W
W
W
W
735,343
100,224
44,171
93,765
55,606
87,420
124,164
239,978
1,223,499
197,034
25,013
168,660
1,392,931
78,021
89,023
2,813,431
1,443,741
948,816
247,289
124,279
776,733
707,161
1,682,585
197,508
413,716
799,407
550,220
129,461
123,109
210,554
563,598
446,997
3-156
-------
San Joaquin Valley Area
Scranton-Wilkes-Barre Area
Scranton-Wilkes-Barre Area
Scranton-Wilkes-Barre Area
Scranton-Wilkes-Barre Area
Sheboygan Area
South Bend-Elkhart Area
South Bend-Elkhart Area
Springfield (W. Mass) Area
Springfield (W. Mass) Area
Springfield (W. Mass) Area
Springfield (W. Mass) Area
St. Louis Area
St. Louis Area
St. Louis Area
St. Louis Area
St. Louis Area
St. Louis Area
St. Louis Area
St. Louis Area
St. Louis Area
State College Area
Steubenville-Weirton Area
Steubenville-Weirton Area
Steubenville-Weirton Area
Sutter County (part) (Sutter Buttes)
Area
Tioga County Area
Toledo Area
Toledo Area
Ventura County (part) Area
Washington Area
Washington Area
Washington Area
CA
PA
PA
PA
PA
Wl
IN
IN
MA
MA
MA
MA
IL
IL
IL
IL
MO
MO
MO
MO
MO
PA
OH
WV
WV
CA
PA
OH
OH
CA
DC
MD
MD
2/Serious
Subpart
2/Serious
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart
2/M ode rate
Subpart 1
Subpart 1
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
Tula re Co
Lackawanna Co
Luzerne Co
Monroe Co
Wyoming Co
Sheboygan Co
Elkhart Co
St Joseph Co
Berkshire Co
Franklin Co
Hampden Co
Hampshire Co
Jersey Co
Madison Co
Monroe Co
St Clair Co
Franklin Co
Jefferson Co
St Charles Co
St Louis
St Louis Co
Centre Co
Jefferson Co
Brooke Co
Hancock Co
Sutter Co
Tioga Co
Lucas Co
Wood Co
Ventura Co
Entire District
Calvert Co
Charles Co
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
P
W
W
W
P
W
W
W
368,021
213,295
319,250
138,687
28,080
112,646
182,791
265,559
134,953
71,535
456,228
152,251
21,668
258,941
27,619
256,082
93,807
198,099
283,883
348,189
1,016,315
135,758
73,894
25,447
32,667
1
41,373
455,054
121,065
753,197
572,059
74,563
120,546
3-157
-------
Washington Area
Washington Area
Washington Area
Washington Area
Washington Area
Washington Area
Washington Area
Washington Area
Washington Area
Washington Area
Washington Area
Washington Area
Washington County (Hagerstown)
Area
Wheeling Area
Wheeling Area
Wheeling Area
York Area
York Area
Youngstown-Warren-Sharon Area
Youngstown-Warren-Sharon Area
Youngstown-Warren-Sharon Area
Youngstown-Warren-Sharon Area
MD
MD
MD
VA
VA
VA
VA
VA
VA
VA
VA
VA
MD
OH
WV
WV
PA
PA
OH
OH
OH
PA
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart
2/M ode rate
Subpart 1 - EAC
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Subpart 1
Frederick Co
Montgomery Co
Prince George's
Co
Alexandria
Arlington Co
Fairfax
Fairfax Co
Falls Church
Loudoun Co
Manassas
Manassas Park
Prince William
Co
Washington Co
Belmont Co
Marshall Co
Ohio Co
Adams Co
York Co
Columbiana Co
Mahoning Co
Trumbull Co
Mercer Co
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
195,277
873,341
801,515
128,283
189,453
21,498
969,749
10,377
169,599
35,135
10,290
280,813
131,923
70,226
35,519
47,427
91,292
381,751
112,075
257,555
225,116
120,293
a) Under the CAA these nonattainment areas are further classified as subpart 1 or subpart 2 (subpart 2 is further
classified as marginal, moderate, serious, severe or extreme) based on their design values. An Early Action
Compact (EAC) area is one that has entered into a compact with the EPA and has agreed to reduce ground level
ozone pollution earlier than the CAA would require in exchange the EPA will defer the effective date of the
nonattainment designation. The severe designation is denoted as severe-15 or severe-17 based on the maximum
attainment date associated with the classification.
b) Boston-Manchester-Portsmouth (SE), NH has the same classification as Boston-Lawrence- Worcester (E. MA),
MA.
3-158
-------
Appendix 3C: PM Nonattainment
Table 3C-1. PMi.s Nonattainment Areas and Populations (data is current through October
2006 and the population numbers are from 2000 census data)
PM2s Nonattainment Area
Atlanta, GA
Baltimore, MD
Birmingham, AL
Canton-Massillon, OH
Charleston, WV
Chattanooga, AL-TN-GA
Chicago-Gary-Lake County, IL-IN
Cincinnati-Hamilton, OH-KY-IN
Cleveland-Akron-Lorain, OH
Columbus, OH
Dayton-Springfield, OH
Detroit-Ann Arbor, Ml
Evansville, IN
Greensboro-Winston Salem-High Point, NC
Harrisburg-Lebanon-Carlisle, PA
Hickory, NC
Huntington-Ashland, WV-KY-OH
Indianapolis, IN
Johnstown, PA
Knoxville, TN
Lancaster, PA
Libby, MT
Liberty-Clairton, PA
Los Angeles-South Coast Air Basin, CA
Louisville, KY-IN
Macon, GA
Martinsburg, WV-Hagerstown, MD
NewYork-N. New Jersey-Long Island, NY-NJ-CT
Parkersburg-Marietta, WV-OH
Philadelphia-Wilmington, PA-NJ-DE
Pittsburgh-Beaver Valley, PA
Reading, PA
Rome, GA
San Joaquin Valley, CA
St. Louis, MO-IL
Steubenville-Weirton, OH-WV
Washington, DC-MD-VA
Wheeling, WV-OH
York, PA
Total
Population
4,231,750
2,512,431
807,612
378,098
251,662
423,809
8,757,808
1,850,975
2,775,447
1,448,503
851,690
4,833,493
277,402
568,294
585,799
141,685
340,776
1,329,185
164,431
599,008
470,658
2,626
21,600
14,593,587
938,905
154,837
207,828
19,802,587
152,912
5,536,911
2,195,054
373,638
90,565
3,191,367
2,486,562
132,008
4,377,935
153,172
381,751
88,394,361
3-159
-------
Table 3C-2. PMi0 Nonattainment Areas and Populations (data is current through March
2006 and the population numbers are from 2000 census data)
PM10 Nonattainment Areas Listed Alphabetically
Classification Number 2000 EPA State
of Population Region
Counties (thousands)
NAA
Ajo (Pima County), AZ
Anthony, NM
BonnerCo (Sandpoint), ID
Butte, MT
Clark Co, NV
Coachella Valley, CA
Columbia Falls, MT
Coso Junction, CA
Douglas (Cochise County), AZ
Eagle River, AK
El Paso Co, TX
Eugene-Springfield, OR
Flathead County; Whitefish and vicinity, MT
Fort Hall Reservation, ID
Hayden/Miami, AZ
Imperial Valley, CA
Juneau, AK
Kalispell, MT
LaGrande, OR
Lake Co, OR
Lame Deer, MT
Lane Co, OR
Libby, MT
Los Angeles South Coast Air Basin, CA
Medford-Ashland, OR
Missoula, MT
Mono Basin, CA
Mun. of Guaynabo, PR
New York Co, NY
Nogales, AZ
Ogden, UT
Owens Valley, CA
Paul Spur, AZ
Phoenix, AZ
Pinehurst, ID
Poison, MT
Portneuf Valley, ID
Rillito, AZ
Ronan, MT
Sacramento Co, CA
Salt Lake Co, UT
San Bernardino Co, CA
San Joaquin Valley, CA
Moderate
Moderate
Moderate
Moderate
Serious
Serious
Moderate
Moderate
Moderate
Moderate
Moderate
Moderate
Moderate
Moderate
Moderate
Serious
Moderate
Moderate
Moderate
Moderate
Moderate
Moderate
Moderate
Serious
Moderate
Moderate
Moderate
Moderate
Moderate
Moderate
Moderate
Serious
Moderate
Serious
Moderate
Moderate
Moderate
Moderate
Moderate
Moderate
Moderate
Moderate
Serious
1
1
1
1
1
1
1
1
1
1
1
1
1
2
2
1
1
1
1
1
1
1
1
4
1
1
1
1
1
1
1
1
1
2
1
1
2
1
1
1
1
1
7
8
3
37
35
1,376
182
4
7
16
195
564
179
5
1
4
120
14
15
12
3
1
3
3
14,594
78
52
0
92
1,537
25
77
7
1
3,112
2
4
66
1
3
1,223
898
199
3,080
9
6
10
8
9
9
8
9
9
10
6
10
8
10
9
9
10
8
10
10
8
10
8
9
10
8
9
2
2
9
8
9
9
9
10
8
10
9
8
9
8
9
9
AZ
NM
ID
MT
NV
CA
MT
CA
AZ
AK
TX
OR
MT
ID
AZ
CA
AK
MT
OR
OR
MT
OR
MT
CA
OR
MT
CA
PR
NY
AZ
UT
CA
AZ
AZ
ID
MT
ID
AZ
MT
CA
UT
CA
CA
3-160
-------
Sanders County (part);Thompson Falls and vicinity,
MT
Sheridan, WY
Shoshone Co, ID
Trona, CA
Utah Co, UT
Washoe Co, NV
Weirton, WV
Yuma, AZ
Moderate
Moderate
Moderate
Moderate
Moderate
Serious
Moderate
Moderate
1
1
1
1
1
1
2
1
1
16
10
4
369
339
15
82
8
8
10
9
8
9
3
9
MT
WY
ID
CA
UT
NV
WV
AZ
51 Total Areas 51 28,674
3-161
-------
Appendix 3D: Visibility Tables
Table 3D-1. List of 156 Mandatory Class I Federal Areas Where Visibility is an Important
Value (As Listed in 40 CFR 81)*
State
Area Name
Acreage
Federal
Land
Manager
Alabama
Alaska
Arizona
Arkansas
California
Sipsey Wilderness Area
Bering Sea Wilderness Area
Denali NP (formerly Mt. McKinley NP)
Simeonof Wilderness Area
Tuxedni Wilderness Area
Chiricahua National Monument Wilderness
Area
Chiricahua Wilderness Area
Galiuro Wilderness Area
Grand Canyon NP
Mazatzal Wilderness Area
Mount Baldy Wilderness Area
Petrified Forest NP
Pine Mountain Wilderness Area
Saguaro Wilderness Area
Sierra Ancha Wilderness Area
Superstition Wilderness Area
Sycamore Canyon Wilderness Area
Caney Creek Wilderness Area
Upper Buffalo Wilderness Area
Agua Tibia Wilderness Area
Caribou Wilderness Area
Cucamonga Wilderness Area
Desolation Wilderness Area
Dome Land Wilderness Area
Emigrant Wilderness Area
Hoover Wilderness Area
John Muir Wilderness Area
Joshua Tree Wilderness Area
Kaiser Wilderness Area
Kings Canyon NP
Lassen Volcanic NP
Lava Beds Wilderness Area
Marble Mountain Wilderness Area
Minarets Wilderness Area
Mokelumme Wilderness Area
Pinnacles Wilderness Area
Point Reyes Wilderness Area
Redwood NP
12,646
41,113
1,949,493
25,141
6,402
9,440
18,000
52,717
1,176,913
205,137
6,975
93,493
20,061
71,400
20,850
124,117
47,757
4,344
9,912
15,934
19,080
9,022
63,469
62,206
104,311
47,916
484,673
429,690
36,300
22,500
459,994
105,800
28,640
213,743
109,484
50,400
12,952
25,370
27,792
USDA-FS
USDI-FWS
USDI-NPS
USDI-FWS
USDI-FWS
USDI-NPS
USDA-FS
USDA-FS
USDI-NPS
USDA-FS
USDA-FS
USDI-NPS
USDA-FS
USDI-FS
USDA-FS
USDA-FS
USDA-FS
USDA-FS
USDA-FS
USDA-FS
USDA-FS
USDA-FS
USDA-FS
USDA-FS
USDA-FS
USDA-FS
USDA-FS
USDI-NPS
USDI-BLM
USDA-FS
USDI-NPS
USDI-NPS
USDI-NPS
USDA-FS
USDA-FS
USDA-FS
USDI-NPS
USDI-NPS
USDI-NPS
3-162
-------
State
Area Name
Acreage
Federal
Land
Manager
Colorado
Florida
Georgia
Hawaii
Idaho
Kentucky
Louisiana
Maine
Michigan
San Gabriel Wilderness Area
San Gorgonio Wilderness Area
San Jacinto Wilderness Area
San Rafael Wilderness Area
Sequoia NP
South Warner Wilderness Area
Thousand Lakes Wilderness Area
Ventana Wilderness Area
Yolla Bolly-Middle Eel Wilderness Area
Yosemite NP
Black Canyon of the Gunnison Wilderness
Area
Eagles Nest Wilderness Area
Flat Tops Wilderness Area
Great Sand Dunes Wilderness Area
La Garita Wilderness Area
Maroon Bells-Snowmass Wilderness Area
Mesa Verde NP
Mount Zirkel Wilderness Area
Rawah Wilderness Area
Rocky Mountain NP
Weminuche Wilderness Area
West Elk Wilderness Area
Chassahowitzka Wilderness Area
Everglades NP
St. Marks Wilderness Area
Cohotta Wilderness Area
Okefenokee Wilderness Area
Wolf Island Wilderness Area
Haleakala NP
Hawaii Volcanoes NP
Craters of the Moon Wilderness Area3
Hells Canyon Wilderness Area
Sawtooth Wilderness Area
Selway-Bitterroot Wilderness Areab
Yellowstone NPC
Mammoth Cave NP
Breton Wilderness Area
Acadia National Park
Moosehorn Wilderness Area
Edmunds Unit
Baring Unit
Isle Royale NP
Seney Wilderness Area
36,137
56,722
37,980
20,564
142,722
386,642
68,507
15,695
95,152
111,841
42,000
759,172
11,180
133,910
235,230
33,450
48,486
71,060
51,488
72,472
26,674
263,138
400,907
61,412
23,360
1,397,429
17,745
33,776
343,850
5,126
27,208
217,029
43,243
83,800
216,383
988,770
31,488
51,303
5,000+
37,503
7,501
2,706
4,680
542,428
25,150
USDA-FS
USDA-FS
USDI-BLM
USDA-FS
USDA-FS
USDI-NS
USDA-FS
USDA-FS
USDA-FS
USDA-FS
USDI-BLM
USDI-NPS
USDI-NPS
USDA-FS
USDA-FS
USDI-NPS
USDA-FS
USDA-FS
USDI-NPS
USDA-FS
USDA-FS
USDI-NPS
USDA-FS
USDA-FS
USDI-FWS
USDI-NPS
USDI-FWS
USDA-FS
USDI-FWS
USDI-FWS
USDI-NPS
USDI-NPS
USDI-NPS
USDA-FS
USDA-FS
USDA-FS
USDI-NPS
USDI-NPS
USDI-FWS
USDI-NPS
USDI-FWS
USDI-FWS
USDI-FWS
USDI-NPS
USDI-FWS
3-163
-------
State
Area Name
Acreage
Federal
Land
Manager
Minnesota
Missouri
Montana
Nevada
New Hampshire
New Jersey
New Mexico
North Carolina
North Dakota
Oklahoma
Oregon
Boundary Waters Canoe Area Wilderness
Area
Voyageurs NP
Hercules-Glades Wilderness Area
Mingo Wilderness Area
Anaconda-Pintlar Wilderness Area
Bob Marshall Wilderness Area
Cabinet Mountains Wilderness Area
Gates of the Mtn Wilderness Area
Glacier NP
Medicine Lake Wilderness Area
Mission Mountain Wilderness Area
Red Rock Lakes Wilderness Area
Scapegoat Wilderness Area
Selway-Bitterroot Wilderness Aread
U. L. Bend Wilderness Area
Yellowstone NPe
Jarbidge Wilderness Area
Great Gulf Wilderness Area
Presidential Range-Dry River Wilderness
Area
Brigantine Wilderness Area
Bandelier Wilderness Area
Bosque del Apache Wilderness Area
Carlsbad Caverns NP
Gila Wilderness Area
Pecos Wilderness Area
Salt Creek Wilderness Area
San Pedro Parks Wilderness Area
Wheeler Peak Wilderness Area
White Mountain Wilderness Area
Great Smoky Mountains NPf
Joyce Kilmer-Slickrock Wilderness Area9
Linville Gorge Wilderness Area
Shining Rock Wilderness Area
Swanquarter Wilderness Area
Lostwood Wilderness
Theodore Roosevelt NP
Wichita Mountains Wilderness
Crater Lake NP
Diamond Peak Wilderness
Eagle Cap Wilderness
Gearhart Mountain Wilderness
Hells Canyon Wilderness3
Kalmiopsis Wilderness
747,840
114,964
12,315
8,000
157,803
950,000
94,272
28,562
1,012,599
1 1 ,366
73,877
32,350
239,295
251,930
20,890
167,624
64,667
5,552
20,000
6,603
23,267
80,850
46,435
433,690
167,416
8,500
41,132
6,027
31,171
273,551
10,201
7,575
13,350
9,000
5,557
69,675
8,900
160,290
36,637
293,476
18,709
108,900
22,700
76,900
USDA-FS
USDI-NPS
USDA-FS
USDI-FWS
USDA-FS
USDA-FS
USDA-FS
USDA-FS
USDI-NPS
USDI-FWS
USDA-FS
USDI-FWS
USDA-FS
USDA-FS
USDI-FWS
USDI-NPS
USDA-FS
USDA-FS
USDA-FS
USDI-FWS
USDI-NPS
USDI-FWS
USDI-NPS
USDA-FS
USDA-FS
USDI-FWS
USDA-FS
USDA-FS
USDA-FS
USDI-NPS
USDA-FS
USDA-FS
USDA-FS
USDI-FWS
USDI-FWS
USDI-NPS
USDI-FWS
USDA-NPS
USDA-FS
USDA-FS
USDA-FS
USDA-FS
USDI-BLM
USDA-FS
3-164
-------
State
Area Name
Acreage
Federal
Land
Manager
South Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
Virgin Islands
Virginia
Washington
West Virginia
Wyoming
New Brunswick,
Canada
Mountain Lakes Wilderness
Mount Hood Wilderness
Mount Jefferson Wilderness
Mount Washington Wilderness
Strawberry Mountain Wilderness
Three Sisters Wilderness
Cape Remain Wilderness
Badlands Wilderness
Wind Cave NP
Great Smoky Mountains NPf
Joyce Kilmer-Slickrock Wilderness9
Big Bend NP
Guadalupe Mountains NP
Arches NP
Bryce Canyon NP
Canyonlands NP
Capitol Reef NP
Zion NP
Lye Brook Wilderness
Virgin Islands NP
James River Face Wilderness
Shenandoah NP
Alpine Lakes Wilderness
Glacier Peak Wilderness
Goat Rocks Wilderness
Mount Adams Wilderness
Mount Rainer NP
North Cascades NP
Olympic NP
Pasayten Wilderness
Dolly Sods Wilderness
Otter Creek Wilderness
Bridger Wilderness
Fitzpatrick Wilderness
Grand Teton NP
North Absaroka Wilderness
Teton Wilderness
Washakie Wilderness
Yellowstone NPh
Roosevelt Campobello International Park
23,071
14,160
100,208
46,116
33,003
199,902
28,000
64,250
28,060
241,207
3,832
708,118
76,292
65,098
35,832
337,570
221,896
142,462
12,430
12,295
8,703
190,535
303,508
464,258
82,680
32,356
235,239
503,277
892,578
505,524
10,215
20,000
392,160
191,103
305,504
351,104
557,311
686,584
2,020,625
2,721
USDA-FS
USDA-FS
USDA-FS
USDA-FS
USDA-FS
USDA-FS
USDI-FWS
USDI-NPS
USDI-NPS
USDI-NPS
USDA-FS
USDI-NPS
USDI-NPS
USDI-NPS
USDI-NPS
USDI-NPS
USDI-NPS
USDI-NPS
USDA-FS
USDI-NPS
USDA-FS
USDI-NPS
USDA-FS
USDA-FS
USDA-FS
USDA-FS
USDI-NPS
USDI-NPS
USDI-NPS
USDA-FS
USDA-FS
USDA-FS
USDA-FS
USDA-FS
USDI-NPS
USDA-FS
USDA-FS
USDA-FS
USDI-NPS
i
* U.S. EPA (2001) Visibility in Mandatory Federal Class I Areas (1994-1998): A Report to Congress.
EPA-452/R-01-008. This document is available in Docket EPA-HQ-OAR-2005-0036.
3-165
-------
a) Hells Canyon Wilderness Area, 192,700 acres overall, of which 108,900 acres are in Oregon and
83,800 acres are in Idaho.
b) Selway Bitterroot Wilderness Area, 1,240,700 acres overall, of which 988,700 acres are in Idaho and
251,930 acres are in Montana.
c) Yellowstone National Park, 2,219,737 acres overall, of which 2,020,625 acres are in Wyoming,
167,624 acres are in Montana, and 31,488 acres are in Idaho
d) Selway-Bitterroot Wilderness Area, 1,240,700 acres overall, of which 988,770 acres are in Idaho and
251,930 acres are in Montana.
e) Yellowstone National Park, 2,219,737 acres overall, of which 2,020,625 acres are in Wyoming,
167,624 acres are in Montana, and 31,488 acres are in Idaho.
f) Great Smoky Mountains National Park, 514,758 acres overall, of which 273,551 acres are in North
Carolina, and 241,207 acres are in Tennessee.
g) Joyce Kilmer-Slickrock Wilderness Area, 14,033 acres overall, of which 10,201 acres are in North
Carolina, and 3,832 acres are in Tennessee.
h) Yellowstone National Park, 2,219,737 acres overall, of which 2,020,625 acres are in Wyoming,
167,624 acres are in Montana, and 31,488 acres are in Idaho.
i) Chairman, RCIP Commission.
Abbreviations Used in Table:
USDA-FS: U.S. Department of Agriculture, U.S. Forest Service
USDI-BLM: U.S. Department of Interior, Bureau of Land Management
USDI-FWS: U.S. Department of Interior, Fish and Wildlife Service
USDI-NPS: U.S. Department of Interior, National Park Service
3-166
-------
Table 3D-2. Current (1998-2002) Visibility, Projected (2015) Visibility, and Natural
Background Levels for the 20% Worst Days at 116 IMPROVE Sites
Class I Area Name"
Acadia
Agua Tibia
Alpine Lakes
Anaconda - Pintler
Arches
Badlands
Bandelier
Big Bend
Black Canyon of the Gunnison
Bob Marshall
Boundary Waters Canoe Area
Bridger
Brigantine
Bryce Canyon
Cabinet Mountains
Caney Creek
Canyonlands
Cape Remain
Caribou
Carlsbad Caverns
Chassahowitzka
Chiricahua NM
Chiricahua W
Craters of the Moon
Desolation
Dolly Sods
Dome Land
Eagle Cap
Eagles Nest
Emigrant
Everglades
Fitzpatrick
Flat Tops
Galiuro
Gates of the Mountains
Gila
Glacier
Glacier Peak
Grand Teton
Great Gulf
Great Sand Dunes
Great Smoky Mountains
Guadalupe Mountains
Hells Canyon
Isle Royale
State
ME
CA
WA
MT
UT
SD
NM
TX
CO
MT
MN
WY
NJ
UT
MT
AR
UT
SC
CA
NM
FL
AZ
AZ
ID
CA
WV
CA
OR
CO
CA
FL
WY
CO
AZ
MT
NM
MT
WA
WY
NH
CO
TN
TX
OR
MI
1998-2002 Baseline
Visibility
(deciviews)b
22.7
23.2
18.0
12.3
12.0
17.3
13.2
18.4
11.6
14.2
20.0
11.5
27.6
12.0
13.8
25.9
12.0
25.9
14.8
17.6
25.7
13.9
13.9
14.7
12.9
27.6
20.3
19.6
11.3
17.6
20.3
11.5
11.3
13.9
11.2
13.5
19.5
14.0
12.1
23.2
13.1
29.5
17.6
18.1
21.1
2015 CAIR Control
Case Visibility0
(deciviews)
21.0
23.2
17.4
12.2
12.1
16.8
13.2
18.3
11.4
14.0
19.0
11.3
25.4
11.9
13.4
24.1
12.0
23.9
14.6
17.9
23.0
13.9
13.9
14.7
12.8
23.9
19.9
19.0
11.4
17.4
19.2
11.3
11.4
14.1
10.8
13.5
19.1
13.8
12.0
21.2
13.0
26.1
17.5
18.0
20.1
Natural
Background
(deciviews)
11.5
7.2
7.9
7.3
7.0
7.3
7.0
6.9
7.1
7.4
11.2
7.1
11.3
7.0
7.4
11.3
7.0
11.4
7.3
7.0
11.5
6.9
6.9
7.1
7.1
11.3
7.1
7.3
7.1
7.1
11.2
7.1
7.1
6.9
7.2
7.0
7.6
7.8
7.1
11.3
7.1
11.4
7.0
7.3
11.2
3-167
-------
Class I Area Name"
James River Face
Jarbidge
Joshua Tree
Joyce Kilmer - Slickrock
Kalmiopsis
Kings Canyon
La Garita
Lassen Volcanic
Lava Beds
Linville Gorge
Lostwood
Lye Brook
Mammoth Cave
Marble Mountain
Maroon Bells - Snowmass
Mazatzal
Medicine Lake
Mesa Verde
Mingo
Mission Mountains
Mokelumne
Moosehorn
Mount Hood
Mount Jefferson
Mount Rainier
Mount Washington
Mount Zirkel
North Cascades
Okefenokee
Otter Creek
Pasayten
Petrified Forest
Pine Mountain
Presidential Range - Dry
Rawah
Red Rock Lakes
Redwood
Rocky Mountain
Roosevelt Campobello
Salt Creek
San Gorgonio
San Jacinto
San Pedro Parks
Sawtooth
Scapegoat
Selway - Bitterroot
Seney
Sequoia
Shenandoah
State
VA
NV
CA
NC
OR
CA
CO
CA
CA
NC
ND
VT
KY
CA
CO
AZ
MT
CO
MO
MT
CA
ME
OR
OR
WA
OR
CO
WA
GA
WV
WA
AZ
AZ
NH
CO
WY
CA
CO
ME
NM
CA
CA
NM
ID
MT
MT
MI
CA
VA
1998-2002 Baseline
Visibility
(deciviews)b
28.5
12.6
19.5
29.5
14.8
23.5
11.6
14.8
16.6
27.9
19.6
23.9
30.2
17.1
11.3
13.1
17.7
12.8
27.5
14.2
12.9
21.4
14.0
15.7
18.9
15.7
11.7
14.0
26.4
27.6
14.7
13.5
13.1
23.2
11.7
12.1
16.5
14.1
21.4
17.7
21.5
21.5
11.4
13.6
14.2
12.3
23.8
23.5
27.6
2015 CAIR Control
Case Visibility0
(deciviews)
25.1
12.8
20.3
26.1
14.4
24.1
11.5
14.6
16.5
24.6
18.7
21.1
27.0
16.8
11.3
13.5
17.1
12.8
25.9
14.0
12.8
20.3
13.7
15.2
19.4
15.2
11.8
14.0
24.7
24.0
14.5
13.8
13.4
20.9
11.7
12.1
16.5
14.1
20.1
17.3
22.1
21.4
11.4
13.5
14.1
12.1
22.6
24.1
23.4
Natural
Background
(deciviews)
11.2
7.1
7.1
11.5
7.7
7.1
7.1
7.3
7.5
11.4
7.3
11.3
11.5
7.7
7.1
6.9
7.3
7.1
11.3
7.4
7.1
11.4
7.8
7.8
7.9
7.9
7.1
7.8
11.5
11.3
7.8
7.0
6.9
11.3
7.1
7.1
7.8
7.1
11.4
7.0
7.1
7.1
7.0
7.2
7.3
7.3
11.4
7.1
11.3
3-168
-------
Class I Area Name"
Sierra Ancha
Sipsey
South Warner
Strawberry Mountain
Superstition
Swanquarter
Sycamore Canyon
Teton
Theodore Roosevelt
Thousand Lakes
Three Sisters
UL Bend
Upper Buffalo
Voyageurs
Weminuche
West Elk
Wind Cave
Wolf Island
Yellowstone
YollaBolly- Middle Eel
Yosemite
Zion
State
AZ
AL
CA
OR
AZ
NC
AZ
WY
ND
CA
OR
MT
AR
MN
CO
CO
SD
GA
WY
CA
CA
UT
1998-2002 Baseline
Visibility
(deciviews)b
13.4
28.7
16.6
19.6
14.7
24.6
16.1
12.1
17.6
14.8
15.7
14.7
25.5
18.4
11.6
11.3
16.0
26.4
12.1
17.1
17.6
13.5
2015 CAIR Control
Case Visibility0
(deciviews)
13.7
26.1
16.5
19.2
15.0
21.9
16.6
12.1
16.8
14.6
15.2
14.1
24.3
17.6
11.4
11.3
15.4
24.9
12.1
16.9
17.4
13.3
Natural
Background
(deciviews)
6.9
11.4
7.3
7.5
6.9
11.2
7.0
7.1
7.3
7.3
7.9
7.2
11.3
11.1
7.1
7.1
7.2
11.4
7.1
7.4
7.1
7.0
a) 116 IMPROVE sites represent 155 of the 156 Mandatory Class I Federal Areas. One isolated Mandatory Class I
Federal Area (Bering Sea, an uninhabited and infrequently visited island 200 miles from the coast of Alaska), was
considered to be so remote from electrical power and people that it would be impractical to collect routine aerosol
samples. U.S. EPA (2003) guidance for Tracking Progress Under the Regional Haze Rule. EPA-454/B-03-004.
This document is available in Docket EPA-HQ-OAR-2005-0036.
b) The deciview metric describes perceived visual changes in a linear fashion over its entire range, analogous to the
decibel scale for sound. A deciview of 0 represents pristine conditions. The higher the deciview value, the worse the
visibility, and an improvement in visibility is a decrease in deciview value.
c) The 2015 modeling projections are based on the Clear Air Interstate Rule analyses (EPA, 2005).
3-169
-------
References for Chapter 3
1 U. S. EPA (2003) National Air Quality and Trends Report, 2003 Special Studies Edition.
Office of Air Quality Planning and Standards, Research Triangle Park, NC. Publication No. EPA
454/R-03-005. http://www.epa.gov/air/airtrends/aqtrnd03/. This document is available in Docket
EPA-HQ-OAR-2005-0036.
2 U. S. EPA (2004) Air Toxics Website, http://www.epa.gov/ttn/atw/stprogs.html
3 U. S. EPA (2004) National Air Toxics Monitoring Strategy, Draft. Office of Air Quality
Planning and Standards, Research Triangle Park, NC. January 2004.
http://www.epa.gov/ttn/amtic/files/ambient/airtox/atstratl04.pdf
4 Environmental Protection Agency (2006) National Air Toxics Trends Stations. [Online at
http ://www. epa. gov/ttn/amtic/natts.html]
5 Kenski, D; Koerber, M.; Hafner, H. et al. (2005) Lessons learned from air toxics data. A
national perspective. Environ Manage. June 2005: 19-22.
6 Touma, J.S.; Cox, W.M.; Tikvart, J.A. (2006) Spatial and temporal variability of ambient air
toxics data. J Air & Waste Manage Assoc (in press).
7 Reiss, R. (2006) Temporal trend and weekend-weekday differences for benzene and 1,3-
butadiene in Houston, Texas. Atmos Environ 40: 4711-4724.
8 Harley, R.A.; Hooper, D.S.; Kean, A.J.; Kirchstetter, T.W.; Hesson, J.M.; Balberan, N.T.;
Stevenson, E.D.; Kendall, G.R. (2006) Effects of reformulated gasoline and motor vehicle fleet
turnover on emissions and ambient concentrations of benzene. Environ Sci Technol 40: 5084-
5088.
9 Aleksic, N.; Boynton, G.; Sistla, G.; Perry, J. (2005) Concentrations and trends of benzene in
ambient air over New York State during 1990-2003. Atmos Environ 39: 7894-7905.
10 California Air Resources Board (2005) The California Almanac of Emissions and Air Quality
- 2005 Edition, http://www.arb.ca.gov/aqd/almanac/almanac05/almanac2005all.pdf.
11 Oommen, R.; Hauser, J.; Dayton, D; Brooks, G. (2005) Evaluating HAP trends: A look at
emissions, concentrations, and regulation analyses for selected metropolitan statistical areas.
Presentation at the 14th International Emissions Inventory Conference: Transforming Emission
Inventories Meeting Future Challenges Today. April 12-14, 2005.
12 Clayton, C.A.; Pellizzari, E.D.; Whitmore, R.W.; et al. (1999) National Human Exposure
Assessment Survey (NHEXAS): distributions and associations of lead, arsenic, and volatile
organic compounds in EPA Region 5. J Exposure Analysis Environ Epidemiol 9: 381-392.
13 Gordon, S.M.; Callahan, P.J.; Nishioka, M.G.; et al. (1999) Residential environmental
measurements in the National Human Exposure Assessment Survey (NHEXAS) pilot study in
Arizona: preliminary results for pesticides and VOCs. J Exposure Anal Environ Epidemiol 9:
456-470.
3-170
-------
14 Payne-Sturges, D.C.; Burke, T.A.; Beysse, P.; et al. (2004) Personal exposure meets risk
assessment: a comparison of measured and modeled exposures and risk in an urban community.
Environ Health Perspect 112: 589-598.
15 Weisel, C.P.; Zhang, J.; Turpin, B.J.; et al. (2005) Relationships of Indoor, Outdoor, and
Personal Air (RIOPA). Part I. Collection methods and descriptive analyses. Res Rep Health
Effects Inst 130.
16 Adgate, J.L.; Eberly, L.E.; Stroebel, C.; et al. (2004) Personal, indoor, and outdoor VOC
exposures in a probability sample of children. J Exposure Analysis Environ Epidemiol 14: S4-
S13.
17 Adgate, J.L.; Church, T.R.; Ryan, A.D.; et al. (2004) Outdoor, indoor, and personal exposures
to VOCs in children. Environ Health Perspect 112: 1386-1392.
18 Sexton, K.; Adgate, J.L.; Church, T.R.; et al. (2005) Children's exposure to volatile organic
compounds as determined by longitudinal measurements in blood. Environ Health Persepct 113:
342-349.
19 Kinney, P.L.; Chillrud, S.N.; Ramstrom, S.; et al. (2002) Exposures to multiple air toxics in
New York City. Environ Health Perspect 110 (suppl 4): 539-546.
20 Sax, S.N.; Bennett, D.H.; Chillrud, S.N.; et al. (2004) Differences in source emission rates of
volatile organic compounds in innter-city residences of New York City and Los Angeles. J
Exposure Analysis Environ Epidemiol 14: S95-S109.
21 Sax, S.N.; Bennett, D.J.; Chillrud, S.N.; Ross, J.; Kinney, P.L.; Spengler, J.D. (2006) A cancer
risk assessment of inner-city teenagers living in New York City and Los Angeles. Environ
Health Perspect (in press) doi: 10.1289/ehp.8507. [Online at http://dx.doi.org/1
22 http://www.epa.gov/eogaptil/module3/categorv/category.htm. This document is available in
Docket EPA-HQ-OAR-2005-0036.
23Kittleson, D.; Watts, W.; Johnson, J. (2002) Diesel aerosol sampling methodology.
Coordinating Research Council report E-43. http://www.crcao.com . This document is available
in Docket EPA-HQ-OAR-2005-0036.
24 Zhang, K.M.; Wexler, A.S.; Zhu, Y.F.; et al. (2004) Evolution of particle number distribution
near roadways. Part II: the'Road-to-Ambient'process. Atmos Environ 38: 6655-6665.
25 Zhu, Y.; Hinds, W.C.; Kim, S.; et al. (2002) Study of ultrafine particles near a major highway
with heavy-duty diesel traffic. Atmos Environ 36: 4323-4335.
26 Zhu, Y.; Hinds, W.C.; Kim, S.; Sioutas, C. (2002) Concentration and size distribution of
ultrafine particles near a major highway. J Air & Waste Manage Assoc 52: 1032-1042.
27 Kittelson, D.B.; Watts, W.F.; Johnson, J.P. (2004) Nanoparticle emissions on Minnesota
highways. Atmos Environ 38: 9-19.
28 Reponen, T.; Grinshpun, S.A.; Trakumas, S.; et al. (2003) Concentration gradient patterns of
aerosol particles near interstate highways in the Greater Cincinnati airshed. J Environ Monit 5:
557-562.
3-171
-------
29 Bunn, H.J.; Dinsdale, D.; Smith, T.; Grigg, J. (2001) Ultrafme particles in alveolar
macrophages from normal children. Thorax 56: 932-934.
30Kittelson, D.B.; Watts, W.F.; Johnson, J.P. (2004) Nanoparticle emissions on Minnesota
highways. Atmos Environ 38: 9-19.
31 Westerdahl, D.; Fruin, S.; Sax, T.; Fine, P.M.; Sioutas, C. (2005) Mobile platform
measurements of ultrafme particles and associated pollutant concentrations on freeways and
residential streets in Los Angeles. Atmos Environ 39: 3597-3610.
32Kittelson, D.B.; Watts, W.F.; Johnson, J.P. (2004) Nanoparticle emissions on Minnesota
highways. Atmos Environ 38: 9-19.
33 Sanders, P.O.; Xu, N; Dalka, T.M.; Maricq, M.M. (2003) Airborne brake wear debris: size
distributions, composition, and a comparison of dynamometer and vehicle tests. Environ Sci
Technol 37: 4060-4069.
34 Kamens, R.M.; Jang, M.; Lee, S.; et al. (2003) Secondary organic aerosol formation: some
new and exciting insights. American Geophysical Union 5: 02915.
35Kupiainen, K.J.; Tervahattu, H; Raisanen, M.; et al. (2005) Size and composition of airborne
particles form pavement wear, tires, and traction sanding. Environ Sci Technol 39: 699-706.
36 Sanders, P.O.; Xu, N; Dalka, T.M.; Maricq, M.M. (2003) Airborne brake wear debris: size
distributions, composition, and a comparison of dynamometer and vehicle tests. Environ Sci
Technol 37: 4060-4069.
37 Hitchins, J.; Morawska, L.; Wolff, R.; Gilbert, D. (2000) Concentrations of submicrometre
particles from vehicle emissions near a major road. Atmos Environ 34: 51-59.
38 Zhang, K.M.; Wexler, A.S. (2004) Evolution of particle number distribution near roadways -
Part I: analysis of aerosol dynamics and its implications for engine emission measurement.
Atmos Environ 38: 6643-6653.
39 Janssen, N.A.H.; van Vliet, P.H.N.; Aarts, F.; et al. (2001) Assessment of exposure to traffic
related air pollution of children attending schools near motorways. Atmos Environ 35: 3875-
3884.
40 Roorda-Knape, M.C.; Janssen, N.A.H.; De Hartog, J.J.; et al. (1998) Air pollution from traffic
in city districts near major motorways. Atmos Environ 32: 1921-1930.
41 Kwon, J. (2005) Development of a RIOPA database and evaluation of the effect of proximity
on the potential residential exposure to VOCs from ambient sources. Rutgers, the State
University of New Jersey and University of Medicine and Dentistry of New Jersey. PhD
dissertation. This document is available in Docket EPA-HQ-OAR-2005-0036.
42 Liu, W.; Zhang, J.; Kwon, J.; Weisel, C.; Turpin, B.; Zhang, L.; Korn, L.; Morandi, M.; Stock,
T.; Colome, S. (2006) Concentrations and source characteristics of airborne carbonyl compounds
measured outside urban residences. J Air & Waste Manage Assoc 56: 1196-1204.
43 Skov, H.; Hansen, A.B.; Lorenzen, G.; et al. (2001) Benzene exposure and the effect of traffic
pollution in Copenhagen, Denmark. Atmos Environ 35: 2463-2471.
3-172
-------
44 Jo, W.; Kim, K.; Park, K.; et al. (2003) Comparison of outdoor and indoor mobile source-
related volatile organic compounds between low- and high-floor apartments. Environ Res 92:
166-171.
45 Fischer, P.H.; Joek, G.; van Reeuwijk, H.; et al. (2000) Traffic-related differences in outdoor
and indoor concentrations of particle and volatile organic compounds in Amsterdam. Atmos
Environ 34: 3713-3722.
46 Ilgen, E.; Karfich, N.; Levsen, K.; et al. (2001) Aromatic hydrocarbons in the atmospheric
environment: part I. Indoor versus outdoor sources, the influence of traffic. Atmos Environ 35:
1235-1252.
47 Rodes, C.; Sheldon, L.; Whitaker, D.; et al. (1998) Measuring concentrations of selected air
pollutants inside California vehicles. Final report to California Air Resources Board. Contract
No. 95-339.
48 Sapkota, A.; Buckley, TJ. (2003) The mobile source effect on curbside 1,3-butadiene,
benzene, and particle-bound polycyclic aromatic hydrocarbons assessed at a tollbooth. J Air
Waste Manage Assoc 53: 740-748.
49 Sapkota, A.; Buckley, TJ. (2003) The mobile source effect on curbside 1,3-butadiene,
benzene, and particle-bound polycyclic aromatic hydrocarbons assessed at a tollbooth. J Air
Waste Manage Assoc 53: 740-748.
50 http://www.mde.state.md.us/Programs/AirPrograms/airData/. This document is available in
Docket EPA-HQ-OAR-2005-0036.
51 Ilgen, E.; Karfich, N.; Levsen, K.; et al. (2001) Aromatic hydrocarbons in the atmospheric
environment: part I. Indoor versus outdoor sources, the influence of traffic. Atmos Environ 35:
1235-1252.
52 Hoek G.; Meliefste K.; Cyrys J.; et al. (2002) Spatial variability of fine particle concentrations
in three European areas. Atmos. Environ. 36: 4077-4088.
53 Etyemezian V.; Kuhns H.; Gillies J.; et al. (2003) Vehicle-based road dust emission
measurement (III): effect of speed, traffic volume, location, and season on PM10 road dust
emissions in the Treasure Valley, ID. Atmos. Environ. 37: 4583-4593.
54 Harrison R.M.; Tilling R.; Romero M.S.C.; et al. (2003) A study of trace metals and
polycyclic aromatic hydrocarbons in the roadside environment. Atmos. Environ. 37: 2391-2402.
55
Zhang K.M.; Wexler A.S.; Zhu Y.F.; et al. (2004) Evolution of particle number distribution
near roadways. Part II: the 'Road-to-Ambient' process. Atmos. Environ. 38: 6655-6665.
56 Zhu Y.F.; Hinds W.C.; Shen S.; Sioutas C. (2004) Seasonal trends of concentration and size
distribution of ultrafine particles near major highways in Los Angeles. Aerosol Sci. Technol. 38:
5-13.
57 Zhu, Y.; Hinds, W.C.; Kim, S.; et al. (2002) Study of ultrafine particles near a major highway
with heavy-duty diesel traffic. Atmos Environ 36: 4323-4335.
58
Zhu, Y.; Hinds, W.C.; Kim, S.; Sioutas, C. (2002) Concentration and size distribution of
ultrafine particles near a major highway. J Air & Waste Manage Assoc 52: 1032-1042.
3-173
-------
59 Riediker, M.; Williams, R.; Devlin, R.; et al. (2003) Exposure to particulate matter, volatile
organic compounds, and other air pollutants inside patrol cars. Environ Sci Technol 37: 2084-
2093.
60 Zielinska, B.; Fujita, E.M.; Sagebiel, J.C.; et al. (2002) Interim data report for Section 211(B)
Tier 2 high end exposure screening study of baseline and oxygenated gasoline. Prepared for
American Petroleum Institute. November 19, 2002. This document is available in Docket EPA-
HQ-OAR-2005-0036.
61 Rodes, C.; Sheldon, L.; Whitaker, D.; et al. (1998) Measuring concentrations of selected air
pollutants inside California vehicles. Final report to California Air Resources Board. Contract
No. 95-339.
62 Fitz, D. R.; Winer, A. M.; Colome, S.; et al. (2003) Characterizing the Range of Children's
Pollutant Exposure During School Bus Commutes. Final Report Prepared for the California
Resources Board. This document is available in Docket EPA-HQ-OAR-2005-0036.
63 Sabin, L.D.; Behrentz, E.; Winer, A.M.; et al. (2005) Characterizing the range of children's air
pollutant exposure during school bus commutes. J Expos Anal Environ Epidemiol 15: 377-387.
64 Behrentz, E.; Sabin, L.D.; Winer, A.M.; et al. (2005) Relative importance of school bus-
related microenvironments to children's pollutant exposure. J Air & Waste Manage Assoc 55:
1418-1430.
65 Batterman, S.A.; Peng, C.Y.; and Braun, J. (2002) Levels and composition of volatile organic
compounds on commuting routes in Detroit, Michigan. Atmos Environ 36: 6015-6030.
66 Fruin, S.A.; Winer, A.M.; Rodes, C.E. (2004) Black carbon concentrations in California
vehicles and estimation of in-vehicle diesel exhaust particulate matter exposures. Atmos Environ
38:4123-4133.
67 Adams, H.S.; Nieuwenhuijsen, M.J.; Colvile, R.N. (2001) Determinants of fine particle
(PM2.5) personal exposure levels in transport microenvironments, London, UK. Atmos Environ
35: 4557-4566.
68 Leung, P.-L.; Harrison, R.M. (1999) Roadside and in-vehicle concentrations of monoaromatic
hydrocarbons. Atmos Environ 33: 191-204.
69 Weinhold, B. (2001) Pollutants lurk inside vehicles. Environ Health Perspec 109 (9): A422-
A427.
70 Riediker, M.; Williams, R.; Devlin, R.; et al. (2003) Exposure to particulate matter, volatile
organic compounds, and other air pollutants inside patrol cars. Environ Sci Technol 37: 2084-
2093.
71 Van Wijnen J.H.; Verhoeff A.P.; Jans H.W.A.; Van Bruggen M. (1995) The exposure of
cyclists, car drivers and pedestrians to traffic-related air pollutants. Int Arch Occup Environ
Health 67: 187-193.
72 Chan C.-C.; Ozkaynak H.; Spengler J.D.; Sheldon L. (1991) Driver Exposure to Volatile
Organic Compounds, CO, Ozone, andNO2 under Different Driving Conditions. Environ. Sci.
Technol. 25: 964-972.
3-174
-------
73 Shikiya, D.C., C.S. Liu, M.I. Kahn, et al. (1989) In-vehicle air toxics characterization study
in the south coast air basin. South Coast Air Quality Management District, El Monte, CA. May,
1989. This document is available in Docket EPA-HQ-OAR-2005-0036.
74 Chan C.-C., Spengler J. D., Ozkaynak H., and Lefkopoulou M. (1991) Commuter Exposures to
VOCs in Boston, Massachusetts. J. Air Waste Manage. Assoc. 41, 1594-1600.
75 U.S. EPA (2000) Development of microenvironmental factors for the HAPEM4 in support of
the National Air Toxics Assessment (NATA). External Review Draft Report Prepared by ICF
Consulting and TRJ Environmental, Inc. for the U.S. EPA, Office of Air Quality Planning and
Standards, Research Triangle Park, NC. This document is available in Docket EPA-HQ-OAR-
2005-0036.
76 U.S. EPA (2000) Determination of microenvironmental factors for diesel PM. An addendum
to: Development of microenvironmental factors for the HAPEM4 in support of the National Air
Toxics Assessment (NATA). External Review Draft Report Prepared by ICF Consulting and
TRJ Environmental, Inc. for the U.S. EPA, Office of Air Quality Planning and Standards,
Research Triangle Park, NC. This document is available in Docket EPA-HQ-OAR-2005-0036.
77 Personal communication with FACES Investigators Fred Lurmann, Paul Roberts, and
Katharine Hammond. Data is currently being prepared for publication.
78 Kim, J.J.; Smorodinsky, S.; Lipsett, M.; et al. (2004) Traffic-related air pollution near busy
roads. Am J Respir Crit Care Med 170: 520-526.
79 Janssen, N.A.H.; van Vliet, P.H.N.; Aarts, F.; et al. (2001) Assessment of exposure to traffic
related air pollution of children attending schools near motorways. Atmos Environ 35: 3875-
3884.
80 Roorda-Knape, M.C.; Janssen, N.A.H.; De Hartog, J.J.; et al. (1998) Air pollution from traffic
in city districts near major motorways. Atmos Environ 32: 1921-1930.
81 Van Roosbroeck, S.; Wichmann, J.; Janssen, N.A.H.; Hoek, G.; van Wijnen, J.H; Lebret, E.;
Brunekreef, B. (2006) Long-term personal exposure to traffic-related air pollution among school
children, a validation study. Sci Total Environ 368: 565-573.
82
Kinney, P.L.; Chillrud, S.N.; Ramstrom, S.; et al. (2002) Exposures to multiple air toxics in
New York City. Environ Health Perspect 110 (Suppl 4): 539-546.
83
Chatzis, C.; Alexopoulos, E.G.; and Linos, A. (2005) Indoor and outdoor personal exposure to
benzene in Athens, Greece. Sci Total Environ 349: 72-80.
84 Gulliver, J.; Briggs, D.J. (2004) Personal exposure to particulate air pollution in transport
microenvironments. Atmos Environ 38: 1-8.
85 Van Wijnen, J.H.; Verhoeff, A.P.; Jans, H.W.A.; Van Bruggen, M. (1995) The exposure of
cyclists, car drivers and pedestrians to traffic-related air pollutants. Int Arch Occup Environ
Health 67: 187-193.
86 Chan, C.-C.; Ozkaynak, H.; Spengler, J.D.; Sheldon, L. (1991) Driver Exposure to Volatile
Organic Compounds, CO, Ozone, andNO2 under Different Driving Conditions. Environ. Sci.
Technol. 25: 964-972.
3-175
-------
87 Chan C.-C., Spengler J. D., Ozkaynak H., and Lefkopoulou M. (1991) Commuter Exposures
to VOCs in Boston, Massachusetts. J. Air Waste Manage. Assoc. 41, 1594-1600.
88
Duci, A.; Chaloulakou, A.; Spyrellis N. (2003) Exposure to carbon monoxide in the Athens
urban area during commuting. Sci Total Environ 309: 47-58.
89 Gulliver, J.; Briggs, DJ. (2004) Personal exposure to parti culate air pollution in transport
microenvironments. Atmos Environ 38: 1-8.
90 Ashmore, M.R.; Batty, K.; Machin, F.; et al. (2000) Effects of traffic management and
transport mode on the exposure of schoolchildren to carbon monoxide. Environ Monitoring and
Assessment 65: 49-57.
91 U.S. EPA (1997) Exposure factors handbook. This document is available in Docket EPA-HQ-
OAR-2005-0036. http ://cfpub . epa. gov/ncea/cfm/recordi splav . cfm?deid= 1 2464 .
92 Isbell, M.A.; Stolzberg, R.J.; Duffy, L.K. (2005) Indoor climate in interior Alaska:
simultaneous measurement of ventilation, benzene, and toluene in residential indoor air of two
homes. Sci Total Environ 345: 31-40.
93 Batterman, S.; Jia, C.; Hatzivasilis, G.; Godwin, C. (2006) Simultaneous measurement of
ventilation using tracer gas techniques and VOC concentrations in homes, garages and vehicles.
J Environ Monit 8: 249-256.
94 Morris, S.S. "Influence of attached garages on indoor VOC concentrations in Anchorage
homes." Presented at the annual meeting of the Air & Waste Management Association's Pacific
Northwest Section, November 4, 2004. This document is available in Docket EPA-HQ-OAR-
2005-0036.
95 Sheltersource, Inc. (2002) Evaluating Minnesota homes. Final report. Prepared for Minnesota
Department of Commerce. U.S. Department of Energy grant DE-FG45-96R530335. This
document is available in Docket EPA-HQ-OAR-2005-0036.
96 Fugler, D.; Grande, C.; Graham, L. (2002) Attached garages are likely path for pollutants.
ASHRAE IAQ Applications 3(3): 1-4.
97 Emmerich, S.J.; Gorfain, I.E.; Howard-Reed, C. (2003) Air and pollutant transport from
attached garages to residential living spaces - literature review and field tests. In J Ventilation 2:
265-276.
98 Isbell, M.; Gordian, M.E.; Duffy, L. (2002) Winter indoor air pollution in Alaska: identifying
a myth. Environ Pollution 117: 69-75.
99 U.S. EPA (1987) The Total Exposure Assessment Methodology (TEAM) Study: Summary
and Analysis: Volume I. Office of Research and Development, Washington, D.C. June 1987.
EPA Report No. EPA/600/6-87/002a.
100 Wallace, L. (1996) Environmental exposure to benzene: an update. Environ Health Perspect
104 (Suppl6): 1129-1136.
101 Thomas, K. W.; Pellizzari, E. D.; Clayton, C. A.; Perrit, R.; Dietz, R. N.; Goorich, R. W.;
Nelson, W.; Wallace, L. 1993. Temporal Variability of Benzene Exposures in several New
Jersey homes with attached garages or tobacco smoke. J. Expos. Anal. Environ. Epi. 3: 49-73.
3-176
-------
102 Batterman, S.; Hatzivasilis, G.; Jia, C. (2005) Concentrations and emissions of gasoline and
other vapors from residential vehicle garages. Atmos Environ (in press).
103 George, M.; Kaluza, P.; Maxwell, B.; et al. (2002) Indoor air quality & ventilation strategies
in new homes in Alaska. Alaska Building Science Network. [Online at http://www.absn.com/]
This document is available in Docket EPA-HQ-OAR-2005-0036.
104 Isbell, M.A.; Stolzberg, R.J.; Duffy, L.K. (2005) Indoor climate in interior Alaska:
simultaneous measurement of ventilation, benzene and toluene in residential indoor air of two
homes. Sci Total Environ 345: 31-40.
105 Graham, L.A.; Noseworthy, L.; Fugler, D.; et al. (2004) Contribution of vehicle emissions
from an attached garage to residential indoor air pollution levels. J Air & Waste Manage Assoc
54: 563-584.
106 Schlapia, A.; Morris, S.S. (1998) Architectural, behavioral, and environmental factors
associated with VOCs in Anchorage homes. Proceedings of the Air & Waste Management
Association's 94th Annual Conference & Exhibition. Paper 98-A504.
107 Isbell, M.; Ricker, J.; Gordian, M.E.; Duffy, L.K. (1999) Use of biomarkers in an indoor air
study: lack of correlation between aromatic VOCs with respective urinary biomarkers. Sci Total
Environ 241: 151-159.
108 Isbell, M.A.; Stolzberg, R.J.; Duffy, L.K. (2005) Indoor climate in interior Alaska:
simultaneous measurement of ventilation, benzene and toluene in residential indoor air of two
homes. Sci Total Environ 345: 31-40.
109 Gordon, S.M.; Callahan, P.J.; Nishioka, M.G.; et al. (1999) Residential environmental
measurements in the National Human Exposure Assessment Survey (NHEXAS) pilot study in
Arizona: preliminary results for pesticides and VOCs. J Exposure Analysis Environ Epidemiol
9: 456-470.
110 Bonanno, L.J.; Freeman, N.C.G.; Greenberg, M.; Lioy, PJ. (2001) Multivariate analysis on
levels of selected metals, particulate matter, VOC, and household characteristics and activities
from the Midwestern states NHEXAS. Appl Occup Environ Hygiene 16: 859-874.
111 Tsai, P.; Weisel, C.P. (2000) Penetration of evaporative emissions into a home from an M85-
fueled vehicle parked in an attached garage. J Air & Waste Manage Assoc 50: 371-377.
112 Wilson, A.L.; Colome, S.D.; and Tian, Y. (1991) Air toxics microenvironment exposure and
monitoring study. Final Report. Prepared for South Coast Air Quality Management District and
U.S. Environmental Protection Agency. This document is available in Docket EPA-HQ-OAR-
2005-0036.
113 Zielinska, B.; Fujita, E.M.; Sagebiel, J.C.; et al. (2002) Interim data report for Section 211(B)
Tier 2 high end exposure screening study of baseline and oxygenated gasoline. Prepared for
American Petroleum Institute. November 19, 2002. This document is available in Docket EPA-
HQ-OAR-2005-0036.
114 Lee, S.C.; Chan, L.Y.; and Chiu, M.Y. (1999) Indoor and outdoor air quality investigation at
14 public places in Hong Kong. Environ Int 25: 443-450.
3-177
-------
115 Wong, Y.-c.; Sin, D.W.-m.; and Yeung L.L. (2002) Assessment of the air quality in indoor
car parks. Indoor & Built Environment 11: 134-145.
116 Chaloulakou, A.; Duci, A.; Spyrellis, N. (2002) Exposure to carbon monoxide in enclosed
multi-level parking garages in the central Athens urban area. Indoor & Built Environment 11:
191-201.
117 Srivastava, A.; Joseph, A.E.; and Nair, S. (2004) Ambient levels of benzene in Mumbai city.
Int J Environ Health Res 14 (3): 215-222.
118 Schwar, M.; Booker, J.; Tait, L. (1997) Car Park Air Pollution Exposure of Operatives and
the General Public. Clean Air & Environ Protection 27 (5): 129-137
119 Morillo, P.; Dos Santos, S.G.; Santamaria, J.; et al. (1998) A study of the atmospheric
pollution produced by vehicles in car parks in Madrid, Spain. Indoor Built Environ 7: 156-164.
120 Wilson, A.L.; Colome, S.D.; and Tian, Y. (1991) Air toxics microenvironment exposure and
monitoring study, Final Report. Prepared for South Coast Air Quality Management District and
U.S. Environmental Protection Agency. This document is available in Docket EPA-HQ-OAR-
2005-0036.
121 Hartle, R. (1993) Exposure to methyl tert-butyl ether and benzene among service station
attendants and operators. Environ Health Perspect Supplements: 101 (Suppl. 6): 23-26.
122 Northeast States for Coordinated Air Use Management (1999) RFG/MTBE Findings and
Recommendations. August 1999. This document is available in Docket EPA-HQ-OAR-2005-
0036.
123 Vayghani, S.A.; Weisel, C. (1999) The MTBE air concentrations in the cabin of automobiles
while fueling. J Expos Analysis Environ Epidem 9: 261-267.
124 Egeghy, P.P.; Tornero-Velez, R.T.; Rappaport, S.M. (2000) Environmental and biological
monitoring of benzene during self-service automobile refueling. Environ Health Perspect 108:
1195-1202.
125 Verma, O.K.; Johnson, D.M.; Shaw, M.L.; et al. (2001) Benzene and total hydrocarbon
exposures in the downstream petroleum industries. Am Indust Hygiene Assoc J 62: 176-194.
126
Verma, D.K. and des Tombe, K. (2002) Benzene in gasoline and crude oil: occupational and
environmental implications. Am Indust Hygiene Assoc J 63: 225-230.
127
Baldauf, R.; Fortune,C.; Weinstein. J.; et al. (2005) Air contaminant exposures during the
operation of lawn and garden equipment. J Expos Anal Environ Epidemiol 16: 362-370
128 Eriksson, K.; Tjarner, D.; Marqvardsen, I.; et al. (2003) Exposure to benzene, toluene,
xylenes, and total hydrocarbons among snowmobile drivers in Sweden. Chemosphere 50: 1343-
1347.
129 Kado, N.Y.; Kuzmicky, P.A.; and Okamoto, R.A. (2001) Environmental and occupational
exposure to toxic air pollutants from winter snowmobile use in Yellowstone National Park.
Final report to the Yellowstone Park Foundation, Pew Charitable Trusts, and National Park
Service. This document is available in Docket EPA-HQ-OAR-2005-0036.
3-178
-------
130 NESCAUM (2003) Evaluating the occupational and environmental impact of nonroad diesel
equipment in the Northeast. Interim Report June 9, 2003. This document is available in Docket
EPA-HQ-OAR-2005-0036. http://www.nescaum.org/focus-areas/mobile-sources/mobile-
sources-documents.
131 Davis, M.E.; Smith, T.J.; Laden, F.; Hart, I.E.; Ryan, L.M.; Garshick, E. (2006) Modeling
particle exposure in U.S. trucking terminals. Environ Sci Technol 40: 4226-4232.
132 Atkinson, R.; Arey, J.; Hoover, S.; Preston, K. (2005) Atmospheric Chemistry of Gasoline-
Related Emissions: Formation of Pollutants of Potential Concern. Draft Report Prepared for
California Environmental Protection Agency. September 2005. This document is available in
Docket EPA-HQ-OAR-2005-0036.
133 U. S. EPA (2006) National-Scale Air Toxics Assessment for 1999.
http://www.epa.gov/ttn/atw/natal999.
134 U.S. EPA (2000) User's Guide for the Assessment System for Population Exposure
Nationwide (ASPEN, Version 1.1) Model. Office of Air Quality Planning and Standards,
Research Triangle Park, NC, Report No. EPA-454/R-00-017. This document is available in
Docket EPA-HQ-OAR-2005-0036. http://www.epa.gov/scramOO 1/userg/other/aspenug.pdf
135 Rosenbaum, A. (2005) The HAPEM5 User's Guide: Hazardous Air Pollutant Exposure
Model, Version 5. Prepared by ICF, Inc. for U. S. EPA. This document is available in Docket
EPA-HQ-OAR-2005-0036. http://www.epa.gov/ttn/fera/hapem5/hapem5_guide.pdf.
136 U. S. EPA (2005) Risk - Air Toxics Risk Assessment.
http://www.epa.gov/ttn/fera/risk_atoxic.html.
137 U. S. EPA (1993) Motor Vehicle-Related Air Toxics Study. Report No. EPA420-R-93-005.
http://www.epa.gov/otaq/regs/toxics/tox_archive.htm#2.
138 U. S. EPA (2000) Technical Support Document: Control of Hazardous Air Pollutants from
Motor Vehicles and Motor Vehicle Fuels. Office of Transportation and Air Quality. Report No.
EPA-420-R-00-023. http://www.epa.gov/otaq/toxics.htm and at
http://www.epa.gov/otaq/regs/toxics/r00023.pdf
139 Cook, R., Jones, B., Cleland, J. (2004) A Cohort Based Approach for Characterizing Lifetime
Inhalation Cancer Risk from Time-Varying Exposure to Air Toxics from Ambient Sources.
Environmental Progress 23(2): 120-125.
140 Cook, R., Strum, M., Touma, J., et al. 2002. Trends in Mobile Source-Related Ambient
Concentrations of Hazardous Air Pollutants, 1996-2007. SAE Paper No. 2002-01-1274. This
document is available in Docket EPA-HQ-OAR-2005-0036.
141 Cook, R., Strum, M., Touma, J., Palma, T., Thurman, J., Ensley, D., Smith, R. 2006.
Inhalation Exposure and Risk from Mobile Source Air Toxics in Future Years. Journal of
Exposure Science and Environmental Epidemiology, in press. [Advance online publication, 27
September 2006; doi:10.1038/sj.jes. 7500529,
http ://www.nature. com/j es/j ournal/vaop/ncurrent/index.html. ]
142 U.S. EPA. 2007. The HAPEM6 User's Guide. Prepared for Ted Palma, Office of Air
Quality Planning and Standards, Research Triangle Park, NC, by Arlene Rosenbaum and
3-179
-------
Michael Huang, ICF International, January 2007.
http://www.epa.gov/ttn/fera/human_hapem.html. This document is available in Docket EPA-
HQ-OAR-2005-0036.
143 Byun, D. W., Ching, J. K. S. 1999. Science Algorithms of the EPA Models-3 Community
Multiscale Air Quality (CMAQ) Modeling System. U. S. EPA, Office of Research and
Development, Washington, DC. Report No. EPA/600/R-99/030. This document is available in
Docket EPA-HQ-OAR-2005-0036.
144 Luecken, D. J., Hutzell, W. T., Gipson, G. J. 2005. Development and Analysis of air quality
modeling simulations for hazardous air pollutants. Atmospheric Environment. 40: 5087-5096.
145 Seigneur, C., Pun, B., Lohman, K., Wu, S.-Y. 2003. Regional modeling of the atmospheric
fate and transport of benzene and diesel particles. Environ. Sci. Technol. 37: 5236-5246.
146 U. S. EPA. 2004. User's Guide for the Emissions Modeling System for Hazardous Air
Pollutants (EMS-HAP, Version 3.0), Office of Air Quality Planning and Standards, Research
Triangle Park, NC, Report No. EPA-454/B-00-007. This document is available in Docket EPA-
HQ-OAR-2005-0036. http://www.epa.gov/scram001/userg/other/emshapv3ug.pdf.
147 Battelle. 2003. Estimated background concentrations for the National-Scale Air Toxics
Assessment. Prepared for U. S. EPA, Office of Air Quality Planning and Standards. Contract
No. 68-D-02-061. Work Assignment 1-03. This document is available in Docket EPA-HQ-OAR-
2005-0036.
148 U. S. EPA. 1993. Motor Vehicle-Related Air Toxics Study. Office of Mobile Sources, Ann
Arbor, MI. Report No. EPA 420-R-93-005. This document is available in Docket EPA-HQ-
OAR-2005-0036. http://www.epa.gov/otaq/regs/toxics/tox_archive.htm
149 U. S. EPA. 1999. Analysis of the Impacts of Control Programs on Motor Vehicle Toxics
Emissions and Exposure in Urban Areas and Nationwide. Prepared for U. S. EPA, Office of
Transportation and Air Quality, by Sierra Research, Inc., and Radian International
Corporation/Eastern Research Group. Report No. EPA 420 -R-99-029/030. This document is
available in Docket EPA-HQ-OAR-2005-0036.
http://www.epa.gov/otaq/regs/toxics/tox_archive.htm
150 U. S. EPA. 2002. 1996 National-Scale Air Toxics Assessment.
http://www.epa.gov/ttn/atw/nata/
151 Glen, G., Lakkadi, Y., Tippett, J. A., del Valle-Torres M. 1997. Development of
NERL/CHAD: The National Exposure Research Laboratory Consolidated Human Activity
Database. Prepared by ManTech Environmental Technology, Inc. EPA Contract No. 68-D5-
0049. This document is available in Docket EPA-HQ-OAR-2005-0036.
152 Long, T,; Johnson, T. ; Laurensen, J.; Rosenbaum, A. 2004. Development of Penetration and
Proximity Microenvironment Factor Distributions for the HAPEM5 in Support of the 1999
National-Scale Air Toxics Assessment (NATA). Memorandum from TRJ Consulting and ICF
Consulting, Inc. to Ted Palma, U. S. EPA, Office of Air Quality Planning and Standards, RTF,
NC., April 5, 2004. This document is available in Docket EPA-HQ-OAR-2005-0036.
http://www.epa.gov/ttn/fera/human hapem.html
3-180
-------
153 U.S. EPA. 2007. The HAPEM6 User's Guide. Prepared for Ted Palma, Office of Air
Quality Planning and Standards, Research Triangle Park, NC, by Arlene Rosenbaum and
Michael Huang, ICF International, January 2007.
http://www.epa.gov/ttn/fera/human_hapem.html. This document is available in Docket EPA-
HQ-OAR-2005-0036.
154 Cohen, J.; Cook, R.; Bailey, C.R.; Carr, E. (2005) Relationship between motor vehicle
emissions of hazardous pollutants, roadway proximity, and ambient concentrations in Portland,
Oregon. Environ Modelling & Software 20: 7-12.
155 Pratt, G. C.; Wu, C. Y.;Bock, D.; et al. (2004) Comparing air dispersion model predictions
with measured concentrations of VOCs in urban communities. Environ. Sci. Technol. 38: 1949-
1959.
156 U.S. EPA. 2007. The HAPEM6 User's Guide. Prepared for Ted Palma, Office of Air
Quality Planning and Standards, Research Triangle Park, NC, by Arlene Rosenbaum and
Michael Huang, ICF International, January 2007.
http://www.epa.gov/ttn/fera/human_hapem.html. This document is available in Docket EPA-
HQ-OAR-2005-0036.
157 U. S. EPA. 2001. National-Scale Air Toxics Assessment for 1996: Draft for EPA Science
Advisory Board Review. Report No. EPA-453/R-01-003 This document is available in Docket
EPA-HQ-OAR-2005-0036. http://www.epa.gov/ttn/atw/sab/natareport.pdf
158 U. S. EPA. 2001. National-Scale Air Toxics Assessment for 1996: Draft for EPA Science
Advisory Board Review. Report No. EPA-453/R-01-003 This document is available in Docket
EPA-HQ-OAR-2005-0036. http://www.epa.gov/ttn/atw/sab/natareport.pdf
159 Taylor, M. Memorandum: Revised HAP Emission Factors for Stationary Combustion
Turbines, Prepared by Alpha-Gamma Technologies, Inc for Sims Roy, EPA OAQPS ESD
Combustion Group. August, 2003.Docket ID: OAR-2003-0189.
http://docket.epa.gov/edkpub/do/EDKStaffItemDetailView?obiectId=090007d480271237
160 U. S. EPA. 2004. Benefits of the Proposed Inter-State Air Quality Rule. Office of Air
Quality Planning and Standards, Research Triangle Park, North Carolina. Report No. EPA 4527-
03-001. This document is available in Docket EPA-HQ-OAR-2005-0036.
http://www.epa.gov/CAIR/technical.html or http://www.epa.gov/CAIR/pdfs/tsdO 175.pdf
161 U. S. EPA (2003) Estimated Background Concentrations for the National-Scale Air Toxics
Assessment. Emissions, Monitoring, and Analysis Division, Office of Air Quality Planning and
Standards, Research Triangle Park, NC. This document is available in Docket EPA-HQ-OAR-
2005-0036. http ://www. epa.gov/ttn/atw/nata 19997background.html.
162 U. S. EPA (2005) Supplemental Guidance for Assessing Susceptibility from Early-Life
Exposure to Carcinogens. Report No. EPA/630/R-03/003F. This document is available in
Docket EPA-HQ-OAR-2005-0036.
http ://cfpub. epa. gov/ncea/cfm/recordisplay. cfm?deid= 160003
3-181
-------
163 U. S. EPA. 2006. National Scale Modeling of Air Toxics for the Mobile Source Air Toxics
Rule (Proposed); Technical Support Document," Report Number EPA-454/R-06-002.
164
Science Advisory Board. 2001. NATA - Evaluating the National-Scale Air Toxics
Assessment 1996 Data - An SAB Advisory. Report No. EPA-SAB-EC-ADV-02-001
165 Chen, C. W.; Gibb, H. (2003). Procedures for Calculating Cessation Lag. Reg. Toxicol.
Pharmacol. 38: 157-165.
166 U. S. EPA. (2004) Benefits of the Proposed Inter-State Air Quality Rule. Office of Air
Quality Planning and Standards, Research Triangle Park, North Carolina. Report No. EPA 4527-
03-001. This document is available in Docket EPA-HQ-OAR-2005-0036.
http://www.epa.gov/CAIR/technical.html and at http://www.epa.gov/CAIR/pdfs/tsd0175.pdf
167 Touma, J. S.; Isakov, V.; Ching, J.; Seigneur, C. (2006). Air Quality Modeling of Hazardous
Air Pollutants: Current Status and Future Directions. J. Air and Waste Manage. Assoc., 56:547-
558..
168 Isakov, V.; Venkatram, A. (2005) Resolving neighborhood scale in air toxics modeling: a
case study in Wilmington, California. J Air & Waste Manage. Assoc. 56:559-568.
169 Pratt, G. C.; Wu, C. Y.;Bock, D.; et al. (2004) Comparing air dispersion model predictions
with measured concentrations of VOCs in urban communities. Environ. Sci. Technol. 38: 1949-
1959.
170 Kinnee, E.J.; Touma, J.S.; Mason, R.; Thurman, J.; Beidler, A.; Bailey, C.; Cook, R. (2004)
Allocation of Onroad Mobile Emissions to Road Segments for Air Toxics Modeling in Harris
County, Texas. Transport Res. Part D: Transport and Environ. 9:139-150.
171 Cohen, J.; Cook, R.; Bailey, C.R.; Carr, E. (2005) Relationship between motor vehicle
emissions of hazardous pollutants, roadway proximity, and ambient concentrations in Portland,
Oregon. Environ Modelling & Software 20: 7-12.
172 Touma, J. S.; Isakov, V.; Ching, J.; Seigneur, C. (2006). Air Quality Modeling of Hazardous
Air Pollutants: Current Status and Future Directions. J. Air and Waste Manage. Assoc., 56: 547-
558.
173 Cook, R.; Beidler, A.; Touma, J.S.; Strum M. (2006) Preparing Highway Emissions
Inventories for Urban Scale Modeling: A Case Study in Philadelphia. Transportation Res. Part
D: Transport and Environ. 11: 396-407.
174 U.S. EPA. 1996. Air Quality Criteria for Ozone and Related Photochemical Oxidants,
EPA600-P-93-004aF. This document is available in Docket EPA-HQ-OAR-2005-0036.
175 U.S. EPA. Air Quality Criteria for Ozone and Related Photochemical Oxidants (Final). U.S.
Environmental Protection Agency, Washington, DC, EPA/600/R-05/004aF-cF, 2006.
176 U.S. EPA (2007) Review of National Ambient Air Quality Standards for Ozone, Assessment
of Scientific and Technical Information, OAQPS Staff Paper, EPA-452/R-07-003. This
document is available in Docket EPA-HQ-OAR-2005-0036.
177Bates, D.V.; Baker-Anderson, M.; Sizto, R. (1990) Asthma attack periodicity: a study of
3-182
-------
hospital emergency visits in Vancouver. Environ. Res. 51: 51-70.
178Thurston, G.D.; Ito, K.; Kinney, P.L.; Lippmann, M. (1992) A multi-year study of air
pollution and respiratory hospital admissions in three New York State metropolitan areas:
results for 1988 and 1989 summers. J. Exposure Anal. Environ. Epidemiol. 2:429-450.
179Thurston, G.D.; Ito, K.; Hayes, C.G.; Bates, D.V.; Lippmann, M. (1994) Respiratory hospital
admissions and summertime haze air pollution in Toronto, Ontario: consideration of the role of
acid aerosols. Environ. Res. 65: 271-290.
180Lipfert, F.W.; Hammerstrom, T. (1992) Temporal patterns in air pollution and hospital
admissions. Environ. Res. 59: 374-399.
181Burnett, R.T.; Dales, R.E.; Raizenne, M.E.; Krewski, D.; Summers, P.W.; Roberts, G.R.;
Raad-Young, M.; Dann,T.; Brook, J. (1994) Effects of low ambient levels of ozone and sulfates
on the frequency of respiratory admissions to Ontario hospitals. Environ. Res. 65: 172-194.
182U.S. EPA. Air Quality Criteria for Ozone and Related Photochemical Oxidants (Final). U.S.
Environmental Protection Agency, Washington, DC, EPA/600/R-05/004aF-cF, 2006.
183 U.S. EPA. Air Quality Criteria for Ozone and Related Photochemical Oxidants (Final). U.S.
Environmental Protection Agency, Washington, DC, EPA/600/R-05/004aF-cF, 2006.
184Devlin, R. B.; McDonnell, W. F.; Mann, R.; Becker, S.; House, D. E.; Schreinemachers, D.;
Koren, H. S. (1991) Exposure of humans to ambient levels of ozone for 6.6 hours causes cellullar
and biochemical changes in the lung. Am. J. Respir. CellMol. Biol. 4: 72-81.
185Koren, H. S.; Devlin, R. B.; Becker, S.; Perez, R.; McDonnell, W. F. (1991) Time-dependent
changes of markers associated with inflammation in the lungs of humans exposed to ambient
levels of ozone. Toxicol. Pathol. 19: 406-411.
186Koren, H. S.; Devlin, R. B.; Graham, D. E.; Mann, R.; McGee, M. P.; Horstman, D. H.;
Kozumbo, W. J.; Becker, S.; House, D. E.; McDonnell, W. F.; Bromberg, P. A. (1989a) Ozone-
induced inflammation in the lower airways of human subjects. Am. Rev. Respir. Dis. 139: 407-
415.
187Schelegle, E.S.; Siefkin, A.D.; McDonald, RJ. (1991) Time course of ozone-induced
neutrophilia in normal humans. Am. Rev. Respir. Dis. 143:1353-1358.
188U.S. EPA. Air Quality Criteria for Ozone and Related Photochemical Oxidants (Final). U.S.
Environmental Protection Agency, Washington, DC, EPA/600/R-05/004aF-cF, 2006.
189Hodgkin, I.E.; Abbey, D.E.; Euler, G.L.; Magie, A.R. (1984) COPD prevalence in
nonsmokers in high and low photochemical air pollution areas. Chest 86: 830-838.
190Euler, G.L.; Abbey, D.E.; Hodgkin, I.E.; Magie, A.R. (1988) Chronic obstructive pulmonary
disease symptom effects of long-term cumulative exposure to ambient levels of total oxidants
3-183
-------
and nitrogen dioxide in California Seventh-day Adventist residents. Arch. Environ. Health 43:
279-285.
191 Abbey, D.E.; Petersen, F.; Mills, P.K.; Beeson, W.L. (1993) Long-term ambient
concentrations of total suspended particulates, ozone, and sulfur dioxide and respiratory
symptoms in a nonsmoking population. Arch. Environ. Health 48: 33-46.
192U.S. EPA. Review of the National Ambient Air Quality Standards for Ozone: Policy
Assessment of Scientific and Technical Information, OAQPS Staff Paper- Second Draft. EPA-
452/D-05-002.
193 U.S. EPA. Air Quality Criteria for Ozone and Related Photochemical Oxidants (Final). U.S.
Environmental Protection Agency, Washington, DC, EPA/600/R-05/004aF-cF, 2006.
194U.S. EPA. Air Quality Criteria for Ozone and Related Photochemical Oxidants (Final). U.S.
Environmental Protection Agency, Washington, DC, EPA/600/R-05/004aF-cF, 2006.
195Avol, E. L.; Trim, S. C.; Little, D. E.; Spier, C. E.; Smith, M. N.; Peng, R.-C.; Linn, W. S.;
Hackney, J. D.; Gross, K. B.; D'Arcy, J. B.; Gibbons, D.; Higgins, I. T. T. (1990) Ozone
exposure and lung function in children attending a southern California summer camp. Presented
at: 83rd annual meeting and exhibition of the Air & Waste Management Association; June;
Pittsburgh, PA. Pittsburgh, PA: Air & Waste Management Association; paper no. 90-150.3.
196Higgins, I. T. T.; D'Arcy, J. B.; Gibbons, D. L; Avol, E. L.; Gross, K. B. (1990) Effect of
exposures to ambient ozone on ventilatory lung function in children. Am. Rev. Respir. Dis. 141:
1136-1146.
197Raizenne, M. E.; Burnett, R. T.; Stern, B.; Franklin, C. A.; Spengler, J. D. (1989) Acute lung
function responses to ambient acid aerosol exposures in children. Environ. Health Per sped. 79:
179-185.
198Raizenne, M.; Stern, B.; Burnett, R.; Spengler, J. (1987) Acute respiratory function and
transported air pollutants: observational studies. Presented at: 80th annual meeting of the Air
Pollution Control Association; June; New York, NY. Pittsburgh, PA: Air Pollution Control
Association; paper no. 87-32.6.
199 Spektor, D. M.; Lippmann, M. (1991) Health effects of ambient ozone on healthy children at a
summer camp. In: Berglund, R. L.; Lawson, D. R.; McKee, D. J., eds. Tropospheric ozone and
the environment: papers from an international conference; March 1990; Los Angeles, CA.
Pittsburgh, PA: Air & Waste Management Association; pp. 83-89. (A&WMA transaction series
no. TR-19).
200Spektor, D. M.; Thurston, G. D.; Mao, J.; He, D.; Hayes, C.; Lippmann, M. (1991) Effects of
single- and multiday ozone exposures on respiratory function in active normal children. Environ.
Res. 55: 107-122.
201Spektor, D. M.; Lippman, M.; Lioy, P. J.; Thurston, G. D.;s Citak, K.; James, D. J.; Bock, N.;
3-184
-------
Speizer, F. E.; Hayes, C. (1988a) Effects of ambient ozone on respiratory function in active,
normal children. Am. Rev. Respir. Dis. 137: 313-320.
202U.S. EPA. Air Quality Criteria for Ozone and Related Photochemical Oxidants (Final). U.S.
Environmental Protection Agency, Washington, DC, EPA/600/R-05/004aF-cF, 2006.
203Hazucha, M. J.; Folinsbee, L. J.; Seal, E., Jr. (1992) Effects of steady-state and variable ozone
concentration profiles on pulmonary function. Am. Rev. Respir. Dis. 146: 1487-1493.
204Horstman, D.H.; Ball, B.A.; Folinsbee, L.J.; Brown, J.; Gerrity, T. (1995) Comparison of
pulmonary responses of asthmatic and nonasthmatic subjects performing light exercise while
exposed to a low level of ozone. Toxicol. Ind. Health.
205Horstman, D.H.; Folinsbee, L.J.; Ives, P.J.; Abdul-Salaam, S.; McDonnell, W.F. (1990) Ozone
concentration and pulmonary response relationships for 6.6-hour exposures with five hours of
moderate exercise to 0.08, 0.10, and 0.12 ppm. Am. Rev. Respir. Dis. 142: 1158-1163.
206 U.S. EPA. 2005. Technical Support Document for the Final Clean Air Interstate Rule - Air
Quality Modeling. This document is available in Docket EPA-HQ-OAR-2005-0036.
207 U.S. EPA. 2005. Technical Support Document for the Proposed Mobile Source Air Toxics
Rule - Air Quality Modeling. This document is available in Docket EPA-HQ-OAR-2005-0036.
208 U.S. EPA. 2005. Technical Support Document for the Final Clean Air Interstate Rule - Air
Quality Modeling. This document is available in Docket EPA-HQ-OAR-2005-0036.
209 U.S. EPA. 2005. Clean Air Interstate Rule Emissions Inventory Technical Support
Document. Available from: http://www.epa.gov/cair/technical.html. This document is available
in Docket EPA-HQ-OAR-2005-0036.
210 U.S. EPA. 2005. Technical Support Document for the Proposed Mobile Source Air Toxics
Rule - Ozone Air Quality Modeling. This document is available in Docket EPA-HQ-OAR-2005-
0036.
211 Pielke, R.A., W.R. Cotton, R.L. Walko, et al. 1992. "A Comprehensive Meteorological
Modeling System - RAMS." Meteor. Atmos. Phys., Vol. 49, pp. 69-91.
212 U.S. EPA. 1999. The Benefits and Costs of the Clean Air Act, 1990-2010. Prepared for U.S.
Congress by U.S. EPA, Office of Air and Radiation, Office of Policy Analysis and Review,
Washington, DC, November; EPA report no. EPA410-R-99-001. This document is available in
Docket EPA-HQ-OAR-2005-0036.
213 U.S. EPA. Air Quality Criteria for Ozone and Related Photochemical Oxidants (Final). U.S.
Environmental Protection Agency, Washington, DC, EPA/600/R-05/004aF-cF, 2006. This
document is available in Docket EPA-HQ-OAR-2005-0036.
214 Winner, W.E., and CJ. Atkinson. 1986. "Absorption of air pollution by plants, and
consequences for growth." Trends in Ecology and Evolution 1:15-18.
215 U.S. EPA. Air Quality Criteria for Ozone and Related Photochemical Oxidants (Final). U.S.
Environmental Protection Agency, Washington, DC, EPA/600/R-05/004aF-cF, 2006.This
document is available in Docket EPA-HQ-OAR-2005-0036.
3-185
-------
216 Tingey, D.T., and Taylor, G.E. 1982. "Variation in plant response to ozone: a conceptual
model of physiological events." In: Effects of Gaseous Air Pollution in Agriculture and
Horticulture (Unsworth, M.H., Omrod, D.P., eds.) London, UK: Butterworth Scientific, pp. 113-
138.
217 U.S. EPA. Air Quality Criteria for Ozone and Related Photochemical Oxidants (Final). U.S.
Environmental Protection Agency, Washington, DC, EPA/600/R-05/004aF-cF, 2006.This
document is available in Docket EPA-HQ-OAR-2005-0036.
218 U.S. EPA. Air Quality Criteria for Ozone and Related Photochemical Oxidants (Final). U.S.
Environmental Protection Agency, Washington, DC, EPA/600/R-05/004aF-cF, 2006. This
document is available in Docket EPA-HQ-OAR-2005-0036.
219 U.S. EPA. Air Quality Criteria for Ozone and Related Photochemical Oxidants (Final). U.S.
Environmental Protection Agency, Washington, DC, EPA/600/R-05/004aF-cF, 2006. This
document is available in Docket EPA-HQ-OAR-2005-0036.
220 Ollinger, S.V., J.D. Aber and P.B. Reich. 1997. "Simulating ozone effects on forest
productivity: interactions between leaf canopy and stand level processes." Ecological
Applications 7:1237-1251.
221 Winner, W.E., 1994. "Mechanistic analysis of plant responses to air pollution." Ecological
Applications, 4(4):651 -661.
222 U.S. EPA. Air Quality Criteria for Ozone and Related Photochemical Oxidants (Final). U.S.
Environmental Protection Agency, Washington, DC, EPA/600/R-05/004aF-cF, 2006. This
document is available in Docket EPA-HQ-OAR-2005-0036.
223 U U.S. EPA. Air Quality Criteria for Ozone and Related Photochemical Oxidants (Final).
U.S. Environmental Protection Agency, Washington, DC, EPA/600/R-05/004aF-cF, 2006. This
document is available in Docket EPA-HQ-OAR-2005-0036.
224 Fox, S., and R. A. Mickler, eds. 1996. Impact of Air Pollutants on Southern Pine Forests.
Springer-Verlag, NY, Ecol. Studies, Vol. 118, 513 pp.
225 National Acid Precipitation Assessment Program (NAPAP), 1991. National Acid
Precipitation Assessment Program. 1990 Integrated Assessment Report. National Acid
Precipitation Program. Office of the Director, Washington DC.
226 De Steiguer, I, J. Pye, C. Love. 1990. "Air Pollution Damage to U.S. Forests." Journal of
Forestry, Vol 88(8) pp. 17-22.
227 Pye, J.M. 1988. "Impact of ozone on the growth and yield of trees: A review." Journal of
Environmental Quality 17 pp.347-360.
228 U.S. EPA. Air Quality Criteria for Ozone and Related Photochemical Oxidants (Final). U.S.
Environmental Protection Agency, Washington, DC, EPA/600/R-05/004aF-cF, 2006. This
document is available in Docket EPA-HQ-OAR-2005-0036.
229 U.S. EPA. Air Quality Criteria for Ozone and Related Photochemical Oxidants (Final). U.S.
Environmental Protection Agency, Washington, DC, EPA/600/R-05/004aF-cF, 2006. This
document is available in Docket EPA-HQ-OAR-2005-0036.
3-186
-------
230 McBride, J.R., P.R. Miller, and R.D. Laven. 1985. "Effects of oxidant air pollutants on forest
succession in the mixed conifer forest type of southern California." In: Air Pollutants Effects
On Forest Ecosystems, Symposium Proceedings, St. P, 1985, p. 157-167.
231 Miller, P.R., O.C. Taylor, R.G. Wilhour. 1982. Oxidant air pollution effects on a western
coniferous forest ecosystem. Corvallis, OR: U.S. Environmental Protection Agency,
Environmental Research Laboratory (EPA600-D-82-276). This document is available in Docket
EPA-HQ-OAR-2005-0036.
232 U.S. EPA. Air Quality Criteria for Ozone and Related Photochemical Oxidants (Final). U.S.
Environmental Protection Agency, Washington, DC, EPA/600/R-05/004aF-cF, 2006. This
document is available in Docket EPA-HQ-OAR-2005-0036.
233Kopp, R. J.; Vaughn, W. J.; Hazilla, M.; Carson, R. 1985. "Implications of environmental
policy for U.S. agriculture: the case of ambient ozone standards." J. Environ. Manage. 20:321-
331.
234 Adams, R. M.; Hamilton, S. A.; McCarl, B. A. 1986. "The benefits of pollution control: the
case of ozone and U.S. agriculture." Am. J. Agric. Econ. 34: 3-19.
235 Adams, R. M.; Glyer, J. D.; Johnson, S. L.; McCarl, B. A. 1989. "A reassessment of the
economic effects of ozone on U.S. agriculture." JAPCA 39:960-968.
236 Abt Associates, Inc. 1995. Urban ornamental plants: sensitivity to ozone and potential
economic losses. U.S. EPA, Office of Air Quality Planning and Standards, Research Triangle
Park. Under contract to RADIAN Corporation, contract no. 68-D3-0033, WA no. 6. pp. 9-10.
237 U.S. EPA. 2004. Air Quality Criteria for Particulate Matter. EPA/600/P-99/002aF and
EPA/600/P-99/002bF. This document is available in Docket EPA-HQ-OAR-2005-0036.
238 U.S. EPA. 2005. Review of the National Ambient Air Quality Standard for Particulate Matter:
Policy Assessment of Scientific and Technical Information, OAQPS Staff Paper. EPA-452/R-
05-005. This document is available in Docket EPA-HQ-OAR-2005-0036.
239 U.S. EPA 2006. Provisional Assessment of Recent Studies on Health Effects of Particulate
Matter Exposure. EPA/600/R-06/063. This document is available in Docket EPA-HQ-OAR-
2005-0036.
240 Laden F; Neas LM; Dockery DW; et al. 2000. "Association of fine particulate matter from
different sources with daily mortality in six U.S. cities." Environ. Health Per sped. 108(10):941-
947.
241 Schwartz J; Laden F; Zanobetti A. 2002. "The concentration-response relation between
PM(2.5) and daily deaths." Environ. Health Perspect. 110(10): 1025-1029.
242 Janssen NA; Schwartz J; Zanobetti A.; et al. 2002. "Air conditioning and source-specific
particles as modifiers of the effect of PM10 on hospital admissions for heart and lung disease."
Environ Health Perspect. 110(l):43-49.
243 Dockery, DW; Pope, CA, III; Xu, X; et al. 1993. "An association between air pollution and
mortality in six U.S. cities." N. Engl. J. Med. 329:1753-1759.
3-187
-------
244 Pope, CA, III; Burnett, RT; Thun, MJ; Calle, EE; et al. 2002. "Lung cancer, cardiopulmonary
mortality, and long-term exposure to fine particulate air pollution." J. Am. Med. Assoc. 287:
1132-1141.
245 Krewski, D; Burnett, RT; Goldberg, M S; et al. 2000. "Reanalysis of the Harvard Six Cities
study and the American Cancer Society study of particulate air pollution and mortality. A special
report of the Institute's Particle Epidemiology Reanalysis Project." Cambridge, MA: Health
Effects Institute. This document is available in Docket EPA-HQ-OAR-2005-0036.
246 Jerrett, M; Burnett, RT; Ma, R; et al. 2005. "Spatial analysis of air pollution and mortality in
Los Angeles. Epidemiology 16(6):727-736.
247 Kiinzli, N.; Jerrett, M.; Mack, W.J.; et al. 2005. Ambient air pollution and atherosclerosis in
Los Angeles. Environ Health Perspect. 113:201 -206.
248 Riediker, M.; Cascio, W.E.; Griggs, T.R.; et al. 2004. "Particulate matter exposure in cars is
associated with cardiovascular effects in healthy young men." Am JRespir Crit Care Med 169:
934-940.
249 Van Vliet, P.; Knape, M.; de Hartog, J.; Janssen, N.; Harssema, H.; Brunekreef, B. (1997).
Motor vehicle exhaust and chronic respiratory symptoms in children living near freeways. Env.
Research 74: 122-132.
250 Brunekreef, B., Janssen, N.A.H.; de Hartog, J.; Harssema, H.; Knape, M.; van Vliet, P.
(1997). Air pollution from truck traffic and lung function in children living near roadways.
Epidemiology 8:298-303.
251 Kim, J.J.; Smorodinsky, S.; Lipsett, M.; Singer, B.C.; Hodgson, A.T.; Ostro, B (2004).
Traffic-related air pollution near busy roads: The East Bay children's respiratory health study.
Am. J. Respir. Crit. Care Med. 170: 520-526.
252
U.S. EPA. 2006. Regulatory Impact Analysis for the final PM NAAQS. This document is
available in Docket EPA-HQ-OAR-2005-0036.
253 National Research Council, 1993. Protecting Visibility in National Parks and Wilderness
Areas. National Academy of Sciences Committee on Haze in National Parks and Wilderness
Areas. National Academy Press, Washington, DC. This book can be viewed on the National
Academy Press Website at http://www.nap.edu/books/0309048443/html/ and is available in
Docket EPA-HQ-OAR-2005-0036.
254 U.S. EPA (2004) Air Quality Criteria for Particulate Matter (Oct 2004), Volume I Document
No. EPA600/P-99/002aF and Volume II Document No. EPA600/P-99/002bF. This document is
available in Docket EPA-HQ-OAR-2005-0036.
255 U.S. EPA (2005) Review of the National Ambient Air Quality Standard for Particulate
Matter: Policy Assessment of Scientific and Technical Information, OAQPS Staff Paper. EPA-
452/R-05-005. This document is available in Docket EPA-HQ-OAR-2005-0036.
256 U.S. EPA. 2005. Review of the National Ambient Air Quality Standard for Particulate Matter:
Policy Assessment of Scientific and Technical Information, OAQPS Staff Paper. EPA-452/R-
05-005. This document is available in Docket EPA-HQ-OAR-2005-0036.
3-188
-------
257 U.S. EPA. 1993. Effects of the 1990 Clean Air Act Amendments on Visibility in Class I
Areas: An EPA Report to Congress. EPA452-R-93-014.
258
U.S. EPA (2002) Latest Findings on National Air Quality - 2002 Status and Trends. EPA
454/K-03-001. This document is available in Docket EPA-HQ-OAR-2005-0036.
259 National Park Service. Air Quality in the National Parks, Second edition. NFS, Air
Resources Division. D 2266. September 2002. This document is available in Docket EPA-HQ-
OAR-2005-0036.
260 U.S. EPA (2002) Latest Findings on National Air Quality - 2002 Status and Trends. EPA
454/K-03-001. This document is available in Docket EPA-HQ-OAR-2005-0036.
261 U.S. EPA (2005). Technical Support Document for the Final Clean Air Interstate Rule - Air
Quality Modeling. This document is available in Docket EPA-HQ-OAR-2005-0036.
262
U.S. EPA (2000) Deposition of Air Pollutants to the Great Waters: Third Report to Congress.
Office of Air Quality Planning and Standards. EPA-453/R-00-0005.
263 U.S. EPA (2004) National Coastal Condition Report II. Office of Research and Development/
Office of Water. EPA-620/R-03/002.
264 Gao, Y., E.D. Nelson, M.P. Field, et al. 2002. Characterization of atmospheric trace
elements on PM2.5 particulate matter over the New York-New Jersey harbor estuary. Atmos.
Environ. 36: 1077-1086.
265 Kim, G., N. Hussain, J.R. Scudlark, and T.M. Church. 2000. Factors influencing the
atmospheric depositional fluxes of stable Pb, 210Pb, and 7Be into Chesapeake Bay. J. Atmos.
Chem. 36: 65-79.
266 Lu, R., R.P. Turco, K. Stolzenbach, et al. 2003. Dry deposition of airborne trace metals on
the Los Angeles Basin and adjacent coastal waters. J. Geophys. Res. 108(D2, 4074): AAC 11-1
to 11-24.
267 Marvin, C.H., M.N. Charlton, EJ. Reiner, et al. 2002. Surficial sediment contamination in
Lakes Erie and Ontario: A comparative analysis. J. Great Lakes Res. 28(3): 437-450.
268 Smith, W.H. 1991. Air pollution and forest damage." Chemical Engineering News, 69(45):
30-43.
269 Gawel, I.E.; Ahner, B.A.; Friedland, A.J.; and Morel, F.M.M. 1996. Role for heavy metals in
forest decline indicated by phytochelatin measurements. Nature, 381: 64-65.
270 Cotrufo, M.F.; DeSanto, A.V.; Alfani, A.; et al. 1995. Effects of urban heavy metal pollution
on organic matter decomposition in Quercus ilix L. woods. Environmental Pollution, 89: 81-87.
271 Niklinska, M.; Laskowski, R.; Maryanski, M. 1998. Effect of heavy metals and storage time
on two types of forest litter: basal respiration rate and exchangeable metals. Ecotoxicological
Environmental Safety, 41: 8-18.
272 Mason, R.P. and Sullivan, K.A. 1997. Mercury in Lake Michigan. Environmental Science &
Technology, 31: 942-947. (from Delta Report "Atmospheric deposition of toxics to the Great
Lakes")
3-189
-------
273 Landis, M.S. and Keeler, GJ. 2002. Atmospheric mercury deposition to Lake Michigan
during the Lake Michigan Mass Balance Study. Environmental Science & Technology, 21: 4518-
24.
274 U.S. EPA. 2000. EPA453/R-00-005, "Deposition of Air Pollutants to the Great Waters: Third
Report to Congress," Office of Air Quality Planning and Standards, Research Triangle Park,
North Carolina. This document is available in Docket EPA-HQ-OAR-2005-0036.
275 Callender, E. and Rice, K.C. 2000. The urban environmental gradient: Anthropogenic
influences on the spatial and temporal distributions of lead and zinc in sediments. Environmental
Science & Technology, 34: 232-238.
276 Rice, K.C. 1999. Trace element concentrations in streambed sediment across the
conterminous United States. Environmental Science & Technology, 33: 2499-2504.
277 Ely, JC; Neal, CR; Kulpa, CF; et al. 2001. Implications of platinum-group element
accumulation along U.S. roads from catalytic-converter attrition. Environ. Sci. Technol. 35:
3816-3822.
278 U.S. EPA. 1998. EPA454/R-98-014, "Locating and Estimating Air Emissions from Sources
of Poly cyclic Organic Matter," Office of Air Quality Planning and Standards, Research Triangle
Park, North Carolina. This document is available in Docket EPA-HQ-OAR-2005-0036.
279 U.S. EPA. 1998. EPA454/R-98-014, "Locating and Estimating Air Emissions from Sources
of Poly cyclic Organic Matter," Office of Air Quality Planning and Standards, Research Triangle
Park, North Carolina. This document is available in Docket EPA-HQ-OAR-2005-0036.
280 Simcik, M.F.; Eisenreich, S.J.; Golden, K.A.; et al. 1996. Atmospheric loading of polycyclic
aromatic hydrocarbons to Lake Michigan as recorded in the sediments." Environmental Science
and Technology, 30: 3039-3046.
281 Simcik, M.F.; Eisenreich, S.J.; and Lioy, PJ. 1999. Source apportionment and source/sink
relationship of PAHs in the coastal atmosphere of Chicago and Lake Michigan. Atmospheric
Environment, 33: 5071-5079.
282 Arzayus, K.M.; Dickhut, R.M.; and Canuel, E.A. 2001. Fate of atmospherically deposited
polycyclic aromatic hydrocarbons (PAHs) in Chesapeake Bay." Environmental Science &
Technology, 35, 2178-2183.
283 Park, J.S.; Wade, T.L.; and Sweet, S. 2001. Atmospheric distribution of polycyclic aromatic
hydrocarbons and deposition to Galveston Bay, Texas, USA. Atmospheric Environment, 35:
3241-3249.
284 Poor, N.; Tremblay, R.; Kay, H.; et al. 2002. Atmospheric concentrations and dry deposition
rates of polycyclic aromatic hydrocarbons (PAHs) for Tampa Bay, Florida, USA. Atmospheric
Environment 38: 6005-6015.
285 Arzayus, K.M.; Dickhut, R.M.; and Canuel, E.A. 2001. Fate of atmospherically deposited
polycyclic aromatic hydrocarbons (PAHs) in Chesapeake Bay. Environmental Science &
Technology, 35, 2178-2183.
3-190
-------
286 U.S. EPA. 2000. EPA453/R-00-005, "Deposition of Air Pollutants to the Great Waters: Third
Report to Congress," Office of Air Quality Planning and Standards, Research Triangle Park,
North Carolina. This document is available in Docket EPA-HQ-OAR-2005-0036.
287 Van Metre, P.C.; Mahler, B.J.; and Furlong, E.T. 2000. Urban sprawl leaves its PAH
signature." Environmental Science & Technology, 34: 4064-4070.
288 Cousins, IT.; Beck, A.J.; and Jones, K.C. 1999. A review of the processes involved in the
exchange of semi-volatile organic compounds across the air-soil interface. The Science of the
Total Environment, 228: 5-24.
289 Tuhackova, J., Cajthaml, T.; Novak, K.; et al. 2001. Hydrocarbon deposition and soil
microflora as affected by highway traffic. Environmental Pollution, 113: 255-262.
290 U.S. EPA (2005) Review of the National Ambient Air Quality Standard for Particulate
Matter: Policy Assessment of Scientific and Technical Information, OAQPS Staff Paper. EPA-
452/R-05-005. This document is available in Docket EPA-HQ-OAR-2005-0036.
291 Hoek, G.; Brunekreef, B.; Goldbohm, S.; et al. (2002) Association between mortality and
indicators of traffic-related air pollution in the Netherlands: a cohort study. Lancet 360: 1203-
1209.
292 Finkelstein, M.M.; Jerrett, M.; Sears, M.R. (2004) Traffic air pollution and mortality rate
advancement periods. Am. J. Epidemiol. 160: 173-177.
293
Finkelstein, M.M.; Jerrett, M.; Sears, M.R. (2005) Environmental inequality and circulatory
disease mortality gradients. J. Epidemiol. Community Health 59: 481-486.
294 Hoek, G.; Brunekreef, B.; Goldbohm, S.; et al. (2002) Association between mortality and
indicators of traffic-related air pollution in the Netherlands: a cohort study. Lancet 360: 1203-
1209.
295 Gehring, U.; Heinrich, J.; Kramer, U.; Grote, V.; Hochadel, M.; Dugiri, D.; Kraft, M.;
Rauchfuss, K.; Eberwein, H.G.; Wichmann, H-E. (2006) Long-term exposure to ambient air
pollution and cardiopulmonary mortality in women. Epidemiology 17: 545-551.
296 Lipfert, F.W.; Baty, J.D.; Miller, J.P.; Wyzga, R.E. PM2.5 constitutents and related air
quality variables as predictors of survival in a cohort of U.S. military veterans. Inhalation
Toxicol. 18: 645-657.
297 Maheswaran, M.; Elliott, P. (2003) Stroke mortality associated with living near main roads in
England and Wales. A geographical study. Stroke 34: 2776-2780.
298 Roemer, W.H.; van Wijnen, J.H. (2001) Daily mortality and air pollution along busy streets in
Amsterdam, 1987-1998. Epidemiology 12: 649-653.
299 Heinrich, J.; Wichmann, H-E. (2004) Traffic related pollutants in Europe and their effect on
allergic disease. Current Opin. Clinical Epidemiol. 4: 341-348.
300 Ryan, P.H.; LeMasters, G.; Biagnini, J.; et al. (2005) Is it traffic type, volume, or distance?
Wheezing in infants living near truck and bus traffic. J. Allergy Clin. Immunol. 116: 279-284.
3-191
-------
301 Kim, J.J.; Smorodinsky, S.; Lipsett, M.; et al. (2004) Traffic-related air pollution near busy
roads: the East Bay Children's Respiratory Health Study. Am. J. Respir. Crit. Care Med. 170:
520-526.
302 Lin, S.; Munsie, J.P.; Hwang, S-A.; et al. (2002) Childhood asthma hospitalization and
residential exposure to state route traffic. Environ. Res. 88: 73-81.
303 English, P.; Neutra, R.; Scalf, R.; et al. (1999) Examining associations between childhood
asthma and traffic flow using a geographic information system. Environ. Health Perspect. 107:
761-768.
304 Garshick, E.; Laden, F.; Hart, I.E.; Caron, A. (2003) Residence near a major road and
respiratory symptoms in U.S. veterans. Epidemiology 14: 728-736.
305 Heinrich, J.; Wichmann, H-E. (2004) Traffic related pollutant in Europe and their effect on
allergic disease. Current Opin. ClinicalEpidemiol. 4: 341-348.
306 McConnell, R.; Berhane, K.; Yao. L., Jerrett, M.; Lurmann, F.; Gilliland, F.; Kunzli, N.;
Gauderman, J.; Avol, E.; Thomas, D.; Peters, J. (2006) Traffic, susceptibility, and childhood
asthma. Environ. Health Perspect. 114:766-772.
307 Hoffmann, B.; Moebus, S.; Stang, A.; Beck, E.M.; Dragano, N.; Mohlenkamp, S.;
Schermund, A.; Memmesheimer, M.; Mann, K.; Erbel, R.; Jockel, K.H. (2006) Residence close
to high traffic and prevalence of coronary heart disease. Eur. Heart J. 27: 2696-2702.
308 Peters, A.; von Klot, S.; Heier, M.; et al. (2004) Exposure to traffic and the onset of
myocardial infarction. N. Engl. J. Med. 351: 1721-1730.
309 Riediker, M.; Cascio, W.E.; Griggs, T.R..; et al. (2003) Particulate matter exposures in cars is
associated with cardiovascular effects in healthy young men. Am. J. Respir. Crit. Care Med.
169:934-940.
310 Schwartz, J.; Litonjua, L.; Suh, H.; et al. (2005) Traffic related pollution and heart rate
variability in a panel of elderly subjects. Thorax 60: 455-461.
311
Gold, D.R.; Litonjua, A.A.; Zanobetti, A.; et al. (2005) Air pollution and ST-segment
depression in elderly subjects. Environ. Health Perspect. 113: 883-887.
312
Wilhelm, M.; Ritz, B. (2002) Residential proximity to traffic and adverse birth outcomes in
Los Angeles County, California, 1994-1996. Environ. Health Prospect. 111:207-216.
313 Ritz B; Yu F. (1999) The effect of ambient carbon monoxide on low birth weight among
children born in southern California between 1989 and 1993. Environ. Health Perspect. 107:17-
25.
314 Ritz B; Yu F; Capa G; Fruin S. (2000) Effect of air pollution on premature birth among
children born in southern California between 1989 and 1993. Epidemiology 11:502-511.
315 Ritz B; Yu F; Fruin S; et al. (2002) Ambient air pollution and risk of birth defects in Southern
California. Am. J. Epidemiol. 155:17-25.
316 Perera, F.P.; Rauh, V.; Tsai, W-Y.; et al. (2002) Effect of transplacental exposure to
environmental pollutants on birth outcomes in a multiethnic population. Environ. Health
Perspect. 111:201-205.
3-192
-------
317 Perera, P.P.; Rauh, V.; Whyatt, R.M.; Tsai, W.Y.; Tang, D.; Diaz, D.; Hoepner, L.; Barr, D.;
Tu, Y.H.; Camann, D.; Kinney, P. (2006) Effect of prenatal exposure to airborne polycyclic
aromatic hydrocarbons on neurodevelopment in the first 3 years of life among inner-city
children. Environ. Health Perspect. 114: 1287-1292.
318 U. S. EPA (2005) Supplemental Guidance for Assessing Susceptibility from Early-Life
Exposure to Carcinogens. Report No. EPA/630/R-03/003F. This document is available in
Docket EPA-HQ-OAR-2005-0036.
http ://cfpub. epa. gov/ncea/cfm/recordisplay. cfm?deid= 160003
319 U. S. EPA. 2002. Toxicological Review of Benzene (Noncancer effects). Report No. EPA
635/R-02/001-F.
320 Savitz, D.A.; Feingold, L. (1989) Association of childhood cancer with residential traffic
density. Scand. J. Work Environ. Health 15: 360-363.
321 Pearson, R.L.; Wachtel, H.; Ebi, K.L. (2000) Distance-weighted traffic density in proximity
to a home is a risk factor for leukemia and other childhood cancers. J. Air Waste Manage. Assoc.
50: 175-180.
322 Feychting, M.; Svensson, D.; Ahlbom, A. (1998) Exposure to motor vehicle exhaust and
childhood cancer. Scan. J. Work Environ. Health 24(1): 8-11.
323
Langholz, B.; Ebi, K.L.; Thomas, D.C.; et al. (2002) Traffic density and the risk of childhood
leukemia in a Los Angeles case-control study. Ann. Epidemiol. 12: 482-487.
324
Raaschou-Nielsen, O.; Hertel, O.; Thomsen, B.L.; Olsen, J.H. (2001) Air pollution from
traffic at the residence of children with cancer. Am. J. Epidemiol. 153: 433-443.
325 Reynolds, P.; Von Behren, J.; Gunier, R.B.; et al. (2003) Childhood cancer incidence rates
and hazardous air pollutants in California: an exploratory analysis. Environ Health Perspect 111:
663-668.
326 Reynolds, P.; Von Behren, J.; Gunier, R.B.; et al. (2004) Residential exposure to traffic in
California and childhood cancer. Epidemiology 15: 6-12.
327 Knox, E.G. (2005) Oil combustion and childhood cancers. J. Epidemiol. Community Health
59: 755-760.
328 U.S. EPA. 2007. The HAPEM6 User's Guide. Appendix B. Prepared for Ted Palma, Office
of Air Quality Planning and Standards, Research Triangle Park, NC, by Arlene Rosenbaum and
Michael Huang, ICF International, January 2007.
http://www.epa.gov/ttn/fera/human_hapem.html. This document is available in Docket EPA-
HQ-OAR-2005-0036.
329United States Census Bureau. (2004) American Housing Survey web page. [Online at
http://www.census.gov/hhes/www/housing/ahs/ahs03/ahs03.html] Table IA-6.
330This statistic is based on the odds ratio of being near major transportation sources, compared
to not being near transportation sources for housing units located in different geographic regions.
3-193
-------
331Garshick, E.; Laden, F.; Hart, I.E.; Caron, A. (2003) Residence near a major road and
respiratory symptoms in U.S. veterans. Epidemiology 14: 728-736.
332Green, R.S.; Smorodinsky, S.; Kim, J.J.; McLaughlin, R.; Ostro, B. (2004) Proximity of
California public schools to busy roads. Environ. Health Perspect. 112: 61-66.
333Gunier, R.B.; Hertz, A.; Von Behren, I; Reynolds, P. (2003) Traffic density in California:
socioeconomic and ethnic differences among potentially exposed children. J. Expos. Anal.
Environ. Epidemiol. 13: 240-246.
334 Emmerich, S.; Gorfain, J.C.; Huang, M.: Howard-Reed, C. (2003) Air and pollutant transport
from attached garages to residential living spaces. National Institute of Standards and
Technology report number NISTIR 7072.
335 Gordon, S.M.; Callahan, P.J.; Nishioka, M.G; Brinkman, M.C.; O'Rourke, M.K.; Lebowitz,
M.D.; Moschandreas, DJ. (1999) Residential environmental measurements in the National
Human Exposure Assesment Survey (NHEXAS) pilot study in Arizona: preliminary results for
pesticides and VOCs. J. Expos. Analysis. Environ. Epidemiol. 9: 456-470.
336 Schlapia, A.; Morris, S.S. (1998) Architectural, behavioral and environmental factors
associated with VOCs in Anchorage homes. Proceedings of the Air & Waste Management
Association's 91st Annual Meeting & Exhibition. Paper no. 98-A504.
337 Batterman, S.; Hatzivasilis, G.; Jia, C. (2006) Concentrations and emissions of gasoline and
other vapors from residential vehicle garages. Atmos. Environ. 40: 1828-1844.
338 Memorandum from Tom Long, Ted Johnson, Jim Laurenson, and Arlene Rosenbaum to Ted
Palma. Subject: Development of penetration and proximity microenvironment factor
distributions for the HAPEM5 in support of the 1999 National-scale Air Toxics Assessment
(NATA). EPA contract no. GS-10F-0124J.
http://www.epa.gov/ttn/fera/hapem5/hapem5_me_factor_memo.pdf
339
Palisade Corporation. Guide to Using @Risk. Risk Analysis and Simulation Add-In for
Microsoft Excel. Version 4.5. Newfield: Palisade Corporation. Pages i-iv.
340 Batterman, S.; Jia, C.; Hatzivasilis, G.; Godwin, C. (2006) Simultaneous measurement of
ventilation using tracer gas techniques and VOC concentrations in homes, garages and vehicles.
J Environ. Monit. 8: 249-256.
341 Data obtained from Dr. Clifford Weisel, EOHSI. E-mail correspondence dated November 3,
2006 from Dr. Clifford Weisel, EOHSI to Rich Cook and Chad Bailey, U.S. EPA. Docket
number Docket EPA-HQ-OAR-2005-0036.
342 J.L. Schoor, "Chapter 1: Chemical Fate and Transport in the Environment," in Fate of
Pesticides and Chemicals in the Environment (J.L. Schoor, Ed.). John Wiley & Sons, 1992, pp.
1-24.
343 Murray, D.M.; Burmaster, D.E. (1995) Residential air exchange rates in the United States:
empirical and estimated parametric distributions by season and climatic region. Risk Analysis
15: 459-465.
344 Fugler, D.; Grande, C.; Graham, L. (2002) Attached garages are likely path for pollutants.
ASHRAE IAQ Applications 3(3): 1-4.
3-194
-------
345 Sheltersource, Inc. (2002) Evaluating Minnesota homes. Final report. Prepared for
Minnesota Department of Commerce. U.S. Department of Energy grant DE-FG45-96R530335.
346 Isbell, M.A.; Stolzberg, R.J.; Duffy, L.K. (2005) Indoor climate in interior Alaska:
simultaneous measurement of ventilation, benzene and toluene in residential air of two homes.
Sci. Total Environ. 345: 31-40.
347 Batterman, S.; Jia, C.; Hatzivasilis, G.; Godwin, C. (2006) Simultaneous measurement of
ventilation using tracer gas techniques and VOC concentrations in homes, garages and vehicles.
J. Environ. Monit. 8: 249-256.
348 Batterman, S.; Jia, C.; Hatzivasilis, G. (2006) Migration of volatile organic compounds from
attached garages to residences: a major exposure source. Environ Res (In Press).
349 George, M.; Kaluza, P.; Maxwell, B.; Moore, G.; Wisdom, S. (2002) Indoor air quality &
ventilation strategies in new homes in Alaska. Final Report. Alaska Building Science Network.
350 Tsai, P.; Weisel, C.P. (2000) Penetration of evaporative emissions into a home from an M85-
fueled vehicle parked in an attached garage. J. Air & Waste Manage. Assoc. 50: 371-377.
351 Isbell, M.A.; Stolzberg, R.J.; Duffy, L.K. (2005) Indoor climate in interior Alaska:
simultaneous measurement of ventilation, benzene and toluene in residential indoor air of two
homes. Sci. Total Environ. 345: 31-40.
352 Isbell, M.A.; Stolzberg, R.J.; Duffy, L.K. (2005) Indoor climate in interior Alaska:
simultaneous measurement of ventilation, benzene and toluene in residential indoor air of two
homes. Sci. Total Environ. 345: 31-40.
353 Schlapia, A.; Morris, S.S. (1998) Architectural, behavioral, and environmental factors
associated with VOCs in Anchorage homes. Presented at the Air & Waste Management
Association's 91st Annual Meeting & Exhibition. June 14-18, 1998, San Diego, CA.
354 Isbell, M.A.; Stolzberg, R.J.; Duffy, L.K. (2005) Indoor climate in interior Alaska:
simultaneous measurement of ventilation, benzene and toluene in residential indoor air of two
homes. Sci. Total Environ. 345: 31-40.
355 Batterman, S.; Jia, C.; Hatzivasilis, G. (2006) Migration of volatile organic compounds from
attached garages to residences: a major exposure source. Environ Res. (In Press).
356 Data obtained from Dr. Clifford Weisel, EOHSI. E-mail correspondence dated November 3,
2006 from Dr. Clifford Weisel, EOHSI to Rich Cook and Chad Bailey, U.S. EPA. Docket
number Docket EPA-HQ-OAR-2005-0036.
357 U.S. Energy Information Administration. Residential Energy Consumption Survey, 2001.
Table HCl-4a. Housing Unit Characteristics by Type of Housing Unit, Million U.S. Households,
2001 [Online at http://www.eia.doe.gov/emeu/recs/recs2001/hcjdf/housunits/hcl-
4a_housingunits2001 .pdfl
358 Environmental Protection Agency. (1997) Exposure factors handbook. National Center for
Environmental Assessment report EPA/600/P-95/002Fa-c [Online at
http://www.epa.gov/ncea/pdfs/efh/front.pdf]
3-195
-------
359 Batterman, S.; Jia, C.; Hatzivasilis, G.; Godwin, C. (2006) Simultaneous measurement of
ventilation using tracer gas techniques and VOC concentrations in homes, garages and vehicles.
J. Environ. Monit. 8: 249-256.
360 Graham, L.A.; Noseworthy, L.; Fugler, D.; O'Leary, K.; Karman, D.; Grande, C. (2004)
Contribution of vehicle emission from an attached garage to residential indoor air pollution
levels. J. Air & Waste Manage. Assoc. 54: 563-584.
361 Evans, J.S.; Wolff, S.K.; Phonboon, K.; Levy, J.I.; Smith, K.R. (2002) Exposure efficiency:
an idea whose time has come? Chemosphere 49: 1075-1091.
3-196
-------
Final Regulatory Impact Analysis
Chapter 4: Table of Contents
Chapter 4: Industry Characterization 2
4.1 Light-Duty Vehicle and Light-Duty Truck Market Structure 2
4.1.1 Domestic vs. Foreign Manufacturers 2
4.1.2 Light-Duty Vehicles vs. Light-Duty Trucks 5
4.1.3 Small Volume Manufacturers, Importers, and Alternative Fuel Vehicle Converters. 7
4.2 Petroleum Refining Industry 7
4.2.1 Gasoline Supply 8
4.2.2 Gasoline Demand 8
4.2.3 Industry Organization 9
4.2.4 Gasoline Market Data 9
4.3 Portable Fuel Container Industry 10
4.3.1 Manufacture and Distribution 10
4.3.2 Container Use 10
4.3.3 Market Structure 11
4.3.4 Market Entry 11
4-1
-------
Final Regulatory Impact Analysis
Chapter 4: Industry Characterization
An understanding of the nature of the affected industries is useful in assessing the potential
impact of the proposed emission control program. Information regarding the structure of the market,
including such things as the degree of concentration, entry barriers, and product differentiation, can
help explain the pricing and other policies that exist in that market. This chapter describes the
light-duty vehicle (LDV) and light-duty truck (LDT) manufacturers, the petroleum refining industry,
and the portable fuel container manufacturers.
4.1 Light-Duty Vehicle and Light-Duty Truck Market Structure
The LDV/LDT market is fairly concentrated, with only five of the 19 total generally-recog-
nized manufacturers accounting for almost 82 percent of all sales. LDV/LDT sales numbered more
than 16.9 million vehicles in 2004. The top five companies are the so-called "Big Three" (General
Motors (GM), Ford, and Daimler-Chrysler) plus Toyota and Honda. The remaining 18 percent of
sales are split between the other 14 manufacturers, with none of them achieving more than 2 percent
of total sales. The bottom 10 manufacturers in fact account for only about 4.5 percent of total sales.
Four of these firms, Ferrari, Maserati, Lamborghini, and Lotus, are considered small-volume
manufacturers, since their sales are less than 15,000 vehicles per year.A Table4.1.-l provides sales
figures by manufacturer.
None of the major manufacturers are small businesses. (As discussed later in Chapter 14, the
Small Business Administration (SBA) criterion for a small business in the vehicle manufacturing
industry is 1,000 employees or less.) This is mainly because of the large outlay of capital and other
resources necessary to enter the market. Becoming even a relatively minor player in the industry
requires a great deal of manufacturing capacity to achieve the necessary production volumes, as well
as an extensive distribution and marketing network. There is also a significant amount of brand
loyalty on the part of consumers, because of tradition or perceived differences in the product. These
all combine to make market entry difficult, and the industry is basically dominated by the
established major manufacturers.
As discussed later in Section 4.1.3, there are also a few smaller, lesser-known LDV/LDT
small volume manufacturers, importers and alternative fuel vehicle converters. These have limited
product lines, and account for less than one-tenth of one percent of all U.S. sales. They primarily fill
niche markets of one kind or another. More than half of these firms are small businesses.
4.1.1 Domestic vs. Foreign Manufacturers
EPA defines small volume manufacturers to be those with total U.S. sales of less than 15,000 vehicles per year. This
status allows vehicle models to be certified under a slightly simpler certification process. For certification purposes,
small volume manufacturers also include independent commercial importers (ICIs) and alternative fuel vehicle
converters since they sell less than 15,000 vehicles per year.
4-2
-------
Final Regulatory Impact Analysis
Previously, it has been relatively easy to characterize manufacturers as "domestic" or
4-3
-------
Final Regulatory Impact Analysis
g
LJJ
D
O
LJJ
S
-------
Final Regulatory Impact Analysis
"foreign." However, this is currently much more difficult. For example, the Daimler-Chrysler
merger combined the former Chrysler divisions Chrysler, Dodge and Jeep with the imported
Mercedes line; but it also includes Maybach, a high-end German luxury car. Ford now includes not
only the traditional Ford, Mercury and Lincoln lines, but also the imported marques Jaguar, Volvo,
Land Rover and Aston-Martin. GM sales include the Swedish import Saab.
Conversely, Toyota and Honda, as well as the six other Far Eastern manufacturers, all
maintain a substantial American manufacturing presence, and the majority of their vehicles sold
here, almost 80 percent on average, are manufactured in North America. Sales figures from North
American manufacturing facilities for individual firms range from 95 to 98 percent for Toyota and
Honda, to 52 to 72 percent for some of the smaller manufacturers. Volkswagen, which now also
includes Bentley, is the only European manufacturer with a North American manufacturing opera-
tion. About 55 percent of its sales are manufactured here. BMW, which now includes the formerly
British Rolls-Royce and Mini lines, is 100 percent imported, as is Porsche.
On the other hand, substantial portions of the Ford and GM "domestic" lines are also
imported. Actually, the term "North American-built," meaning "made in the United States, Canada
or Mexico," seems to have replaced the term "domestic" in the sales reports. About 28 percent of all
domestic LDVs sold in the U.S. are considered "imports," i.e., not North-American built, as opposed
to only about 13 percent of all LDTs.
4.1.2 Light-Duty Vehicles vs. Light-Duty Trucks
In earlier years, light-duty vehicles tended to outsell light-duty trucks by a fairly wide
margin. In 1981, for example, LDTs comprised less than 20 percent of total sales, and this had only
grown to about 38 percent by 1993. However, in recent years the gap has been closing rapidly.
LDTs have made considerable gains in the last decade; by the 2000 model year LDVs outsold LDTs
by a margin of only about 52 to 48 percent. By 2001 the split was roughly 50/50, with LDT sales
actually moving slightly ahead by about 100,000 units.l As shown in Table 4.1-1, for the 2004
model year, LDTs outsold LDVs by a 55 to 45 percent margin. The rise of the Sport-Utility Vehicle
(SUV) accounts for much of this change, but stronger sales of the more traditional LDTs account for
a substantial amount of the increase as well.
In general, LDTs and LDVs are produced by the same manufacturers, both foreign and
domestic. The Big Three plus Toyota and Honda account for almost 90 percent of LDT sales. The
Big Three actually account for almost 75 percent of all LDT sales, but only about 45 percent of all
LDV sales. All of the Far Eastern manufacturers, except for Isuzu and Subaru, also make LDTs as
well as LDVs. Isuzu sells only LDTs, in the U.S. while Subaru sells only LDVs. Three European
manufacturers, Volkswagen, BMW, and Porsche, sell both LDTs and LDVs, while the remaining
four European manufacturers sell only LDVs. These four are all small-volume, high-end sports car
manufacturers (Ferrari, Maserati, Lamborghini and Lotus). Figures 4.1-1 and 4.1-2 show market
shares for LDV and LDT manufacturers.
4-5
-------
Final Regulatory Impact Analysis
Figure 4.1-1.
LDV MANUFACTURER SALES
2.4% -,
2.9%
4.0%
2.4% -, r 1.4%
24.4%
5.9%
7.0%
11.0%
13.5%
GENERAL MOTORS
FORD MOTOR CO.
DAIMLER CHRYSLER
TOYOTA MOTOR CO.
AMERICAN HONDA
NISSAN MOTOR CO.
HYUNDAI GROUP.
I | VW of AMERICA
• BMW GROUP
D MAZDA
• SUBARU
• MITSUBISHI
D ALL OTHERS
13.7%
10.5%
Figure 4.1-2.
LOT MANUFACTURER SALES
0.3%
2.5% -,
4.9%
0.8%
0.8%
10.9%
30.1%
GENERAL MOTORS
FORD MOTOR CO.
DAIMLER CHRYSLER
HYUNDAI GROUP.
VW of AMERICA
BMW GROUP
TOYOTA MOTOR CO. D MAZDA
AMERICAN HONDA
NISSAN MOTOR CO.
ALL OTHERS
17.6%
24.8%
-------
Final Regulatory Impact Analysis
For regulatory purposes, LDVs and LDTs are divided into categories based on their gross
vehicle weight ratings (GVWR). This distinction was based on the premise that heavier vehicles
produce more pollutants than do lighter vehicles, making it more difficult to achieve comparable
emission reductions. Standards for the heavier vehicles were therefore less stringent. However,
modern emission-control technologies are virtually the same and equally effective for both the
lighter and the heavier vehicles. Therefore, the Tier 2 emission standards now make no distinction
between weight categories. In addition, Tier 2 applies to medium duty passenger vehicles
(MDPVs), i.e. passenger vehicles between 8,500 and 10,000 Ibs. GVW. These are primarily the
very large SUVs, and passenger vans.
Emission standards were also slightly less stringent for the LDTs than for LDVs, partly
because of weight considerations, and partly because of perceived differences in usage patterns.
Again, the Tier 2 emission standards now make no distinction between LDVs and LDTs, except for
some minor differences in the evaporative emissions standards. In large part this is because LDVs
and LDTs share the same basic emission-control technologies and are primarily used for the same
purpose, for personal transportation. Thus, there does not appear to be a strong rationale for making
distinctions between the two.
4.1.3 Small Volume Manufacturers, Importers, and Alternative Fuel Vehicle Converters
There are a number of lesser-known small volume manufacturers who produce high
performance and other specialized vehicles, such as Roush Industries or the Panoz Auto Develop-
ment Company. These number less than a dozen, and about half are small businesses. In addition to
the manufacturers, there are a handful of Independent Commercial Importers (ICIs) who are issued
certificates to import a limited number of nonconforming vehicles for racing or other purposes, and
to modify these vehicles to meet U.S. standards.8 These ICIs are almost all considered small busi-
nesses, and total sales for all of them are fewer than 500 vehicles per year. There are also a small
number of converters who convert conventional gasoline- or diesel-fueled vehicles to operate on
alternative fuel (e.g., compressed natural gas and liquefied petroleum gas). These are also few in
number, and are almost all small businesses. Altogether, combined sales for these small-volume
manufacturers, importers, and converters accounted for less than one-tenth of one percent of total
sales of LDVs and LDTs for the 2004 model year.
4.2 Petroleum Refining Industry
Early in this rulemaking process, EPA commissioned an analysis of the U.S. gasoline pro-
duction and distribution system from RTI International in order to support economic analyses of the
proposal. The final report of the analysis, entitled "Characterizing Gasoline Markets: A Profile,"
discusses supply and demand issues associated with the refining industry and with gasoline market
ICIs are not required meet the emission standards in effect when the vehicle is modified, but instead they must meet
the emission standards in effect when the vehicle was originally produced (with an annual production cap of a total of
50 light-duty vehicles and trucks).
4-7
-------
Final Regulatory Impact Analysis
behavior.2 The information contained in the report is summarized below, supplemented by addi-
tional information found in this RIA and in other sources.
4.2.1 Gasoline Supply
Detailed descriptions of the refinery processes by which gasoline is produced can be found
in the final report mentioned above and in Chapter 6 of this RIA. Gasoline is the dominant product
for most refineries, constituting almost half of the total product produced by U.S. refineries in 2002.3
Federal and state regulations have resulted in a variety of gasoline formulations. These include the
RFG and CG designations, oxygenated gasoline, octane-based gasoline grades, and volatility
distinctions. Additional variation occurs when different oxygenates are used, though that difference
will lessen significantly in the coming years as MTBE use diminishes and the renewable oxygenate
requirements of the Energy Policy Act of 2005 cause a substantial increase in ethanol use in
gasoline. Some gasoline regulations, such as gasoline sulfur and MSAT1, affect all gasoline and
impact refineries and gasoline production, but do not contribute to additional gasoline types.
Gasoline supply is also affected by the types of crude oils available, and the refining indus-
try's ability to process the different crude types to maximize gasoline production while meeting all
applicable regulations. Sweet, or low sulfur, crude oils are more easily processed, but this factor
increases their cost compared to sour, or high sulfur, crude oils. Some refineries are optimized to
run based on a certain type of crude oil, and have little flexibility in processing other types. Crude
cost is the largest factor in total refining cost and the price of crude can significantly affect the total
cost of product on.
Gasoline and other petroleum products are transported from the refineries to intermediate
points such as terminals, and to the final market by pipeline, truck and barge. Most product is
moved via pipeline, as the cost is extremely low. Pipelines have been able to accommodate the
many gasoline formulations that have resulted from federal and state gasoline regulations, but are
near their limit in handling additional formulations. Modifying schedules and flow rates in order to
get gasoline and non-gasoline products on and off the pipeline contributes to increased costs. The
final step for gasoline transport to retail outlets is via truck.
4.2.2 Gasoline Demand
Gasoline demand is affected by gasoline use and factors that influence consumption. The
vast majority of gasoline is used for private and commercial highway use. About 3 percent is used
in non-highway applications such as lawn and garden or marine use. Light-duty transportation
accounts for over 90% of gasoline used, and most of this is attributable to private automobile use.
Transportation choices, and thus gasoline use, are affected by many factors, including personal
income, geography, gasoline prices and the prices of related goods. Though daily travel increases
with household income, average annual expenditures for gasoline, as a percent of income, showed
little variation by geography or income class. Consumers can respond to gasoline price increases in
many ways, such as reducing the number of miles traveled, or by adjusting their "capital stock," that
4-8
-------
Final Regulatory Impact Analysis
is, for example, by purchasing a car with better fuel economy.
4.2.3 Industry Organization
The refining industry structure is critical to the implementation and impact of the proposed
regulation. Factors such as regional production and shipment patterns and industry concentration
can influence market price and product availability. For instance, because of current fuel formula-
tions and distribution patterns, consideration of regional (PADD) gasoline markets, rather than a
national gasoline market, may be more appropriate for evaluating certain impacts of the proposed
regulatory program.
Market concentration refers to some measure of the market share of competitors in an area.
High market concentration may indicate some ability of competitors in an area to influence prices by
coordinated action, thus resulting in less competition and higher product prices. A recent Federal
Trade Commission analysis has shown that the refining industry is not concentrated or only
moderately concentrated. In addition, the possibility of increased gasoline imports, particularly into
PADDs I and III, can serve to moderate any attempts to set prices.
Refiners serving the same market may have a wide range of total delivered costs. Cost to the
refiner is a function of distance to market, refinery-specific operating costs and gasoline formulation.
Gasoline formulation, as discussed, depends on the crude oil, refinery configuration and
environmental or other gasoline controls. The market price for gasoline is set by the producers with
the highest costs, taking into consideration their full range of products produced.
4.2.4 Gasoline Market Data
An analysis of the impacts of a policy change—in this case, from current gasoline toxics re-
quirements to the proposed fuel benzene standard-requires consideration of the baseline case com-
pared to likely changes expected from the new policy. National and regional (by PADD) consump-
tion and gasoline price, price volatility, international trade, and projected growth (in gasoline con-
sumption) are the primary factors considered in estimating economic impacts of the proposed rule.
Gasoline consumption is estimated to increase by about 1.8 percent annually through 2025.
As discussed above, gasoline consumption, primarily influenced by personal light-duty vehicle use,
is affected by many factors, including retail gasoline price. Gasoline price is a function of distribu-
tion and marketing costs, refining costs, profit, federal and state taxes, and crude oil cost. Crude oil
cost accounts for almost half of the retail price of gasoline. Price volatility is primarily due to the
magnitude of any supply and demand imbalance, and the speed with which new supply can be pro-
vided. These imbalances can be caused by unexpected refinery shutdowns or pipeline disruptions,
or even by relatively planned activity, such as seasonal transitions. Isolated markets, or those re-
quiring unique gasoline blends, are likely to be more susceptible to such supply and demand
imbalances.
4-9
-------
Final Regulatory Impact Analysis
International gasoline trade, that is, imports and exports of gasoline, account for an
extremely small part of all gasoline transactions. However, regional activity, at the PADD level,
shows significant variation. PADD I received over 90% of all gasoline and gasoline blendstock
imports in 2002.4
4.3 Portable Fuel Container Industry
EPA also contracted with RTI International for a characterization of the PFC industry in
support of our economic analyses of the proposal. The final analysis report, entitled "Characterizing
Gas Can Markets: A Profile," discusses production and distribution issues associated with gas
cans.5'6 This report is also summarized below, and is again supplemented by additional information
found in this RIA and in other sources. PFCs include gasoline, kerosene, and diesel containers.
4.3.1 Manufacture and Distribution
PFCs are designed to transport, store and dispense fuel, normally for refueling vehicles when
they run out of gas, or for home applications such as refueling lawnmowers, trimmers, etc. PFCs
include utility jugs that are marketed for use with fuels, which are often used to refuel recreational
products such as personal watercraft and all-terrain vehicles. PFCs range in capacity from a gallon
or less to over 6 gallons. Standard PFCs have three main components: a spout for pouring fuel, a
tank with a fill port to hold the gasoline, and a vent to make pouring the fuel easier. About 98
percent of all containers are made of high-density polyethylene (HOPE) plastic, chosen mainly
because of its fuel-resistant properties. Two main manufacturing processes are used: extrusion blow
molding, which is used for the bodies, in which a molten tube of plastic is forced into a mold by
compressed air; and injection molding, which is used for spouts, caps and other tubes. In injection
molding, plastic material is forced through a heated injection chamber and through a nozzle into a
cold mold. Because of safety regulations in most states, gas cans are colored red during the
manufacturing process. Diesel containers are colored yellow and kerosene containers are colored
blue to help consumers avoid misfueling of equipment. Industry and other sources indicate that gas
cans and diesel and kerosene containers are distributed by manufacturers through their distribution
centers to major retail establishments. Utility jugs are sold in several colors and are more often sold
through online retailers.
4.3.2 Container Use
PFCs allow people to refuel a wide variety of equipment without the inconvenience of taking
it to a retail gasoline station. This equipment can range from lawn and garden equipment such as
tractors, lawnmowers, trimmers and chainsaws to recreational vehicles such as motorcycles, ATVs
and golf carts. We estimate that there are about 80 million gas cans in the U.S., which is similar to
other such estimates.7 Although publicly-available data on gas can usage are scarce, a California Air
Resources Board (CARB) study performed in 1999 indicated that 94 percent of all gas cans in
California were used in households. The remaining 6 percent were used for such commercial
applications as farming, logging, construction, lawn care, and automotive applications such as repair
4-10
-------
Final Regulatory Impact Analysis
shops and gasoline stations. State surveys in California and Texas indicated that between 46 and 72
percent of all households owned gas cans, and that 14 percent of those surveyed had bought one
during the past year. The average number of gas cans ranged from 1.4 per household in Texas to 1.8
per household in California. A typical plastic PFC will have a life expectancy of 3 to 5 years before
it needs to be replaced.
The demand for fuel containers reflects the demand for other goods and services. The gas
can industry has suggested that the sales of gas cans are linked to the sales of gasoline-powered
equipment such as lawn and garden equipment or recreational vehicles. Therefore, factors that
influence the sales of these types of equipment will also influence the sales of gas cans. These
factors can include such things as price, population growth, or changes in personal income.
Gasoline container sales for 2002, the latest year for which we were able to develop data,
were about 24.4 million units (including utility jug sales which were estimated to be about 2.4
million units). Diesel and kerosene container annual sales are estimated to be about 620,000 and 1
million units, respectively. Although the PFC manufacturing industry has become fairly
concentrated, with one firm accounting for more than half of all U.S. container sales, that firm does
not exert significant influence over market prices. This is because there are few barriers to market
entry by other companies, and the products are substantially the same, making for very limited brand
loyalty. Other firms could enter or re-enter the market should the economic conditions seem right.
Imports from Canada, which amount to about 10 percent of annual sales, would also tend to limit
arbitrary pricing practices.
4.3.3 Market Structure
As noted above, the PFC market is fairly concentrated, with only five firms accounting for
the vast majority of sales. These are Blitz USA, Midwest Can, Scepter Manufacturing, Ltd.
(Canadian), No-Spill Research, and Wedco Molded Products, which is owned by the Plastics Group.
All of these companies, except for the parent company Plastics Group, meet the primary Small
Business Administration (SBA) criterion for small businesses (i.e., less than 500 employees). Data
for utility jug manufacturers was scarce, but we believe that there are likely about 5 manufacturers of
these containers, including Scribner Plastics. There are other gasoline container manufacturers, but
they have a very limited market share. Most of their products are designed for industrial use or to fill
a niche market (e.g., racing or safety cans used in an industrial setting), which are not be covered by
the standards. These companies include Eagle Manufacturing and Protectoseal Company. Table
4.3-1 provides relevant data about these firms.
4.3.4 Market Entry
There are very few barriers to entering the PFC market. Only about 2 percent of the
containers sold in the U. S. in 2002 were of metal construction; the vast majority were plastic. These
are produced by a fairly straightforward molding process in much the same manner as hundreds if
not thousands of other plastic products. Plastic PFCs are in fact classified in the U.S. Economic
4-11
-------
Final Regulatory Impact Analysis
Census as "All other plastics product manufacturing." Since manufacturing such PFCs is similar to
manufacturing most other molded plastic products, any firm with that capability could freely enter
the market with a relatively low initial investment, if the economic conditions should appear
advantageous to do so. Since most consumers tend to view gas cans as more or less all the same,
there is not a well-developed brand loyalty to one brand or other, so competition in the industry is
based primarily on price. Finally, safety regulations in most states prevent consumers from using old
paint thinner cans or other such containers as substitutes for gas cans, thus eliminating any potential
reduction in sales from that quarter.
4-12
-------
Final Regulatory Impact Analysis
Table 4.3-1. Manufacturers*
Ultimate Parent
Blitz USA
Eagle Manufacturing
Midwest Can
No-spill Research Inc.
Protectoseal Co.
Scepter Mfg., Ltd.
Scribner Plastics
The Plastics Group
Company name
Blitz USA
Eagle Manufacturing
Midwest Can
No-spill Research Inc.
Protectoseal Co.
Scepter Mfg., Ltd.
Scribner Plastics
Wedco Molded Prod.
Sales (Smillion)
20-50
50-100
20-50
2.5-5
20-50
10-20
5-10
20-50
Employment
200
100-249
45
5
100-249
200
20-49
600
Comments
Consumer market
Primarily Metal Safety Cans
Consumer market
Limited Distribution
Primarily Industrial
Canadian-Consumer
Specialty Containers
Consumer Market
* Businesses Engaged In NAICS Code 326119, All Other Plastic Product Manufacturing, Or NAICS Code 332431, Metal Can
Manufacturing
Source: Characterizing Gas Can Markets, a Profile," RTI International, Final Report, EPA Contract 68-D-99-024.
4-1
-------
Final Regulatory Impact Analysis
References for Chapter 4
1 Source: Ward's "World Motor Vehicle Data, 2004."
2 "Characterizing Gasoline Markets: A Profile," Final Report. EPA Contract Number 68-D-99-
024, prepared for Robert Johnson, USEPA, Office of Transportation and Air Quality, Ann Arbor,
MI by Brooks Depro, et al, RTI International, Research Triangle Park, NC, August 2004.
3 Table 1-1 in Final Report. Data Source: DOE, EIA Petroleum Supply Annual 2002.
4 Table 4-4. Final Report. Source DOE EIA Petroleum Supply Annual 2002.
5 "Characterizing Gas Can Markets: A Profile." Final Report, EPA Contract Number 68-D-99-
024, prepared for Robert Johnson, USEPA, Office of Transportation and Air Quality, Ann Arbor,
MI by Brooks Depro, et al, RTI International, Research Triangle Park, NC, August 2004.
6 "Gas Can Industry Profile Updates", memorandum from RTI to EPA, January 4, 2007.
7 Memorandum from Terrance R. Karels, Consumer Product Safety Commission, January 3, 2003.
4-1
-------
Final Regulatory Impact Analysis
Chapter 5: Table of Contents
Chapter 5: Vehicle Technological Feasibility 2
5.1 Feasibility of Cold Exhaust Emission Standards for Vehicles 2
5.1.1 NMHC Emissions Control Technologies on Tier 2 Gasoline-Fueled Vehicles 2
5.1.1.1 Calibration and Software Control Technologies 4
5.1.1.1.1 Idle Speed and Air Flow Control 4
5.1.1.1.2 Spark Control 5
5.1.1.1.3 Secondary Air Injection Control 5
5.1.1.1.4 Cold Fuel Enrichment 6
5.1.1.1.5 Closed Loop Delay 7
5.1.1.1.6 Transient Fuel Control 8
5.1.1.1.7 Fuel Volatility Recognition 8
5.1.1.1.8 Fuel Injection Timing 9
5.1.1.1.9 Spark Delivery Control 10
5.1.1.1.10 Universal Oxygen Sensor 10
5.1.1.2 Tier 2 Engine and Exhaust Control Technologies 11
5.1.2 Data Supporting Cold NMHC Standard Technical Feasibility 11
5.1.2.1 Certification Emission Level 11
5.1.2.2 EPA Test Programs 18
5.1.2.2.1 2004 Chevrolet Trailblazer Testing 19
5.1.2.2.2 2006 Chrysler 3OOC Testing 22
5.2 Feasibility of Evaporative Emissions Standards for Vehicles 25
5-1
-------
Final Regulatory Impact Analysis
Chapter 5: Vehicle Technological Feasibility
5.1 Feasibility of Cold Exhaust Emission Standards for Vehicles
5.1.1 NMHC Emissions Control Technologies on Tier 2 Gasoline-Fueled Vehicles
Emission control technology has evolved rapidly since the passage of the CAA
Amendments of 1990. Emission standards applicable to 1990 model year vehicles required
roughly 90 percent reduction in exhaust non-methane hydrocarbon (NMHC) emissions compared
to uncontrolled emission levels. The Tier 2 program and before that, the National Low Emission
Vehicle (NLEV) program, contain stringent standards for light-duty vehicles that have resulted
in additional NMHC reductions. Tier 2 vehicles currently in production show overall reductions
in NMHC of more than 98 percent compared to uncontrolled emissions levels. These emission
standards for NMHC are measured under the EPA Federal Test Procedure (FTP), which
measures exhaust emissions from vehicles operating only in the ambient temperature range of
68° F to 86° F.
Table 5.1-1 below lists specific types of NMHC emission controls that EPA projected in
the Tier 2 technological feasibility assessment could be used in order to meet the final Tier 2
standards. It is important to point out that all of the following technologies have not necessarily
been needed to meet the Tier 2 standards. The choices and combinations of technologies have
depended on several factors, such as current engine-out emission levels, effectiveness of existing
emission control systems, and individual manufacturer preferences. In some cases, no additional
hardware from the NLEV level of hardware was needed. Instead, many manufacturers focused
their efforts in the software and calibration controls to achieve stringent emission levels.
Table 5.1-1. Tier 2 Projected Emission Control Hardware and Technologies
Emission Control Technologies
Fast Light-off Exhaust Oxygen Sensors
Retarded Spark Timing at Start-up
More Precise Fuel Control
Individual Cylinder Control
Manifold with Low Thermal Capacity
Air Assisted Fuel Injection
Faster Microprocessor
Secondary Air Injection into Exhaust
Heat Optimized/Insulated Exhaust Pipe
Close-coupled Catalyst
Improved Catalyst Washcoats/Substrates
Increased Catalyst Volume and Loading
Engine Modifications
Universal Exhaust Oxygen Sensor
A number of technological advances and breakthroughs have allowed these significant
emission reductions to occur without the need for expensive emission control equipment. For
example, the California Air Resources Board (ARB) originally projected that many vehicles
would require electrically-heated catalysts to meet their Low Emission Vehicle I (LEV I)
program requirements. Today, with even more stringent standards than LEV I, no manufacturer
needs to use these devices to comply with program requirements. Similarly, the Tier 2 and Low
Emission Vehicle II (LEV II) programs, currently being phased-in, have projected that some
additional emission control hardware and techniques may be required. However, initial
5-2
-------
Final Regulatory Impact Analysis
indications from the Tier 2 vehicles already certified indicate that increases in emission control
hardware have been kept to a minimum, likely to minimize cost.
The Tier 2 program requires reductions in all regulated pollutants, but the largest
reductions are required for oxides of nitrogen (NOx) emissions. To achieve these NOx
reductions, significant improvements in catalyst technologies have been employed, largely in
improved catalyst substrates and washcoats containing the precious metals. In fact, some
manufacturers have even been able to reduce precious metal loadings as compared to previous
generation catalysts because of the new substrate and washcoat improvements developed in
response to Tier 2. These catalyst technologies have generally also resulted in better emission
performance of all regulated pollutants, largely because of improved catalyst light-off times.
The Tier 2 program also includes new tighter non-methane organic gases (NMOG)
standards. Unlike tight NOx controls, manufacturers had significant experience in non-methane
hydrocarbons (NMHC) controls from the stringent NMOG standards (NMOG consists primarily
of NMHC) under the NLEV and LEV I programs. In fact, the NMOG standards for a Tier 2 Bin
5 package are the same for the passenger car and light-duty truck as those established under the
NLEV program. One of the largest challenges manufacturers have encountered under the Tier 2
program is the program's weight neutral standards for all vehicles up to 8500 Ibs. gross vehicle
weight rating (GVWR) and medium-duty passenger vehicles (MDPV) up to 10,000 Ibs. GVWR.
These heavier vehicles may be where new hardware will more likely be required to meet Tier 2
weight neutral standards as they fully phase in to Tier 2.
Some of the most significant technological advances that have facilitated low NMHC
emission levels have occurred in calibration and software-based controls. These controls have
been carefully designed to both minimize exhaust emissions before exhaust aftertreatment has
reached operational temperature and accelerate the usage of the aftertreatment earlier in the
operation of the engine. Additionally, fuel metering controls during the critical period prior to
aftertreatment reaching operating temperature is more precise than previous systems, largely due
to advances in software controls. While some improvements also have been made to base engine
designs, which have resulted in lower overall operating engine-out emissions, controls aimed at
minimizing emissions during the critical period before exhaust aftertreatment readiness have
been accomplished almost exclusively with software based controls. Even with base engine and
exhaust hardware improvements, calibration and software controls of the emission control
hardware remain the most important and powerful emission control technique used by
manufacturers. Calibrations and software controls will continue to become more refined and
sophisticated as manufacturers learn new ways to better utilize existing hardware, particularly in
the remaining Tier 2 phase-in vehicle models.
Today, these emission control strategies are utilized at 75° F to meet stringent Tier 2 and
LEV II NMOG standards. The potential exists for these same software and calibration controls
to be utilized at 20° F and all other cold start temperatures to control NMHC emissions. Most of
these controls are feasible and available today in Tier 2 and LEV II vehicles. With the
implementation of these controls at the colder start temperatures, significant reductions in
NMHC emissions (and therefore air toxics) can be realized. The following sections provide
5-3
-------
Final Regulatory Impact Analysis
details on these software and calibration control strategies, supporting certification results, and
feasibility studies utilizing these existing emission control opportunities.
5.1.1.1 Calibration and Software Control Technologies
Tier 2 vehicles are equipped with very sophisticated emissions control systems. Table
5.1-1 above lists some of the technologies manufacturers have successfully used to meet
stringent Tier 2 emission standards. In addition to hardware technologies, manufacturers have
developed calibration and software control strategies to meet Tier 2 emission standards that also
can be effectively used at 20° F to achieve significant reductions in NMHC and other emissions.
We expect manufacturers will expand the use of these same emission control strategies already
in place on Tier 2 vehicles at 75° F to control NMHC emissions at 20° F. The following
descriptions provide an overview of the calibration and software technologies capable of
reducing exhaust emissions at 20° F.
5.1.1.1.1 Idle Speed and Air Flow Control
Idle speed and air flow control have been utilized very successfully to both reduce
emissions before the catalyst aftertreatment is considered active and to accelerate the activity of
the catalyst. Elevated idle speeds immediately following the start of a vehicle, particularly in
park and neutral, will result in more stable combustion resulting from the improved air and fuel
mixture motion. This is largely due to the higher air velocity entering the combustion chamber
which generally results in a more homogeneous mixture, and therefore, a more fully combustible
air-fuel mixture. The higher engine speed may also increase heat created from piston to cylinder
wall friction, further assisting in transforming fuel droplets to burnable mixtures. The higher
engine speeds cause additional combustion events, in which contribute to the rapid heating of the
combustion chamber. The higher combustion stability can generally result in the ability to run
leaner air-fuel ratios, which reduces the percentage of unburned fuel that would be exhausted
from the engine.
Air flow through the engine, exhausted after combustion, provides the heat required for
the catalyst to become active. Increased air flow through the engine, mainly through elevated
idle speeds, provides the catalyst with supplemental heat. Additionally, this extra exhaust heat is
carried to the catalyst at higher exhaust flow velocities, further shortening the amount of time the
catalyst is inactive. The higher combustion stability from the increased air flow results in less
hydrocarbons from unburned fuel, which can actually quench a catalyst and slow its warming.
The ability to run leaner mixtures can provide the catalyst with the necessary oxygen for the
catalyst to begin oxidation of NMHC and carbon monoxide (CO).
Elevated air flow used off-idle can also produce significant emission benefits. This
elevated air flow is achieved by allowing extra air flow primarily when the throttle is closed, but
also during the transient period when the throttle is in the process of closing. This momentary air
flow increase has been referred to as "dashpot" effect. It typically has been used only for short
durations following a throttle closing to help provide additional air flow, and usually only during
the first few minutes of cold start engine operation. Elevated air flow has also been used to
5-4
-------
Final Regulatory Impact Analysis
provide slightly more closed throttle engine torque to overcome additional loads only
encountered following a cold start. This reduces risk of idle undershoots and stalling.
5.1.1.1.2 Spark Control
Spark control has evolved with modern electronic controls to a highly precise tool to
carefully control when the combustion event is initiated in a spark ignition engine. Retarding the
spark delivery immediately after the start has been highly effective at reducing exhaust
emissions. Retarding the spark, particularly after a cold start, generally reduces engine-out
emissions. This is generally believed to be a result of the longer period of time that the fuel is
under compression and absorbing combustion chamber heat. This assists in more complete
combustion when the fuel is finally spark-ignited. It also is believed that the retarded spark
timing results in lower cylinder peak pressures during the combustion of the air-fuel mixture,
reducing the opportunity for hydrocarbons to migrate to crevices and further helping lower
engine-out hydrocarbon emissions.
Retarded timing also has been used very effectively to accelerate the early usage of the
catalyst by providing supplemental heat, which reduces the time for the catalyst to begin
oxidation. The retarded timing results in peak combustion of the air-fuel mixture occurring later
in the engine operating cycle, leading to significant thermal energy being transferred into the
exhaust. This thermal energy very effectively provides a boost to the catalyst warm-up,
particularly at colder temperatures and for large mass catalyst systems or catalyst systems that
are further from the engine.
The effectiveness of retarded timing can be enhanced significantly when used in
conjunction with elevated idle speeds and/or air flow control. The simultaneous use of the two
features generally results in greater emission reductions than when either feature is used
independently. Additionally, utilizing elevated idle speeds while retarding the timing can offset
any engine vacuum level concerns encountered when only retarded timing is used.
5.1.1.1.3 Secondary Air Inj ection Control
Many Tier 2 vehicles produced today contain secondary air injection systems to comply
with stringent Tier 2 and LEV II standards. These systems reduce vehicle emissions by injecting
ambient air into the rich engine exhaust upstream of the catalyst for a short period of time
immediately after a start. This reduces emissions in two ways. First, the oxygen in the ambient
air being pumped into the exhaust assists in oxidizing HC and CO prior to reaching the catalyst.
Second, this oxidation can generate large amounts of heat that help bring the catalyst to effective
temperatures much sooner. As the catalyst reaches effective temperature, the secondary air can
continue to provide needed oxygen for oxidation in the catalyst until the total system is ready to
go "closed loop," at which time the secondary air injection is ceased.
The secondary air injection technology for controlling emissions is not new. For many
years, manufacturers used secondary air injection systems that ran continuously from a
mechanical belt-driven pump to oxidize HC and CO emissions produced from a rich exhaust
mixture during all modes of operation. With the advent of the three-way catalyst, manufacturers
5-5
-------
Final Regulatory Impact Analysis
began to use engine control modules to activate electric air pumps to only reduce start emissions
at 75° F, typically on vehicle packages with specific cold start emission challenges. For
example, vehicles that have large mass catalysts or catalyst systems located relatively far from
the engine have utilized secondary injection to assist catalyst light-off Further, many Tier 2 and
LEV II packages certified to the cleanest emission levels utilize secondary air injection to
achieve these results. Some Tier 2 packages that appear to have relatively high engine-out
emissions, possibly due to engine design limitations, also have implemented secondary injection
to allow compliance with Tier 2 emission standards.
Many manufacturers that have equipped some of their Tier 2 vehicles with secondary air
injection systems do not appear to consistently utilize this emission control strategy across start
temperature ranges outside of the currently regulated cold start temperature (75° F for Tier 2 and
50° F for LEV II). However, many identical vehicle models that are sold in both Europe and the
U.S. are equipped with secondary air injection that does appear to be used at 20° F on the U.S.
model, based on our analysis of the certification data. This is attributable to shared emission
control technologies with the European market vehicles, where manufacturers are already
required to meet a 20° F NMHC standard.
The activation of the secondary air system is a feasible and effective emission control
technology for 20° F as well as all other interim start temperatures. The use of secondary air
injection technology at 20° F is well proven as an emission control technology, as observed in
the European vehicles. Certain design criteria must be taken into account for the system to
operate robustly at these colder temperatures, but there appears to be no technological challenge
that would prevent these vehicles already equipped with secondary air injection from activating
this emission control technology at 20° F.
Some manufacturers, who do not use secondary air injection systems at 20° F but do
include the systems on some of their U.S.-only models, have expressed concerns with freezing
water in the system. We have investigated this concern with the manufacturers of the secondary
air injection components and found this to be a system design issue that has been addressed by
guidelines on the location and plumbing of the individual secondary air injection components.1
5.1.1.1.4 Cold Fuel Enrichment
Gasoline-fueled spark ignition engines generally require rich air-fuel mixtures (i.e., a
larger amount of fuel for a given amount of air) for some amount of time immediately following
a cold start. Under normal operating conditions, the amount of required enrichment always
increases as start temperature decreases. This is largely because low in-cylinder temperatures for
some period of time following the cold start lead to a lower percentage of liquid fuel vaporizing
to a burnable mixture. The level of enrichment and its duration following the start will vary with
many factors, including base engine hardware design and fuel properties. Fuel property
interactions with engine combustion chamber dynamics are quite complex and can vary with fuel
composition, but typical gasoline fuel available in the U.S. during the cold weather (e.g., 20° F)
is properly formulated for robust cold start operation.
5-6
-------
Final Regulatory Impact Analysis
The level of enrichment should be calibrated to closely match the "winter"-grade fuel
properties that the overwhelming majority of vehicles will be experiencing during the colder start
conditions. Winter-grade fuel is formulated to have a higher Reid vapor pressure (RVP),
specifically to allow the fuel to vaporize at lower cold start temperatures and minimize the need
for additional enrichment. Any fuel enrichment beyond the minimum required level results in
proportional increases in cold start emissions, primarily NMHC and CO. Additionally, over-
fueling can hamper earlier use of the exhaust aftertreatment by quenching the catalyst with the
unburned fuel, effectively cooling the catalyst. This retards the warm-up rate of the catalyst and
also reduces the availability of any excess oxygen that would be used by the catalyst to oxidize
the NMHC and CO.
The amount of required enrichment also can be reduced when used in conjunction with
the previously mentioned elevated idle speed emission control technology. As stated earlier,
elevated idle speeds will result in a more homogeneous mixture which supports more stable
combustion. The improvements in the mixture will allow the enrichment levels to be reduced
accordingly.
5.1.1.1.5 Closed Loop Delay
"Closed loop" operation refers to operation that allows the exhaust oxygen sensor to feed
back to the engine control module and control the air-fuel mixture to an exhaust stoichiometric
ratio. Following start-up of a modern gasoline fueled engine, operation in closed loop is delayed
for some amount of time based on a combination of engine and oxygen sensor readiness criteria.
As stated in the previous section, gasoline-fueled engines require rich air-fuel mixtures for some
amount of time immediately following a start. The amount of time requiring the rich operation
and, therefore, the delay of exhaust stoichiometric operation, will vary with the gasoline engine's
ability to operate smoothly at these air-fuel ratios.
The delay also will be determined by the exhaust oxygen sensor's ability to properly
function. Modern exhaust oxygen sensors, including both conventional switching and universal
linear sensors, contain heating elements to allow them to maintain proper operating sensor
temperatures and also to be used sooner following a cold start. These internal heating elements
require careful control to prevent any potential thermal shock from water or fuel in the exhaust
stream. The water is generated from the combustion process but also can be present in the
exhaust pipe from condensation of water, particularly during certain ambient temperature and
humidity operating conditions. Generally, cold starts at 20° F only require a short delay to allow
the initial heating of the exhaust manifold to vaporize any combustion water. This period is
followed by an electronically controlled and monitored heating of the sensor. Exhaust oxygen
sensors have been designed to have significant protection from water and are typically fully
operational well before the engine is prepared to use their information.
Generally, within approximately one minute of 20° F cold start operation, combustion
chamber temperatures are at levels that vaporize sufficient amounts of the gasoline fuel to
command exhaust stoichiometric operation of the engine. Also within that minute, exhaust
oxygen sensors should have sufficient time to reach operating temperature with any thermal
issues mitigated, allowing closed loop stoichiometric operation. As stated earlier, operating a
5-7
-------
Final Regulatory Impact Analysis
gasoline-fueled engine at stoichiometry provides the exhaust aftertreatment with oxygen required
for oxidation of HC and CO. Therefore, the amount of time requiring enrichment should be
minimized and closed loop operation of the emission control system should be able to occur as
soon as physically possible.
5.1.1.1.6 Transient Fuel Control
The control of the air-fuel ratio during transient maneuvers (i.e., operator-induced throttle
movement) has dramatically improved with modern hardware and software controls. This is
largely due to the improved accuracy of both the measurement sensors and the fuel delivery
devices, but also refined software modeling of both air flow and physical fuel characteristics.
Tier 2 vehicles have highly accurate sensors that measure changes in air flow to predict and
deliver the appropriate amount of metered fuel. Additionally, the software that interprets these
sensor signals has evolved to predict transient behaviors with much higher accuracy than ever
before. Many of these improvements were necessitated by increases in emission stringency in
the recent Tier 2 and LEV II programs, which were much less tolerant of transient errors that
were acceptable in past emission control systems.
With the recent widespread penetration of electronic throttle controls (ETC), partially in
response to the stringent Tier 2 and LEV II 75° F standards, manufacturers have been able to
further reduce variability of transient errors. ETC applications remove the direct mechanical
connection from the accelerator pedal to the engine. Instead, the pedal is simply a sensor that
reports pedal movement to the engine control module (ECM). The ECM interprets the pedal
movement and provides a corresponding controlled movement of the engine throttle.
Transient air-fuel errors can be minimized through advanced approaches to ETC usage.
This is possible because the electronic controls can better synchronize the introduction of the
transient maneuver and closely match required air and fuel amounts. The controls can be
designed and programmed to prevent most of the transient errors experienced with older cable-
driven mechanical systems. The older mechanical systems resulted in reactionary response to
throttle movements, making it significantly more difficult to deliver precise dynamic air-fuel
control. Since the ETC systems control the actual movement of the throttle, they have the ability
to essentially eliminate transient errors by preceding the throttle movement with appropriate fuel
metering amounts. This is particularly important at colder temperatures (i.e., 20° F cold start)
where transient errors can be exaggerated when the engine is operating rich of stoichiometry.
5.1.1.1.7 Fuel Volatility Recognition
Improved modeling of the effect of fuel properties on engine and emission performance
has eliminated the need for a new sensor. For instance, some manufacturers have successfully
designed software models that can determine the percentage of ethanol in the fuel on which the
vehicle is operating. These "virtual sensor" models take into account information from sources
such as existing sensors and use historical data for the determinations. The models use this
information to adjust many outputs including fuel metering and spark ignition control.
5-8
-------
Final Regulatory Impact Analysis
Currently, manufacturers have active software features that are designed to recognize and
recover from a lean condition that can be a precursor to an engine stall. These features use
different input criteria to identify and actively change the air-fuel ratio when an excessively lean
condition may be occurring. These features may look at control parameters such as engine speed
(RPM), engine manifold absolute pressure (MAP), engine mass air flow (MAP), and even engine
misfire-related information to determine if a fuel metering change should occur.
The approaches described above exemplify possible software-based control designs that
can achieve the desired emission and engine performance characteristics. Manufacturers have
extensive experience designing and implementing software features to identify and react to
specific fuel parameters that are deemed important to engine operation. The ability to recognize
fuel volatility and actively adjust the fuel metering accordingly would allow the gasoline-fueled
engine to operate at the lean limit, reducing engine-out emissions, particularly NMHC and CO.
Much like the "virtual sensor" model described above for ethanol content, this model would take
existing sensor information and other information available from the ECM and determine the
fuel volatility characteristics at any given cold start temperature. The modern engine controllers
have the ability to maintain significant historical data that can help predict fuel properties. The
items of importance for fuel volatility may include ambient temperature exposure of fuel, amount
of time since previous start, and other related items.
5.1.1.1.8 Fuel Injection Timing
Fuel injection timing control is another emission control technology that has evolved as a
result of increased computing power of the engine. Depending on the engine design and the
thermal characteristics of the intake port design, significant opportunity may exist for optimizing
fuel preparation prior to combustion.
Generally, there are two fuel injection timing approaches used to optimize fuel
preparation: closed-valve injection and open-valve injection. Closed valve injection is the
traditional method of injecting fuel into the cylinder head intake port. As the name indicates, the
intake valve is closed during the injection time period. This approach allows the fuel to have
residence time in the intake port prior to ingestion into the cylinder. Usually, the fuel injector is
targeted to spray the fuel on the back of the closed intake valve in order to allow the fuel to
absorb any heat conducted through the valve from the combustion events occurring inside the
cylinder chamber. The heat absorbed by the fuel potentially allows more of the fuel to vaporize
either in the port or in the chamber, resulting in higher percentage of vaporized fuel that can be
combusted. If the higher percentage of vaporized fuel burns, less liquid fuel will be exhausted,
effectively reducing the engine-out NMHC levels.
Open-valve injection involves carefully coordinating the fuel injection timing in order to
inject fuel while the intake valve is in some state of opening. This approach attempts to take
advantage of the incoming air velocity as the air is drawn through the port and also the intake air
pressure depression. The mixture motion and depression can help vaporize the fuel and assist in
better mixing of the air and fuel prior to combustion, resulting in improved fuel burn. This
approach is dependent on many aspects, including injector spray design, injector targeting, intake
valve timing, and intake valve lift. Open-valve timing may be used initially after engine start
5-9
-------
Final Regulatory Impact Analysis
followed by a closed-valve approach, described previously, once the intake valve is heated.
Many similar approaches which can be implemented at any cold start temperature are detailed in
past Society of Automotive Engineers (SAE) papers.2
5.1.1.1.9 Spark Delivery Control
With the increases in the computing power of the engine controller, opportunities have
been created for new spark delivery related emission control features. Separate from the retarded
timing benefits described previously, there are other potential controls that may help reduce
engine-out emissions. Many new engines contain individual cylinder ignition coils. With these
individual coils comes the opportunity for individual cylinder-based spark control features
designed to promote more complete combustion. Additionally, some new engines have dual
spark plugs (i.e., two plugs for each cylinder). These dual spark plug systems may have
opportunities for new concepts targeted at emission reductions, particularly following cold start
operation.
Spark energy, the amount of energy delivered to the spark plug that is used to ignite the
air-fuel mixture, can be carefully controlled by modifying the dwell time delivered to the ignition
coil. The dwell time is the amount of time that the ignition coil is allowed to be charged with
electrical energy. An increase in dwell time will generally result in an increase in spark energy
delivered to the spark plug. Higher spark energy typically results in a higher burn rate
particularly in air-fuel mixtures that are not optimized, which is typical of mixtures at start-up.
Other new concepts may include such ideas as multiple spark events on a single engine
cycle. The concept of delivering redundant spark events has been used in the past, primarily for
engine performance. While we do not currently know if redundant spark events are beneficial in
reducing emissions, it could be explored for emissions control. Similarly, dual spark plug
engines or engines with individual cylinder ignition coils can explore other spark delivery related
concepts that may prove to be effective emission control tools. All of these concepts are equally
feasible at cold temperatures since they are not temperature dependent.
5.1.1.1.10 Universal Oxygen Sensor
As listed in Table 5.1-1 above, universal oxygen sensors were projected to be an emission
control hardware that could be used to meet Tier 2 vehicle standards. Several manufacturers did
in fact decide to replace their conventional switching oxygen sensors with these universal
oxygen sensors. Universal oxygen sensors have certain benefits over conventional switching
sensors that should prove substantially beneficial at 20° F. While these sensors require a similar
delay to reach operating temperature following a start, universal oxygen sensors can accurately
control the air-fuel ratio during rich operating conditions prior to commanded closed loop
operation. Conventional switching sensors cannot indicate the actual air-fuel ratio during rich
conditions, therefore preventing them from being used as a control sensor during critical rich
operation. Additionally, universal oxygen sensors can be used to more accurately recover from
air-fuel transient errors during the warm-up due to their ability to measure the magnitude of the
error.
5-10
-------
Final Regulatory Impact Analysis
5.1.1.2 Tier 2 Engine and Exhaust Control Technologies
The Tier 2 technological feasibility assessment described several engine and exhaust
hardware control technologies that could be used to meet stringent Tier 2 emission standards.3
These technologies continue to be very effective emission control strategies to meet Tier 2
standards. We believe that manufacturers will use these same Tier 2 technologies in order to
meet the 20° F NMHC standard. We do not expect that manufacturers will need to utilize
additional emission control hardware. However, if a manufacturer chose to do so, most of these
same Tier 2 technologies can also be used to meet the 20° F NMHC standard.
5.1.2 Data Supporting Cold NMHC Standard Technical Feasibility
Data to support the feasibility of complying with the 20° F NMHC standard are presented
in the following two sections. The first section includes evidence from recent model year
certification emissions data submitted to EPA. Certification data are required to include cold
temperature carbon monoxide emissions data, and some manufacturers have also included
associated cold temperature total hydrocarbon emissions data. The second section provides
evidence from a feasibility evaluation program recently undertaken by EPA. This program
examined the effects of making only calibration modifications to vehicles with 20° F NMHC
levels that were significantly higher than the standards we are finalizing today.
When considering the supporting data, it should be noted that manufacturers generally
design vehicles to incorporate a compliance margin in their exhaust emissions controls systems
to account for operational variability. For example, manufacturers design controls to meet
emissions targets below the standard when using catalytic converters thermally aged to the full
useful life. By ensuring that emission targets are met when testing on artificially aged
converters, manufacturers reduce the probability that in-use vehicles will exceed the relevant
standard throughout the useful life of the vehicles. Put another way, this design attempts to
account for maximum normal operating variability.
However, the data presented in the following sections do not explicitly incorporate a
compliance margin since the cold temperature NMHC data, at the time they were submitted to or
tested by the EPA, were not subject to cold NMHC standards. The data represent the cold
NMHC emissions as tested, and suggest that a significant number of vehicles are within reach of
the standards we are finalizing today.
5.1.2.1 Certification Emission Level
Currently, manufacturers are required to report carbon monoxide (CO) exhaust emissions
test results for compliance with cold temperature CO standards (i.e., the 20° F FTP test) for light-
duty vehicles and light-duty trucks. (Cold CO requirements for medium-duty passenger vehicles
do not begin until model year 2008.4) Many manufacturers have included total hydrocarbon
(THC) cold temperature exhaust emission data that are collected along with cold CO data. In
addition, several of these manufacturers also reported test results for both the THC emission data
and the matching NMHC emission data. Based on these data from manufacturers who have
included both THC and NMHC cold temperature data, non-methane hydrocarbons (NMHCs)
5-11
-------
Final Regulatory Impact Analysis
account for approximately 95% of total hydrocarbon emissions at cold temperatures. Therefore,
a review of the more abundant THC data provides a reasonable means of assessing
manufacturers' cold NMHC emissions performance.
EPA analyzed 2004-2007 model year full useful life certification data for vehicles
certified to nationwide Tier 2 standards, interim non-Tier 2 standards, and California program
standards. Lists were compiled from certification data submissions that reported unrounded cold
THC results and for which an associated FTP full useful life deterioration factor (DF) was
available. The DF was incorporated into the emissions result to estimate emissions at the full
useful life of the vehicle. The DF was applied to the unrounded test result, and that result was
rounded to one decimal point. This calculation was then compared to the cold temperature
NMHC standards of 0.3 g/mi for LDV/LLDTs, and 0.5 g/mi for HLDT/MDPVs.
Table 5.1-2 shows the number of car lines for which the resulting calculation for total
hydrocarbons was at or below the 0.3 g/mi NMHC standard for LDV/LLDTs, and at or below
the 0.5 g/mi NMHC standard for HLDT/MDPVs. Again, these data only reflect an analysis of
those car lines for which manufacturers voluntarily provide cold THC data.
Tables 5.1-3 through 5.1-6 show, by model year, the total hydrocarbon emission levels
(calculated according to the method described above) for LDV/LLDTs at or below 0.3 g/mi, and
HLDT/MDPVs at or below 0.5 g/mi. For each manufacturer, the data were grouped according to
car lines with the same calculated cold THC emission result. Where a range is shown for the
emission level, tests on multiple configurations within the car line yielded a range of results.
Table 5.1-2. Number of Car Lines with one or more Engine Families whose Certification
Data for Total Hydrocarbons was at or below the Proposed Cold NMHC Standards
Year
2004
2005
2006
2007
LDV/LLDTs
41
42
44
39
HLDT/MDPVs
-------
Final Regulatory Impact Analysis
vehicles in the certification data are interim non-Tier 2 vehicles. We would expect hydrocarbon
levels to be somewhat lower as these vehicles fully phase-in to Tier 2.
Table 5.1-3. 2004 Model Year Vehicles with Certification Data
for Total Hydrocarbons at or below the Cold NMHC Standard
MANUFACTURER
CAR LINE
COLD TOTAL HC LEVEL
LDV/LLDTs
ACURA
ACURA
ACURA
AUDI
BMW
BMW
CADILLAC
CHEVROLET
HONDA
HONDA
HONDA
HONDA
HONDA
HONDA
HYUNDAI
MAZDA
MAZDA
MERCEDES-BENZ
MERCEDES-BENZ
MITSUBISHI
MITSUBISHI
NISSAN
NISSAN
SATURN
TOYOTA
TOYOTA
VOLKSWAGEN
VOLVO
1.7EL, TL
MDX 4WD
RSX
A4 QUATTRO
3251 SPORT WAGON, 330CI CONVERT.
X3
CTS
CORVETTE
ACCORD
CIVIC
CIVIC HYBRID, INSIGHT
CR-V 4WD, ELEMENT 4WD, S2000
ODYSSEY 2WD
PILOT 4WD
XD-5DR
MAZDA 3
MAZDA 6, MAZDA 6 SPORT WAGON, MPV
C240 (WAGON), C-CLASS SEDAN/WAGON, S-CLASS
E320 4MATIC (WAGON), S500 (GUARD)
GALANT
LANCER SPORTBACK
ALTIMA
SENTRA
VUE AWD
CAMRY
PRIUS, RAV4 4WD
JETTA, JETTA WAGON, BEETLE CONVERT.
V70
0.1
0.2
0.3
0.3
0.1
0.2
0.2
0.2
0.1 -0.3
0.1-0.2
0-0.1
0.2
0.3
0.2-0.3
0.3
0.2-0.3
0.3
0.3
0.2
0.1-0.2
0.3
0.3
0.2-0.3
0.2
0.3
0.2
0.2
0.2-0.3
HLDT/MDPVs
BENTLEY
BMW
CHEVROLET
CHEVROLET
GMC
HIREUS
MERCEDES-BENZ
PORSCHE
ROLLS-ROYCE
VOLKSWAGEN
VOLVO
CONTINENTAL GT
X5
ASTRO AWD(C) CONV
Kl 5 SLV HYBRID 4WD
K1500 SIERRA AWD
RR01
G500, ML350
CAYENNE, CAYENNE S
PHANTOM
TOUAREG
XC90
0.3
0.3
0.5
0.4
0.4
0.3
0.4
0.3
0.3
0.4
0.3,0.5
5-13
-------
Final Regulatory Impact Analysis
Table 5.1-4. 2005 Model Year Vehicles with Certification Data
for Total Hydrocarbons at or below the Cold NMHC Standard
MANUFACTURER
CAR LINE
COLD TOTAL HC LEVEL
LDV/LLDTs
ACURA
ACURA
AUDI
BMW
BMW
BUICK
CADILLAC
HONDA
HONDA
HONDA
HONDA
HONDA
HYUNDAI
HYUNDAI
HYUNDAI
MAZDA
MAZDA
MERCEDES-BENZ
MERCEDES-BENZ
MERCEDES-BENZ
MITSUBISHI
MITSUBISHI
NISSAN
SATURN
SATURN
TOYOTA
TOYOTA
VOLKSWAGEN
1.7EL, MDX4WD
RL, RSX
A4 QUATTRO
3251 SPORT WAGON, 330CI CONVERTIBLE
X3
LACROSSE/ALLURE
CTS
ACCORD
ACCORD HYBRID
CIVIC
CIVIC HYBRID
CR-V 4WD, ODYSSEY 2WD, S2000
JM(2WD)
JM(4WD)
XD-5DR
MAZDA 3
MPV
C240 (WAGON), C32 AMG, E320 4MATIC (WAGON), S55 AMG
C320
S430 4MATIC
GAL ANT
LANCER, LANCER SPORTBACK
SENTRA
RELAY AWD
VUE AWD
CAMRY, SCION XB
PRIUS, RAV4 4WD
JETTA, JETTA WAGON, BEETLE CONVERT., V70
0.1
0.2
0.3
0.1
0.2
0.3
0.2
0.1 -0.2
0.2
0.1 -0.2
0-0.1
0.2
0.3
0.2
0.3
0.2-0.3
0.2
0.3
0.2
0.1
0.2-0.3
0.3
0.2
0.3
0.2
0.3
0.2
0.2
HLDT/MDPVs
BENTLEY
BMW
CHEVROLET
CHEVROLET
GMC
LAND ROVER LTD
LEXUS
MERCEDES-BENZ
MERCEDES-BENZ
PORSCHE
ROLLS-ROYCE
TOYOTA
VOLVO
CONTINENTAL GT
X5
ASTRO AWD(C) CONV, C2500 SLVRADO 2WD, K1500 SUB'N 4WD
K15SLV HYBRID 4WD
G3500 SAVANA(P), K1500 SIERRA AWD
LR3
GX470
G500, ML350
G55 AMG
CAYENNE
PHANTOM
TOYOTA TUNDRA 4WD
XC90
0.3
0.3
0.5
0.4
0.4
0.4
0.4
0.4
0.2
0.3
0.3
0.5
0.3
5-14
-------
Final Regulatory Impact Analysis
Table 5.1-5. 2006 Model Year Vehicles with Certification Data
for Total Hydrocarbons at or below the Cold NMHC Standard
MANUFACTURER
CAR LINE
COLD TOTAL HC LEVEL
LDV/LLDTs
ACURA
ACURA
AUDI
BUICK
CADILLAC
CHEVROLET
CHRYSLER
HONDA
HONDA
HONDA
HONDA
HONDA
HYUNDAI
HYUNDAI
LEXUS
MAZDA
MAZDA
MERCEDES-BENZ
MERCEDES-BENZ
MERCEDES-BENZ
MITSUBISHI
MITSUBISHI
NISSAN
SATURN
SATURN
SUZUKI
TOYOTA
VOLKSWAGEN
VOLKSWAGEN
VOLVO
MDX 4WD
RL, RSX
A4 QUATTRO
LACROSSE/ALLURE
CTS
COBALT, IMP ALA
TOWN & COUNTRY 2WD
ACCORD
CIVIC, CR-V 4WD, ODYSSEY 2WD
CIVIC HYBRID
INSIGHT
S2000
JM(2WD), XD-4DR/5DR
JM(4WD)
GS 300 4WD, RX 400H 4WD
MAZDA 3, MAZDA 5, MPV
MAZDA 6, MAZDA 6 SPORT WAGON
B200 TURBO, S350
S430 4MATIC
S55 AMG
GAL ANT
LANCER, LANCER SPORTBACK
ALTIMA, SENTRA
RELAY AWD
VUE AWD
FORENZA WAGON
CAMRY, CAMRY SOLARA, YARIS
JETTA WAGON
PASSAT WAGON
V70
0.1
0.2
0.3
0.3
0.3
0.3
0.3
0.1 -0.2
0.2
0.1
0-0.1
0.3
0.3
0.2
0.3
0.2
0.3
0.2
0.1
0.3
0.2-0.3
0.3
0.3
0.3
0.2
0
0.3
0.2
0.3
0.2
HLDT/MDPVs
CADILLAC
CHEVROLET
CHEVROLET
DODGE
GMC
GMC
HONDA
JEEP
LAND ROVER LTD
LEXUS
LEXUS
MERCEDES-BENZ
FUNERAL COACH/HEARS, SRX AWD
C2500 SLVRADO 2WD
K15SLV HYBRID 4WD
DAKOTA PICKUP 4WD, RAM 1500 PICKUP 2WD
ENVOY XUV 4WD, G1525 SAVANA CONV
Kl 5 YUKON XL AWD
RIDGELINE 4WD
GRAND CHEROKEE 4WD
LR3
GX470
LX470
R500
0.5
0.5
0.3
0.5
0.5
0.3
0.2
0.4
0.5
0.4
0.5
0.2
5-15
-------
Final Regulatory Impact Analysis
PORSCHE
PORSCHE
ROLLS-ROYCE
TOYOTA
VOLKSWAGEN
VOLVO
CAYENNE, CAYENNE S
CAYENNE TURBO KIT
PHANTOM
TOYOTA TUNDRA 4WD
PHAETON
XC90
0.3
0.5
0.3
0.5
0.5
0.3
5-16
-------
Final Regulatory Impact Analysis
Table 5.1-6. 2007 Model Year Vehicles with Certification Data
for Total Hydrocarbons at or below the Cold NMHC Standard
MANUFACTURER
CAR LINE
COLD TOTAL HC LEVEL
LDV/LLDTs
BMW
DAIMLER-CHRYSLER
GENERAL MOTORS
GENERAL MOTORS
GM/DAEWOO
HONDA
HONDA
HONDA
HONDA
HYUNDAI
LAMBORGHINI
LOTUS
MAZDA
MAZDA
MAZDA
MERCEDES-BENZ
MERCEDES-BENZ
MITSUBISHI
MITSUBISHI
PORSCHE
TOYOTA
TOYOTA
TOYOTA
VOLKSWAGEN
VOLVO
335X1
CROSSFIRE ROADSTER
IMPALA, UPLANDER FWD
VUE AWD
FORENZA WAGON
ACCORD HYBRID
ACCORD
CIVIC, CIVIC HYBRID, ELEMENT 4WD, S2000
CR-V 4WD, FIT, MDX 4WD, ODYSSEY 2WD, RDX 4WD, RL, TL
HD-4DR, MC(3/4DR)
GALLARDO COUPE, MURCIELAGO
ELISE/EXIGE
MAZDA 3
MAZDAS
MAZDA 6, MAZDA 6 SPORT WAGON
SL55 AMG
CLK550 (CABRIOLET)
GALANT
ECLIPSE SPYDER
CAYMAN S
CAMRY HYBRID
PRIUS
YARIS, RX 400H 4WD
PASSAT WAGON
S40
0.3
0.2
0.3
0.2
0
0.1
0.1-0.2
0.3
0.2
0.3
0.1
0.3
0.2-0.3
0.2
0.3
0.2
0.3
0.2-0.3
0.3
0.2
0.1
0.2
0.3
0.3
0.1
HLDT/MDPVs
BENTLEY MOTORS LTD.
DAIMLER-CHRYSLER
GENERAL MOTORS
GENERAL MOTORS
GENERAL MOTORS
HONDA
MERCEDES-BENZ
MERCEDES-BENZ
TOYOTA
TOYOTA
VOLKSWAGEN
VOLVO
CONTINENTAL GTC
RAM 1500 PICKUP 2WD
K1500 AVALANCHE 4WD
CHEVY C1500 CLASSIC PICKUP 2WD, ISUZU ASCENDER SUV
4WD, FULL SIZE CONVERSION VAN AWD
ESCALADE EXTAWD
RIDGELINE 4WD
R500 4MATIC
MAYBACH 62, GL450
GX470
LX 470, TOYOTA TUNDRA 4WD
TOUAREG
XC90
0.3
0.4-0.5
0.3-0.4
0.4
0.5
0.2
0.2
0.5
0.4
0.5
0.5
0.3, 0.5
5-17
-------
Final Regulatory Impact Analysis
5.1.2.2
EPA Test Programs
To further assess the feasibility of meeting the new cold temperature NMHC standards
through changes to engine and emission control system calibration, EPA performed a test
program involving two Tier 2 vehicles. We considered several key aspects when selecting test
vehicles for a feasibility study. First, the vehicles currently produce 20° F NMHC levels that are
significantly higher than the standards we are finalizing today. Second, since higher vehicle
weight poses additional challenge, we considered heavier GVWR vehicles. Finally, the
technological approach chosen by the manufacturer to meet stringent 75° F Tier 2 standards was
also considered. Specifically, we considered secondary air injection technology and close
coupled catalyst technology. Specifications for the vehicles included in the test program are
provided in Table 5.1-6.
Table 5.1-6. EPA Test Vehicle Specifications
Vehicle
2004
Chevrolet
Trailblazer
2006
Chrysler
300C
Engine Family
4GMXT04.2185
6CRXV05.7VEO
Powertrain
4.2L 16
4-speed auto
Rear 2-WD
5.7LV8
5-speed auto
Rear 2-WD
GVWR
5550 Ibs.
5300 Ibs.
Emission
Class
Tier 2 Bin 5
Tier 2 Bin 5
Mileage
36,500
2,000
The vehicles were tested at 20° F following EPA cold FTP test procedures established in
40 CFR 86.230-94. In addition to regulated pollutant measurements, we also measured NMHC,
NOx, and direct particulate matter (PM). NMOG analysis also produced measurements of 13
carbonyls. PM measurement was performed following 40 CFR 86.110-94 procedures. A
detailed diagram of the emission and PM sampling system can be seen in the docket.A The road
load force target coefficient settings, contained in Table 5.1-7, are 10% higher than the vehicle's
75° F target coefficients as established procedure in EPA guidance letter CD-93-01.B
Table 5.1-7. EPA 20° F Cold Test Vehicle Settings
Vehicle
2004 Chevrolet
Trailblazer
2006 Chrysler
300C
Test Weight
5000 Ibs.
4500 Ibs.
20° F Target
Coefficients
A=38.97
B=1.2526
C=.02769
A=61.09
B=.3105
C=.0247
A "Cold Chamber Sampling System Diagram," PDF file from test lab.
B Available at www.epa.gov/otaq/cert/dearmfr/dearmfr.htm.
5-18
-------
Final Regulatory Impact Analysis
5.1.2.2.1 2004 Chevrolet Trailblazer Testing
The 2004 Chevrolet Trailblazer was chosen as a test vehicle for several reasons. First, it
is certified as a Tier 2 Bin 5 package, which represents what can be considered the "typical" or
average 75° F emission level once Tier 2 phase-in is complete. The Bin 5 emission standards
represent the required EPA fleet average for NOx, and therefore the hardware used on the
Trailblazer to comply with Bin 5 standards represents what we might expect from many
manufacturers and vehicle lines. Second, while it was certified to the expected average Tier 2
emission levels, its NMHC emission performance at 20° F was substantially worse than the
industry averages. Finally, due to its GVWR, it represents vehicles that are very close to 6000
Ibs. GVWR (i.e., the HLDT emission category). Different Trailblazer models fall above and
below 6000 Ibs. GVWR, but have no discernable differences in the emission control hardware.
The Trailblazer engine control system is representative of typical Tier 2 systems. The
system includes an electronic engine control module (ECM), individual cylinder fuel injectors,
individual cylinder ignition coils, heated exhaust gas oxygen sensors (HEGO) before and after
the catalyst, electronic throttle control, variable valve timing and several other necessary
supporting sensors. The aftertreatment hardware consists of a single, under-floor catalyst and a
secondary air injection system.
The secondary air injection system is composed of an electric air pump and an electric
solenoid valve. The air pump is located under the vehicle's driver-side floor board where it is
mounted to a frame bracket. The electric solenoid valve is mounted to the engine cylinder head
directly above the exhaust manifold on the passenger side of the vehicle. Clean air is drawn by
the air pump from the air cleaner assembly in the engine compartment through a pipe, and then it
is pumped back to the electric solenoid valve through a second pipe. The two pipes used to
transport the air are fairly long, due primarily to the air pump location.
The secondary air injection system on the Trailblazer appears to operate on cold starts
above 40° F only. The system operates for approximately 20 to 45 seconds after the start,
depending on start-up coolant temperature, and is deactivated when the emission control system
goes into closed loop operation. Some manufacturers have indicated that operation of the
secondary air injection system is not currently performed on cold start temperatures at and below
freezing due to potential ice issues. However, this is not universal across all manufacturers,
since several manufacturers do, in fact, operate their secondary air injection system at 20° F cold
start temperatures and above. They have addressed the issue of water collecting and freezing by
design aspects primarily concentrated around system plumbing and location of the components.
On some European vehicle models, these manufacturers effectively use the secondary air
injection systems to comply with a 20° F NMHC standard in Europe.5
A key element of the feasibility test program was to imitate software initiated emission
control system behaviors which are observed at the currently regulated start temperatures of 75°
F and 50° F (California-only requirement). These software features do not entail any new
hardware. In the case of the Trailblazer, while not all behaviors could be demonstrated, several
of the most important behaviors were replicated. First, the secondary air injection system was
5-19
-------
Final Regulatory Impact Analysis
operated during the cold temperature test. Second, elevated idle speeds, similar to what the
Trailblazer currently uses after the start at the regulated start temperatures, were also used at cold
temperature.
Activation of the secondary air injection was accomplished through circuit overrides of
the air pump and solenoid valve control circuits, completely external to the ECM. The air pump
and the solenoid valve are each powered by a relay normally only controlled by the ECM output
signals. The two relays were forced on to activate the secondary air injection system during the
desired period following the cold start. We tested several delay periods from the start of the
engine until the secondary air system was activated in order to measure benefits of earlier
introduction of the air injection. The secondary air was always run until the ECM induced
closed loop operation (approximately 60 seconds after the start). At the completion of the
desired period of operation, control of the relays was returned to the ECM.
The elevated idle speed was performed by allowing a manually controlled vacuum leak
into the intake manifold during the first 30 to 60 seconds following engine start. The controlled
vacuum leak targeted 1550 to!600 RPM idle speed in park/neutral, mimicking the same
acceptable idle speed the ECM commands at 50° F cold starts. Typically, idle speeds increase
with drops in start temperature, but the observed desired idle speeds in the Trailblazer were
lower at 20° F (1350 RPM) than at the warmer 50° F starts (1550 RPM). An On-board
Diagnostics 2 (OBD2) dealership diagnostic tool was capable of electronically elevating the idle
speed to the equivalent 50° F cold start idle speed but internal tool or ECM software prevented
this idle speed and air flow for the full 30 to 60 second time period. Hence it was determined
that for the purposes of this test program, a vacuum leak would accurately demonstrate the effect
of an acceptable idle speed and air flow following a start. Manufacturers today control to a
desired idle speed through control of electronic throttle or other air bleed devices.
Table 5.1-8 below contains the weighted test total (3 bags) emission results of the
different test configurations attempted on the Trailblazer. Test #7 and #8 also included defroster
operation starting at 130 seconds into the test and remaining on for the rest of the test. Since the
methods used to control cold start NMHC emissions were used only in the first minute of
operation, prior to defroster activation, the NMHC and PM emission results with defroster
operation remain representative of emission control opportunities. It is important to note the
consistent reductions in NMHC with early activation of the secondary air injection system, as
seen in the test sequence from test #3 through test #6, and also in the defroster tests. The tests
with defroster operation were included to assess any emission impacts of defroster-on, which is
required in the fuel economy rule.c
While NOx emissions are not part of the controls investigation, the NOx levels appeared
to increase with the NMHC control methods. After some modal investigation, it was determined
that the NOx increases were occurring after the NMHC controls had performed the majority of
their benefits. The NOx emissions were brought back almost to the baseline levels by shortening
the elevated idle speed and air bleed time. This can be observed in the results of test #6 and #7.
In fact, test #6 produced the largest NMHC reduction with essentially the same NOx levels as the
baseline tests. Manufacturers would be able to better synchronize their controls through their
Fuel Economy Final Rule, 71 FR 77872, December 27, 2006, Defroster Operation Requirement for Cold FTP.
5-20
-------
Final Regulatory Impact Analysis
ECM to control NMHC and NOx emissions simultaneously, as compared to this test program's
limitations.
CO and direct PM measurements were also significantly reduced when NMHC controls
were activated. CO, the only currently regulated pollutant at 20° F, was consistently reduced
from baseline levels with each of the control combinations. PM was also generally reduced;
however, this is less obvious when reported as test total results. Since the emissions are recorded
over the three-phase test with each phase composed of an individual bag measurement, PM
reductions can be better evaluated in Table 5.1-9, which contains the emission results for only
the first phase (bag 1) of the three-phase emission test.
Table 5.1-8. Trailblazer Test Configuration and 20° F FTP Weighted Test Total
Results
Test
Number
Air
Injection
Elevated Idle &
air bleed time
Standard < 6000 Ibs GVWR
Standard > 6000 Ibs GVWR
1 -baseline
2-baseline
3 -controls
4-controls
5 -controls
6-controls
7-defrost on
8-defrost on
none
none
5 s delay
2 s delay
1 s delay
0 s delay
1 s delay
0 s delay
none
none
60s
60s
60s
30s
30s
45 s
NMHC
g/mi
.3
.5
1.08
1.03
.59
.42
.35
.29
.38
.32
CO
g/mi
7.8
9.5
5.2
5.5
5.2
5.1
6.9
6.4
NOx
g/mi
.05
.04
.15
.19
.17
.06
.08
.13
PM
g/mi
.024
.015
.025
.013
.014
.013
.012
.013
Fuel Economy
mi/gallon
13.82
13.64
13.87
13.56
13.71
13.64
13.17
13.25
As can be seen in Table 5.1-8, control test #6 provided a NMHC level that would have
allowed the Trailblazer to comply with the standard for the < 6000 Ibs GVWR class (i.e.,
0.3g/mi). While this vehicle was tested as the lower GVWR class at 5000 Ibs test weight, the
Trailblazer also is sold as an over 6000 Ibs. GVWR model that would have been tested at 5500
lbs°. We believe that if tested at the higher weight, the emission results likely would not have
increased much, reflecting a large margin (.2 g/mi) for this vehicle when certified to the heavier
class. We recognize that manufacturers will need to account for a compliance margin, but we
believe this vehicle can achieve a comfortable compliance margin for the more stringent standard
(i.e., 0.3 g/mi) with some additional minor calibration changes. We also recognize that this
feasibility study does not constitute a production calibration and that additional development
effort would be needed to achieve manufacturer functional objectives for cold starts. This test
program simply demonstrates that in the case of this typical secondary air injection equipped
vehicle, additional emission reduction opportunities exist without the requirement for additional
hardware.
D Tier 2 vehicles are tested (inertia setting) at loaded vehicle weight (LVW). LVW=curb weight + 300 Ibs.
5-21
-------
Final Regulatory Impact Analysis
Emissions results for the 20° F cold CO test are reported as a weighted three-bag average.
However, bag one (the first 505 seconds of the test) provides a better indication of emission
reductions achieved with controls, because almost all of the emissions at 20° F are emitted in the
first few minutes of operation, and all control changes were attempted only during the first
minute of operation. Table 5.1-9 presents only the bag 1 emission results. This table highlights
the emission reductions from the control changes by not diluting the improvements over the
second and third phase (bag 2 and 3) of the emission test.
As observed below in Table 5.1-9, the results clearly show NMHC, CO and PM
reductions. NMHC and CO reductions occur with all the control attempts, but control tests #6
and #8, in which secondary air injection was activated immediately upon engine cranking,
achieve the best results. PM reductions follow similar behavior as NMHC, but they appear to be
very sensitive to delayed secondary air injection.
Table 5.1-9. Trailblazer Test Configuration and 20° F FTP Phase 1 Only Results
Test
Number
1 -baseline
2-baseline
3 -controls
4-controls
5 -controls
6-controls
7-defrost on
8-defrost on
Air
Injection
none
none
5 s delay
2 s delay
1 s delay
0 s delay
1 s delay
0 s delay
Elevated Idle &
air bleed time
none
none
60s
60s
60s
30s
30s
45 s
NMHC
g/mi
5.18
4.92
2.81
1.96
1.63
1.34
1.75
1.47
CO
g/mi
27.3
31.7
18.6
15.0
13.6
13.3
14.8
13.2
NOx
g/mi
.22
.16
.72
.85
.81
.29
.35
.61
PM
g/mi
.055
.040
.043
.033
.026
.022
.010
.022
Fuel Economy
mi/gallon
11.55
11.47
11.29
11.30
11.40
11.45
11.23
11.27
While the emissions reductions were fairly substantial with the best control combination
in test #6, we believe that even greater emission reductions can be achieved with more precise
use of the secondary air system and additional control measures described earlier in the
calibration and controls technology section. The ability to more precisely provide the ideal air-
fuel mixture for the secondary air injection system likely would have resulted in faster catalyst
light-off and subsequently even greater reductions in emissions, especially NMHC.
Additionally, retarded timing was not tested due to the limited capability to modify engine
operation. Typically this would further compound the rate of heating the catalyst, particularly on
secondary air injection systems, and thus, would be expected as an additional opportunity to
reduce NMHC.
5.1.2.2.2
2006 Chrysler 300C Testing
A second vehicle, a 2006 Chrysler 300C, was chosen because it has specific engine
related challenges and a different method of controlling emissions than the first feasibility
5-22
-------
Final Regulatory Impact Analysis
vehicle. Manufacturers have commented that the higher displacement engines typically used in
heavier vehicles pose additional challenges because they require more fuel to start and maintain
idle stability at cold temperatures than smaller displacement engines.6 The 300C is a light-duty
vehicle equipped with a large displacement V8 engine (5.7 liter). The same large displacement
V8 engine used in the 300C is also found in a broad range of vehicles, including full size trucks
and several sport utility vehicles above 6000 Ibs. GVWR. Unlike the Trailblazer, it does not use
secondary air injection, and instead relies more heavily on controlling fuel in the combustion
chamber. Like the Trailblazer, it is certified as a Tier 2 Bin 5 package. This represents what can
be considered the "typical" or average 75° F emission level once Tier 2 phase-in is complete.
The hardware used on the 300C to comply with Tier 2 Bin 5 standards also represents what we
might expect from many manufacturers and vehicle lines that do not have secondary air
injection.
The 300C emission control system is representative of typical Tier 2 systems. The
system includes an electronic engine control module (ECM), individual cylinder fuel injectors,
individual cylinder ignition coils, heated exhaust gas oxygen sensors (HEGO) before and after
the catalysts, electronic throttle control, and several other necessary supporting sensors. The
aftertreatment hardware consists of a dual catalysts (one per bank) located in close proximity to
the engine exhaust manifold for optimum catalyst heating. Additionally, the 300C contains
several new technologies, including cylinder deactivation and dual spark plugs per cylinder.
The 300C is offered as both a US model (sold in North America) and a European model
(sold in Europe) without any discernable differences in the emission control hardware (i.e.,
catalysts, oxygen sensors, etc.), with the exception of the ECM. Vehicles sold in the European
market are required to comply with the European Union (EU) type VI test, which is a cold start
test performed at 20° F. In addition to CO emission standards, the EU cold temperature test also
requires stringent HC and NOx emission control. For this reason, any differences observed in
the emissions between a US model and a European model are likely due to software and
calibration differences targeted specifically at HC, CO and NOx emission reductions to meet the
European cold temperature emissions standards.
These emission controls could be implemented in the US model 300C and other US
packages also sharing the same large displacement V8 engine. While all the different
applications were not directly tested, the control techniques used in the 300C to reduce cold start
emissions for EU compliance should also be applicable to the other vehicle platforms that share
the same engine. In fact, some of the other vehicle platforms that share the same V8 engine are
also sold in Europe, which indicates that emission control techniques are likely already leveraged
across various European models sharing a common engine.
In order to determine any emission control opportunities that exist in the European 300C,
an ECM containing the 2006 European software and calibration was purchased and installed in
the 300C. No other changes were made to the emission control system. The US and European
model configurations were each tested over the 20° F FTP. The tests were replicated to ensure
that the emissions levels were consistent. The testing was performed with defroster operation as
specified in the fuel economy ruleE to incorporate any potential emission impacts of defroster or
'• Fuel Economy Final Rule, 71 FR 77872, December 27, 2006, Defroster Operation Requirement for Cold FTP.
5-23
-------
Final Regulatory Impact Analysis
heater-on operation. The 300C utilizes an automatic interior climate control system which, as
expected, did not ramp up fan speed until the coolant temperature exceeded a threshold
established for driver comfort. Since the methods used to control cold start NMHC emissions
are primarily used only in the first minute of operation prior to the ramp-up of the automatically
controlled interior fan, the emission results are likely not impacted by defroster operation.
However, as with the first feasibility vehicle test program, it was logical that the testing be
performed with the defroster usage as specified in the fuel economy rule to include any potential
emissions impact.
Table 5.1-10 below contains the weighted test total (3 bags) emission results of the two
different test configurations attempted on the 300C. All emissions were significantly reduced
when tested in the European configuration. CO, the only currently regulated pollutant at 20° F in
the US, was consistently reduced below baseline levels with each test of the European
calibration. In the European model, NMHC, was reduced 32% reduced when compared to the
US model. The European models NMHC levels approached the new fleet average standard of 0.3
g/mi for lighter vehicles under 6000 Ibs. GVWR. The 300C is considered a worst-case model in
the LDV category due to its large displacement V8 engine and high test weight. With the fleet
averaging provision provided in the final rule, this vehicle would likely be certified to a higher
PEL. However, we fully expect that with the lead time provided, further calibration
improvements to reduce NMHC levels to the 0.3 g/mi level are possible but we did not attempt
this capability in the limited test program.
Table 5.1-10. 300C Test Configuration and 20° F FTP Weighted Test Total Results
Test Number
Standard < 6000 Ibs GVWR
Standard > 6000 Ibs GVWR
1-US model
2-US model
Average
1-EU model
2-EU model
Average
% Change
NMHC
g/mi
0.3
0.5
0.574
0.540
0.557
0.398
0.384
0.379
-32%
CO
g/mi
2.25
1.99
2.12
1.26
1.36
1.23
-42%
NOx
g/mi
0.06
0.07
0.07
0.04
0.05
0.04
-43%
PM
g/mi
0.011
0.009
0.010
0.005
0.006
0.006
-40%
Fuel
Economy
mi/gallon
14.90
14.89
14.90
14.24
14.93
14.64
-2%
While this vehicle was tested at 4500 Ibs test weight representing the lower GVWR class,
as noted above, the same engine configuration is also sold in several over 6000 Ibs. GVWR
models. Vehicles over 6000 Ibs. GVWR are required to test at slightly higher weights (i.e., 5000
5-24
-------
Final Regulatory Impact Analysis
Ibs. to 6000 lbs.)F. We expect that if tested at the higher test weight required for vehicles over
6000 Ibs. GVWR, the emission results would likely increase but would maintain an acceptable
margin below the 0.5 g/mi standard for heavier vehicles. As indicated previously, we believe
this vehicle can likely be brought into compliance with the LDV/LLDT fleetwide standard (i.e.,
0.3 g/mi) with some additional calibration changes, given the lead time provided in the program.
Since almost all of the emissions at 20° F are emitted in the first few minutes of operation
and mostNMHC controls are attempted only during the first few minutes of operation, Table
5.1-11 presents only the bag 1 emission results.
Table 5.1-11. 300C Test Configuration and 20° F FTP Bag 1 Only Results
Test Number
1-US model
2-US model
Average
1-EU model
2-EU model
Average
% Change
NMHC
g/mi
2.75
2.49
2.62
1.897
1.830
1.864
-39%
CO
g/mi
10.4
9.1
9.8
5.6
6.0
5.8
-41%
NOx
g/mi
0.14
0.20
0.17
0.10
0.10
0.10
-41%
PM
g/mi
0.037
0.033
0.035
0.016
0.021
0.019
-54%
Fuel Economy
mi/gallon
13.26
13.41
13.34
12.42
12.74
12.58
-6%
As observed above in Table 5.1-11, all measured emissions with the European
configuration were significantly reduced in bag 1 of the 20° F FTP test. NMHC, CO, NOx and
PM reductions can be clearly seen from the bag 1 results. As described earlier, these reductions
are attributed entirely to calibration and software changes in the European model as all other
emission control hardware remained the same with the US model. These calibration and
software changes could easily be adapted to the US model to achieve significant emissions
reductions. Additional emission reduction opportunities may exist with further software and
calibration refinement. We believe this test program demonstrates that significant emissions
reductions are available through only calibration on the most challenging Tier 2 vehicles (i.e.,
300C) and that most Tier 2 vehicles can achieve these standards. While the standards will be
challenging for some manufacturers across their product lines, we believe the lead time and other
program flexibilities provided will allow compliance with the new standards.
5.2 Feasibility of Evaporative Emissions Standards for Vehicles
Tier 2 vehicles are tested (inertia setting) at loaded vehicle weight (LVW). LVW=curb weight + 300 Ibs.
5-25
-------
Final Regulatory Impact Analysis
The standards for evaporative emissions, which are equivalent to the California LEV II
standards, are technologically feasible now. As discussed in Section V of the preamble for
today's rulemaking, the California LEV II program contains numerically more stringent
evaporative emissions standards compared to existing EPA Tier 2 standards, but because of
differences in testing requirements, we believe the programs are essentially equivalent. This
view is supported by manufacturers and current industry practices. (See Section V.C.5 of the
preamble for further discussion of such test differences (e.g., test temperatures and fuel
volatilities).) A review of recent model year certification results indicates that essentially all
manufacturers certify 50-state evaporative emission systems.7 Therefore, harmonizing with
California's LEV-II evaporative emission standards will codify the approach manufacturers have
already indicated they are taking for 50-state evaporative systems.
5-26
-------
Final Regulatory Impact Analysis
References for Chapter 5
1 Memo to docket "Discussions Regarding Secondary Air System Usage at 20° F with European
Automotive Manufacturers and Suppliers of Secondary Air Systems," December 2005.
2 Meyer, Robert and John B. Heywood, "Liquid Fuel Transport Mechanisms into the Cylinder of
a Firing Port-Injected SI Engine During Start-up," SAE 970865, 1997.
3
For a more detailed description of these technologies see the Tier 2 final rule at 65 FR 6698-
6822, February 10,2000, and Regulatory Impact Analysis Chapter IV: Technical Feasibility.
4 40 CFR 86.1811-04, paragraphs (g) and (k).
5 Memo to docket "Discussions Regarding Secondary Air System Usage at 20° F with European
Automotive Manufacturers and Suppliers of Secondary Air Systems," December 2005.
6 Comments submitted by Alliance of Automobile Manufacturers (Alliance), OAR-2005-0036-
0881.1
7 Update for FRM: U.S. EPA, Evaporative Emission Certification Results for Model Years 2004
to 2007, Memorandum to Docket EPA-HQ-OAR-2005-0036 from Bryan Manning, January 4,
2007.
5-27
-------
Final Regulatory Impact Analysis
Chapter 6: Table of Contents
Chapter 6: Feasibility of the Benzene Control Program 3
6.1 Overview of Refinery Flow 3
6.2 What are the Benzene Levels in Gasoline Today? 6
6.3 Where Does Gasoline Benzene Come From? 16
6.3.1 How Do Reformers work? 17
6.3.2 How Can Benzene Levels be Reduced in Gasoline? 22
6.3.2.1 Pre-Fractionation to Reroute Benzene Precursors 22
6.3.2.2 Benzene Saturation via Isomerization 24
6.3.2.3 Reformate Post-Fractionation with Benzene Saturation 25
6.3.2.4 Benzene Extraction 27
6.3.2.5 Low-Pressure Reformer Operation 29
6.3.2.6 Pre-fractionation Combined with Low-Pressure Reformer Operation 30
6.4 Experience Using Benzene Control Technologies 30
6.4.1 Benzene Levels Achievable through Reformate Benzene Control 31
6.4.2 Other Benzene Controls 33
6.5 Averaging, Banking, and Trading (ABT) Program 40
6.5.1 Starting Gasoline Benzene Levels 40
6.5.2 Model-Predicted Refinery Compliance Strategies 41
6.5.3 Predicted Reductions in Gasoline Benzene 43
6.5.3.1 Early Operational Changes in Benzene Control Technology 45
6.5.3.2 Early Small Capital Investments in Benzene Control Technology 45
6.5.3.3 Compliance with the 1.3 vol% Maximum Average Standard 46
6.5.3.4 Small Refiner Compliance with the Benzene Standards 47
6.5.3.5 Full Program Implementation / Ultimate Compliance with the 0.62 vol%
Standard 47
6.5.4 Credit Generation/Use Calculations & Considerations 48
6.5.4.1 What factors impact refiners' decisions to make early process changes? 48
6.5.4.2 How are early credits calculated? 49
6.5.4.3 How many early credits do we predict will be generated? 50
6.5.4.4 How many early credits will be demanded? 50
6.5.4.5 How are standard credits calculated? 52
6.5.4.6 How much additional lead time would be generated by standard credits
generated during the early credit "lag"? 52
6.5.4.7 How do we estimate ongoing standard credit generation/demand? 53
6.5.4.8 What are the credit use provisions? 54
6.5.4.9 Are there any geographic restrictions on credit trading? 54
6.5.4.10 What are the credit life provisions? 55
6.5.4.11 Consideration of credit availability 55
6.5.4.12 What is the economic value of the ABT program? 60
6.6 Feasibility for Recovering Octane 62
6.7 Will the Benzene Standard Result in Any New Challenges to the Fuel Distribution
System or End-Users? 65
6.8 Impacts on the Engineering and Construction Industry 66
6.8.1 Design and Construction Resources Related to Benzene Reduction Equipment67
6-1
-------
Final Regulatory Impact Analysis
6.8.2 Number and Timing of Benzene Reduction Units 67
6.8.3 Timing of Projects Starting Up in the Same Year 68
6.8.4 Timing of Design and Construction Resources Within a Project 68
6.8.5 Projected Levels of Design and Construction Resources 70
6.9 Time Needed to Comply with a Benzene Standard 74
6.10 Will the Benzene Standards Be More Protective Than Current Programs? 76
6.10.1 Modeling Approach 77
6.10.1.1 Choice of Analysis Cases and Data Sources 78
6.10.1.2 Adjustment of Fuel Parameters for Future Years 80
6.10.1.3 Conversion of Production Properties to In-Use Properties 82
6.10.1.4 Running the MOBILE Model 86
6.10.2 Interpretation of Results 89
6.10.3 Conclusions 90
6.11 MS AT Fuel Effects Test Program 90
6.11.1 Overview of Test Program 90
6.11.2 Key Findings and Next Steps 94
6.12 Analysis of Future Need for RFG Surveys of Toxics and NOX Performance under
MSAT2 96
6.12.1 Total Toxics Reduction 96
6.12.2 NOX Reduction 97
Appendix 6A: Additional Background on Refining and Gasoline 99
6A. 1 Petroleum Refining 99
6A.2 Crude Oil 100
6A.2.1 Crude Desalting 101
6A.2.2 Atmospheric Crude Unit 102
6A.2.3 Preflash 102
6A.2.4 Crude Unit 103
6A.2.5 Atmospheric Tower Gasoil and Residuum; Vacuum Unit 104
6A.2.6 Naphtha Splitter 107
6A.2.7 Hydrotreating 107
6A.2.8 Fluid Catalytic Cracker 108
6A.2.9 Alkylation 110
6A.2.10 Thermal Processing 110
6A.3 Gasoline 110
6A.3.1 Gasoline as a Complex Mixture Ill
6A.3.2 Octane 114
6A.4 Kerosene and Diesel 117
6-2
-------
Final Regulatory Impact Analysis
Chapter 6: Feasibility of the Benzene Control Program
This chapter summarizes our assessment of the feasibility of complying with a benzene
control standard. It begins with an overview of refining followed by a summary of the benzene
levels of gasoline today and where that benzene comes from. The various technologies which
reduce benzene levels in gasoline are described along with an assessment of the levels of
benzene achievable by the application of these technologies and their potential to be applied by
refineries. This assessment of the benzene levels achieved by applying control technologies is
used to assess the feasibility of complying with the benzene control program. Next the lead time
to apply the various control technologies and to comply with the new standards is evaluated.
Finally, the energy and supply impacts of the program are assessed.
6.1 Overview of Refinery Flow
Figure 6.1-1 shows a process flow diagram for a typical complex refinery, capable of
making a wide product slate (shown on the right side of the figure) from crude oil (input on the
left). Following the figure is a brief description of key units and streams focusing more on the
gasoline producing units. It's important to note that not all refineries have all of these units,
which is a key factor in both the variation in their baseline benzene levels as well as their cost of
benzene control.
6-3
-------
Final Regulatory Impact Analysis
Figure 6.1-1. Process Flow Diagram for a Typical Complex Refinery
Natural
Gas
Crude
Oil
* Fuel Gas
LPG
Gasoline
Aromatics
Crude Tower
Kerosine
^ Jet Fuel
On-Highway
Diesel
Off-roadDiesel
* Fuel Oil
»• Resid
Vacuum Tower
Coke
Crude Tower
The purpose of the crude tower is to perform a distillation separation of crude oil into
different streams for additional processing in the refinery and for the production of specific
products. Crude oil is shipped to the refinery via pipeline, ship, barge, rail, or truck, whereupon
it is sampled, tested, and approved for processing. The crude oil is heated to between 650° F and
700° F and fed to crude distillation tower. Crude components vaporize and flow upward through
the tower. Draw trays are installed at specific locations up the tower from which desired side
cuts or fractions are withdrawn. The first side-cut above the flash zone is usually atmospheric
gasoil (AGO), then diesel and kerosene/jet fuel are the next side-cuts, in that order The lightest
components, referred to here as straight run naphtha, remain in the vapor phase until they exit the
tower overhead, following which they are condensed and cooled and sent to the naphtha
splitter.l
Naphtha Splitter
The purpose of the naphtha splitter is to perform a distillation separation of straight run
naphtha into light straight run naphtha and heavy straight run naphtha. The feed can be split
between the Cs's and Ce's in order to assure the Ce's and heavier were fed to the reformer.2
6-4
-------
Final Regulatory Impact Analysis
Isomerization Unit
The purpose for the isomerization unit is to convert the light naphtha from straight chain
hydrocarbons to branched chain hydrocarbons, increasing the octane of this stream. The
isomerate is sent to gasoline blending.3
Reformer
The purpose of the reformer unit is to convert C6 to C8 or C9 hydrocarbons into aromatic
and other higher octane compounds (benzene is one of the aromatic compounds produced),
typically necessary to produce gasoline with sufficient octane. Heavy straight run naphtha is
hydrotreated and fed to the reformer. As the reformer converts the feed hydrocarbons to
aromatics, hydrogen and light gases are produced as byproducts. The liquid product, known as
reformate, is sent directly to gasoline blending, or to aromatics extraction.4
Aromatics Extraction Unit
The purpose of aromatics extraction is to separate the aromatic compounds from the rest
of the hydrocarbons in reformate using chemical extraction with a solvent to concentrate the
individual aromatic compounds, (mainly xylene and benzene) for sale to the chemicals market.5
Vacuum Tower
The purpose of the vacuum distillation tower unit is to enable a refinery to produce more
gasoline and diesel fuel out of a barrel of crude oil. It separate the heavy vacuum gasoil
(HVGO), which is fed to the FCC unit, from the vacuum tower bottoms (VTB) which is sent to
the coker, or in other refineries is made into asphalt.
Fluidized Catalytic Cracker
The purpose of the fluidized catalytic cracker is to convert heavy hydrocarbons, which
have very low value, to higher value lighter hydrocarbons. AGO and HVGO are the usual feeds
to a fluid catalytic cracker (FCC). The full boiling range cracked product leaves the reactor and
is sent to a fractionator. The overhead includes propane, propylene, butane, butylene, fuel gas
and FCC naphtha, which contains some benzene. There are two heavy streams; light cycle oil
(LCO), which can be hydrotreated and blended into diesel fuel or hydrocracked into gasoline;
and heavy cycle oil, sometimes called slurry oil, which can be used for refinery fuel.6
Gas Plant
The purpose of the gas plant is to use a series of distillation towers to separate various
light hydrocarbons for further processing in the alkylation or polymerization units or for sale.
Alkylation Unit
The purpose of the alkylation unit is to chemically react light hydrocarbons together to
6-5
-------
Final Regulatory Impact Analysis
produce a high quality, heavy gasoline product. Alkylation uses sulfuric or hydrofluoric acid as
catalysts to react butylene or propylene together with isobutane. Following the main reaction
and product separation, the finished alkylate is sent to gasoline blending. Alkylate is low in RVP
and high in octane.7
Polymerization Unit
The purpose of the polymerization unit is to react light hydrocarbons together to form a
gasoline blendstock. A polymerization unit, often referred to as a "cat poly" is somewhat similar
to an alkylation unit, in that both use light olefins to produce gasoline blendstocks. The feed is
generally propylene and/or butylene from the gas plant. The product, called polygas is sent to
gasoline blending.
Coker Unit
The purpose of the coker unit is to process vacuum tower bottoms (VTB) to coke and to
crack a portion to various lighter hydrocarbons. The hydrocarbons produced by the coker
include cracked gases, coker naphtha, coker distillate and gas oil. The gas is fed to the gas plant,
the naphtha to the reformer hydrotreater, and the distillate either to distillate hydrotreating or to
the hydrocracker.
Hydrocracker
The purpose of the hydrocracker is to crack and "upgrade" the feedstock into higher
value products. The feedstock to the hydrocracker is usually light cycle oil (LCO) and coker
distillate, poor quality distillate blendstocks, which are upgraded to diesel fuel, or cracked to
gasoline. Heavier hydrocarbons such as AGO and HVGO can be feedstocks as well.
A more complete description for reforming is contained in Section 6.3. Other refinery
units are described in more detail in the Appendix.
6.2 What are the Benzene Levels in Gasoline Today?
EPA receives information on gasoline quality, including benzene, from each refinery in
the U.S. under the reporting requirements of the Reformulated Gasoline and Antidumping
Programs. Benzene levels averaged 0.97 volume percent for gasoline produced in and imported
into the U.S. in 2004, which is the most recent year for which complete data was available at the
time of this analysis. The benzene levels differ depending on different volumes of interest. We
assessed the 2004 benzene levels by conventional versus reformulated gasoline, winter versus
summer, and with and without California and Imports. Table 6.2-1 contains the benzene levels
for these various gasoline types by season and aggregated.
6-6
-------
Final Regulatory Impact Analysis
Table 6.2-1. Summary of U.S. Benzene Levels by Gasoline Type and Season for 2004
CG Summer
CG Winter
Total CG
% total volume
RFG Summer
RFC Winter
Total RFG
% total volume
Summer CG & RFG
Winter CG & RFG
Total CG & RFG
% of total volume
U.S. Production
(excl. California)
1.132
1.076
1.103
64.3
0.587
0.622
0.606
20.3
1.006
0.964
0.984
84.6
Imports
0.949
0.756
0.828
1.9
0.677
0.696
0.688
2.1
0.800
0.725
0.754
4.0
Production +
Imports
1.128
1.065
1.095
66.2
0.594
0.629
0.613
22.4
0.998
0.952
0.973
88.6
California
0
0.620
0.620
0.620
11.4
0.620
0.620
0.620
11.4
All Gasoline
1.128
1.065
1.095
66.2
0.603
0.626
0.616
33.8
0.955
0.914
0.933
100.0
Individual refinery gasoline benzene levels can vary significantly from the national
average with trends forming in specific regions of the country. Therefore, it is useful to
understand how the benzene levels vary by individual refinery as well as regionally. Figure 6.2-
1 contains a summary of annual average gasoline benzene levels by individual refinery for
conventional gasoline and reformulated gasoline versus the cumulative volume of gasoline
produced (not including California refineries for which EPA does not receive data).
6-7
-------
Final Regulatory Impact Analysis
Figure 6.2-1. Benzene Content of RFG and Conventional Gasoline, 2004.
4.5 -i
4n
o
§on
N o.U
O
OQ
c Zb
0
o
o o n
0. -i.U
3 1 ^
O
i n
1 .U
0.5
y
^^
___x^
.—- --' — "" — "~~"^
pp.
RFG
. u i i i i i i i i
0 10 20 30 40 50 60 70 80 90 100
Cumulative Gasoline Supplied (Billion Gallons)
Figure 6.2-1 shows that the annual average benzene levels of conventional gasoline
produced by individual refineries varies from 0.3 to 4.2 volume percent. The volume-weighted
average is 1.10 volume percent. As expected, the annual average benzene levels of reformulated
gasoline as produced by individual refineries are lower ranging from 0.1 to 1.0 volume percent.
The volume-weighted average benzene content for U.S. reformulated gasoline (not including
California) is 0.61 volume percent.
The information presented for annual average gasoline benzene levels does not indicate
the variability in gasoline batches produced by each refinery. We also evaluated the batch-by-
batch gasoline benzene levels for individual refineries. This information is obtainable from data
provided to EPA under the reporting requirements of the RFG program. To illustrate the degree
of variability within different refineries, in Figure 6.1-2 through 6.2-7 we provide the data for 3
different refineries which produce both conventional and reformulated gasoline and 3 refineries
which produce solely conventional gasoline. For the RFG producing refineries we summarize
the data by gasoline type as these refineries produce both RFG and CG. For the CG refineries
we break out the data by premium grade, regular grade and midgrade gasoline, if the refinery
produces it. We arbitrarily labeled the refineries in these figures refineries A through F to
facilitate the discussion about this data.
-------
Final Regulatory Impact Analysis
Figure 6.2-2. RFG and CG Batch-by-Batch Benzene Levels for Refinery "A'
(volume percent benzene in 2003 gasoline)
2.5
2.0
2 1.5
1.C
o.e
o.o
Jan-03
** x. x
CG X RFG
Apr-03
Jun-03
Batch Date
Sep-03
Dec-03
6-9
-------
Final Regulatory Impact Analysis
Figure 6.2-3. RFG and CG Batch-by-Batch Benzene Levels for Refinery "Br
(volume percent benzene in 2003 gasoline)
3.5
3.0
2.5
0)
2.0
0)
.a
S 1.5
CG X RFG
Jan-03
Apr-03
Jun-03
Batch Date
Sep-03
Dec-03
6-10
-------
Final Regulatory Impact Analysis
Figure 6.2-4. Batch-by-Batch Benzene Levels for Refinery "C" that Produces both RFG
and CG Gasoline (volume percent benzene in 2003 gasoline)
o
2.5
2.0
1.5
1.0
0.5
0.0
Jan-03
• CG
XRFG
•
X
Apr-03
Jun-03
Batch Date
Sep-03
Dec-03
6-11
-------
Final Regulatory Impact Analysis
Figure 6.2-5. Premium and Regular Grade Gasoline Batch-by-Batch Benzene Levels for
Refinery "D" (volume percent benzene in 2003 gasoline)
3.0
2.5
2.0
nzene
J| 1.5
1.0
.
*
• *
0.0
Jan-03
x Premium Grade • Regular Grade
Apr-03
Jun-03
Batch Date
Sep-03
Dec-03
6-12
-------
Final Regulatory Impact Analysis
Figure 6.2-6. Premium, Midgrade and Regular Grade Batch-by-Batch Benzene Levels for
Refinery "E" (volume percent benzene in 2003 gasoline)
c
A
enzene
0 4
& J 1
ss
0
o
1
• Regular Grade A Mid-Grade x Premium Grade
A
A A A A
A A A
A*
A A * AA A^ 4«
A ^ A • A A
' V '. . /* .A A^AA A ^ A A .
' ' ' -. ' A-"- -"Jl -* A." «AV * '
• -. - - ' - .- r^. -.;>/••• .- • - ..
x x • ,
X v X * . X x XX
^x^xxxx xxx xxXxx x x ^x^^x^x * 3< ^ x x* x^
x x
Jan-03 Apr-03 Jun-03 Sep-03 Dec-03
Batch Date
6-13
-------
Final Regulatory Impact Analysis
Figure 6.2-7. Premium and Regular Grade Gasoline Batch-by-Batch Benzene Levels for
Refinery "F" (volume percent benzene in 2003 gasoline)
1
x Premium Grade • Regular Grade
n-03
Apr-03
Jun-03
Batch Date
Sep-03
Dec-03
Most of the refineries that we studied produced substantially different batch-to-batch
benzene levels. As expected, the RFG batches were consistently lower than the CG batches.
Two of the RFG producing refineries had a wide variability in benzene levels. The gasoline
batch benzene levels for refineries A and B varied by over an order of magnitude. Refinery C's
gasoline batch benzene levels varied less than those of refinery A and B. Most all of refinery
C's batches were under 0.5 volume percent benzene except for a very few which were much
higher and were sold as CG. Also, refinery C's gasoline batches had similar benzene levels for
both RFG and CG, a very different trend than refineries A and B.
Of the three CG refineries, refineries labeled E and F have widely varying gasoline batch
benzene levels. Refinery E's gasoline batch benzene levels were consistently higher than the
rest, ranging from under 1 percent to over 4 percent. Refinery F had no clear trend for either the
regular or premium grade of gasoline; the benzene levels varied for both by about an order of
magnitude. Refinery E did have an interesting trend for specific refinery grades. Premium grade
tended to have lower benzene levels than the other grades, midgrade had the highest benzene
levels and regular grade's benzene levels were in between the other two grades. Evaluated all
together, the various grades of refinery E also varied by an order of magnitude. The gasoline
6-14
-------
Final Regulatory Impact Analysis
batch benzene levels for refinery D were consistently under 0.5 volume percent for most of the
batches, although a very small fraction of the batches had much higher benzene levels. The
lower variability in refinery D's batches was similar for both premium and regular grades of
gasoline.
There are several reasons for the variability in refinery gasoline benzene levels across all
the refineries. First, crude oil varies greatly in aromatics content. Since benzene is an aromatic
compound, its concentration tends to vary consistent with the aromatics content of crude oil. For
example Alaskan North Slope (ANS) crude oil contains a high percentage of aromatics. A
refiner processing ANS crude oil in their refineries shared with us that their straight run naphtha
off the atmospheric crude distillation column contains on the order of 3 volume percent benzene.
This is one reason why the gasoline in PADD 5 outside of California is high in benzene.
Conversely, refiners with very paraffmic crude oils (low in aromatics) may have benzene levels
as low as 0.3 volume percent benzene in their straight run naphtha.
The second reason why benzene levels vary is due to the types of units in their refinery.
Different refinery streams contain widely different concentrations of benzene, with reformate
typically contributing the most. If a refinery relies on the reformer for virtually all of their
octane needs, especially the type which operates at higher pressures and temperatures that tends
to produce more benzene, they will likely have a high benzene level in their gasoline. Refineries
with a reformer and without an FCC unit are particularly prone to higher benzene levels.
However, refineries which can rely on several different units or means for boosting their
gasoline octane can usually run their reformers at a lower severity resulting in less benzene in
their gasoline pool. Examples of octane-boosting refinery units include the alkylation unit, the
isomerization unit, and units which produce oxygenates. Refiners may have these units in their
refineries, or in many cases, the gasoline blendstocks produced by these units can be purchased
from other refineries or third-party producers. The blending of alkylate, isomerate, and
oxygenates into the gasoline pool provides a significant octane contribution which would allow
refiners to rely less on the octane from reformate. The variation in gasoline blendstock content
across different batches of gasoline is likely the reason for the drastically differing benzene
levels between batches of gasoline.
Finally, many refiners may be operating their refinery today to intentionally have less
benzene in their gasoline. They could be doing this by operating the refinery with that end in
mind such as for the Federal or California RFG programs. Refiners which are currently
producing reformulated gasoline are targeting to reduce their gasoline benzene levels to less than
0.95 volume percent for the Federal RFG program or lower for the California RFG program, and
are using benzene control technologies to produce gasoline with lower benzene levels. If they
are producing conventional gasoline along with the reformulated gasoline, their conventional
gasoline is usually lower in benzene as well compared with the conventional gasoline produced
by other refineries. Alternatively, some refiners add specific refinery units such as benzene
extraction which intentionally removes benzene and concentrates it for the profit it earns. The
profit gained by extraction is due to the much higher price that benzene earns on the benzene
chemical market compared to the price of gasoline. In most cases, refineries with extraction
units are also marketing their low benzene gasoline as RFG.
6-15
-------
Final Regulatory Impact Analysis
Table 6.2-2 shows the variations in gasoline benzene levels as produced by refineries in,
and as imported into, refining regions called Petroleum Administrative for Defense Districts
(PADD) for 2004.8 The information is presented for both conventional gasoline and
reformulated gasoline.
Table 6.2-2. 2004 Benzene Levels by Gasoline Type and by PADD as Supplied in the U.S
Conventional
Gasoline
Reformulated
Gasoline
Gasoline
Average
PADD1
0.84
0.63
0.72
PADD 2
1.33
0.81
1.24
PADD 3
0.94
0.54
0.87
PADD 4
1.55
N/A
1.55
PADD 5
1.75
N/A
1.75
CA
0.62
0.61
0.62
U.S.
1.10
0.63
0.94
Table 6.2-2 shows that benzene levels vary fairly widely across different regions of the
country. PADD 1 and 3 benzene levels are lower because the refineries in these regions produce
a high percentage of reformulated gasoline for both the Northeast and Gulf Coast. About 60
percent of PADD 1's gasoline is reformulated, while 20 percent of PADD 3's gasoline is
reformulated. Reformulated gasoline must meet a 0.95 volume percent average benzene
standard, and a 1.3 volume percent cap standard. Another reason why the benzene levels are so
low in these two regions is because 35 percent of the refineries in these two regions, are
extracting benzene for sale to the petrochemicals market. When refiners are extracting benzene
from their gasoline, they extract as much benzene as possible to take maximum advantage of the
expensive cost of capital associated with extraction units. This is likely the reason why the CG
in PADDs 1 and 3 is low in benzene as well. In other parts of the U.S., where little to no
reformulated gasoline is being produced and little extraction exists, the benzene levels are much
higher.
6.3 Where Does Gasoline Benzene Come From?
The portion of the crude oil barrel which boils within the gasoline boiling range is called
naphtha. There are two principal sources of naphtha. The first principal source of naphtha is
straight run naphtha which comes directly off of the crude oil atmospheric tower. The second
principal source of naphtha is from the cracking reactions. Each type of naphtha provides a
source of benzene to gasoline.
Straight run naphtha which comes directly from the distillation of crude oil contains
anywhere from 0.3 to 3 volume percent benzene. While straight run naphtha is in the correct
distillation range to be usable as gasoline, its octane value is typically 70 octane numbers which
is too low for blending directly into gasoline. Thus, the octane value of this material must be
increased to enable it to be sold as gasoline. The primary means for increasing the octane of
6-16
-------
Final Regulatory Impact Analysis
naphtha is reforming. In the process of increasing the octane of this straight run material, the
reformer increases the benzene content of this stream.
There are two primary cracking processes in the refinery. One is called the fluidized
catalytic cracking (FCC) unit and the second is called hydrocracking. Other cracking units
include cokers and thermal crackers. These various cracked naphthas contain anywhere from 0.5
to 5 volume percent benzene.
The attached table summarizes the range in benzene content and typical percentage of
gasoline of the various refinery intermediate streams used to blend up gasoline.
Table 6.3-1. Benzene Content and Typical Gasoline Fraction of Various Gasoline
Blendstocks.
Process or Blendstock
Name
Reformate
FCC Naphtha
Alkylate
Isomerate
Hydrocrackate
Butane
Light Straight Run
MTBE/Ethanol
Natural Gasoline
Coker Naphtha
Benzene Level
(volume %)
3-11
0.5-2
0
0
1-5
0
0.3-3
0.05
0.3-3
3
Typical Volume in
Gasoline (volume %)
30
36
12
4
3
4
4
3
3
1
Estimated Contribution to Gasoline
Benzene Content (volume %)
77
15
-
-
4
-
2
-
1
1
Table 6.3-1 shows that the principal contributor of benzene to gasoline is reformate. This
is due both to the high benzene content of reformate and the relatively large gasoline fraction
that it comprises of the gasoline pool. For this reason, reducing the benzene in reformate is the
focus for the various benzene reduction technologies available to refiners.
6.3.1 How Do Reformers work?
Reformers have been the dominant gasoline high octane producing units since they first
came into operation in the 1940's.9 An indication of their importance in refining is that every
U.S. refinery except one has a reformer. Prior to the lead phase-down in the early 1980's
reformers operated at fairly moderate severities and produced product octane numbers around 85
RON (see the Appendix for a discussion of octane). After the phase-down and eventual phase-
out of lead from gasoline, and as the demand for high-octane premium fuel grew, octane
numbers for reformate increased to a range from a RON in the low 90s to 104. The reforming
process works by rearranging, e.g., "reforming" the chemical structure of straight-chain and
cycloparaffin molecules in a given feedstock, to produce a variety of high-octane benzene,
substituted aromatic, and isoparaffmic molecules. The reforming process uses a combination of
heat, pressure, and catalyst, to produce high octane, high-value finished blendstocks from a low-
octane, (about 50 RON in some cases) low-value feedstock.
6-17
-------
Final Regulatory Impact Analysis
Reformer Chemical Reactions
The chief means by which reformers increase octane is through the formation of aromatic
compounds, including benzene. Aromatic compounds are distinguished from other hydrocarbon
compounds by their structure which cannot be described without at least a very rudimentary
discussion of organic chemistry. All hydrocarbons can be categorized into two groups, saturated
and unsaturated. Saturated compounds have single bonds between carbons with the other bonds
to carbon being made with hydrogen. Unsaturated hydrocarbons contain a double bond between
one or more carbon atoms thus, there are fewer hydrogen atoms attached to the carbons.
Aromatic compounds are unsaturated ring hydrocarbons with six carbons forming the ring.
Benzene is the most basic of the aromatic compounds having a structure of CeHe. Other
aromatic compounds are variants of the benzene ring. Toluene has a methyl group replacing one
hydrogen molecule attached to the six carbon ring of benzene. Xylenes have two methyl groups
replacing two of the hydrogens of the benzene ring.
Five reactions take place in a reformer: 1) The dehydrogenation (hydrogen removal) of
naphthenes; 2) The dehydroisomerization (hydrogen removal and conversation of hydrocarbons
from straight chain to branched chain) of alkyl cyclopentanes; 3) The isomerization (conversion
of hydrocarbons from straight chain to branched chain) of paraffins and aromatics; 4) The
dehydrocyclization (hydrogen removal and conversion of hydrocarbons from straight chain to
cyclic) of paraffins; and 5) The hydrocracking (conversion of hydrocarbons to smaller molecules
with hydrogen as a reactant) of paraffins and naphthenes. Reactions numbered 1, 2 and 4 form
aromatic compounds, while reaction number 3 can alter aromatic types. There are two very
important reactions which result in the formation of benzene. Reaction number 1 forms benzene
from cyclohexane. Reaction number 2 forms benzene from methyl cyclopentane. Reactions
numbered 1, 2, & 4 produce hydrogen as a by-product. Reaction number 3 neither produces nor
consumes hydrogen. Reaction number 5 consumes hydrogen.10,11
Reformer Feed and Operations
The feed to the reformer comes from the splitter bottom as we described previously; in
some cases, the feed may come directly from the crude tower. Until recently, the reformer feed
boiling point range was about 180° F to 370° F. The 180° F initial boiling point temperature sets
the cut between the hexanes and pentanes in the crude tower overhead. If the initial boiling point
of the feed is lower than 180° F, pentanes that are normally not considered good feed will be
pulled into the reformer. The 180° F temperature has varied somewhat according to the crude
from which the feed comes and also according to a particular refiner's economics.
Feed boiling point (FBP) adjustments often have to do with economics. The maximum
FBP for reformer feed is about 390° F to 400° F. The catalyst will coke (accumulate carbon) at
370° F, but as the feed FBP's rise above 370° F the coking rate rises increasingly more rapidly,
until at the 390° F to 400° F range, the catalyst cycle length is far to short to even be considered.
On the other hand, the reformer feed portion that boils above about 340° F could be cut into
kerosene, jet fuel, or diesel. In other words, the price-spread between gasoline and diesel may
warrant cutting some of the heavy straight run into diesel. Under other economics, it may pay to
6-18
-------
Final Regulatory Impact Analysis
run the reformer feed FBP up as high as possible in order to maximize gasoline make. During
summer months the demand for gasoline grows while the demand for diesel fuel drops. To stay
in balance, a refiner may raise the FBP of the HSR to as high as 390° F. This move would
increase the reformer feed volume and at the same time reduce the kerosene and ultimately the
diesel make. If the refiner has a jet fuel contract, he may not be able to make such a change.
Increasing the initial boiling point can reduce the benzene make in the reformer. This is covered
in the next section discussing the technologies for reducing gasoline benzene levels.
Different crude oil types affect the quality and volume of feed to the reformer. Light,
sweet crude, such as that produced in southwestern Wyoming, is reported to have had as much as
35% to 45% by volume of heavy straight run (HSR) naphtha that is high naphthenes and
aromatics and consequently a fairly rich feed. By contrast, there are heavy asphaltic crudes
produced from off the California coast with almost no HSR.A Reformer feed often includes
intermediate streams from hydrocrackers and cokers. Coker naphtha ordinarily must be
hydrotreated at conditions well beyond the severity of the common reformer hydrotreater before
it is fed to a reformer. HSR from a hydrocracker is usually very clean with regard to most
critical contaminants, but as a rule must be reformed because it has a very low octane.
Occasionally a refiner must consider reforming a poorer feed (e.g., feed from paraffinic crude).
In such cases, the refiner may need to load two or three different catalysts into his reactors in
stacked-beds in order to provide for all the necessary reactions. Paraffinic feedstocks are
ordinarily difficult to reform.
A reformer consists of essentially three separate components: the naphtha hydrotreater
section, the reformer section, and the product stabilization section. The reformer section
contains a catalyst which is usually bi-metallic; platinum and rhenium are two that are often
used. Consequently, the catalyst is quite expensive.
The feed to the reformer is hydrotreated to reduce contaminants, such as sulfur, nitrogen,
and arsenic. Arsenic poisons the catalyst, from which the catalyst activity cannot be recovered;
sulfur and nitrogen deactivate the catalyst and to some degree activity can be regained through
regeneration. The process conditions of the hydrotreater are ordinarily not severe; using
common hydrotreating catalysts, temperatures around 600° F and pressures of around 400 psi.
The hydrotreater reactor effluent is fed to a stabilizer/splitter to remove light products
and gaseous contaminants, such as hydrogen sulfide formed in the hydrotreating process. The
stabilizer bottoms are heated against reformer reactor effluent in feed/effluent exchangers, and
subsequently fed to the first pass of the reformer feed furnace. There are typically four reactors
IA & IB, II, and III, in series. The feed is heated to a feed temperature of about 930° F in the
first pass and fed down-flow to reactors IA & IB, where several endothermic reactions take
place; the reactor effluent is then fed to the second furnace pass and reheated to the same reactor
inlet temperature as for the first set of reactors. It is subsequently fed to reactor II. The effluent
is heated once again, and fed to the third furnace to be reheated and fed to the third reactor.
Effluent from the third reactor is cooled against first-pass furnace feed in the
Internal document.
6-19
-------
Final Regulatory Impact Analysis
feed/effluent exchangers and fed to the high pressure separator. One of the principal byproducts
of the reforming reactions is hydrogen. Volumes in excess of 1000 scf per barrel of feed have
been reported. The high pressure separator is used to separate the hydrogen from the cooled
reactor effluent liquid. Part of the hydrogen is recycled back to the reformer; mole ratios of five
moles of hydrogen to one mole of feed are usually required to suppress catalyst coking. Some of
the excess hydrogen is fed to the naphtha hydrotreater and the balance is available for other units
in the refinery that may need it; e.g., cat feed hydrotreaters or distillate hydrotreaters are
examples. The liquid reactor effluent is reheated and fed to a stabilizer to control the Reid Vapor
Pressure (RVP) of the final reformate. The stabilizer is ordinarily a total-reflux unit, the pressure
of which is controlled by a gas controller on the tower overhead drum. Light hydrocarbons in
the off-gas, released to maintain pressure control, are sent to either the gas plant or to fuel gas.
The light hydrocarbons in the off-gas includes methane, ethane, propane and butanes in small
volumes.
Different reformer operating conditions result in the production of different qualities of
reformate, different hydrogen production levels and can change the reformer cycle length (time
between catalyst replacements or regeneration). For example, low reactor pressure increases
yield and octane but increases the production of coke. Increased hydrogen partial pressure, that
is the ratio of hydrogen to hydrocarbon, suppresses coke formation, it promotes hydrogen yield
and product octane, but it also promotes hydrocracking. Reducing the space-velocity, that is the
rate at which the reactor volume of the hydrocarbon changes per unit time, favors aromatic
production, but also promotes cracking. Higher activity catalysts increase cycle lengths and
usually yields, but sometimes they are more expensive.12
Certain tools are available to refiners to tailor the reforming process to their needs. There
are several proprietary processes, including catalysts, from which refiners can choose to treat the
specific qualities of their heavy naphtha. In most cases, a few laboratory tests allow vendors to
estimate, with reasonable accuracy, how well their processes can reform a given feedstock.
However, in some cases, vendors insist on running pilot plant tests before they will guarantee
their process's performance. A common lab test, known as a PONA, is used to determine
paraffin, olefm, aromatic, and naphthene content; API gravity, sulfur, nitrogen, and metals are
also important. From these test results, most vendors have computer-based process simulators
that, for a given RON, can estimate the finished product and hydrogen yield, off-gas composition
at several different Reid Vapor Pressures (RVP), reformate octanes, and catalyst cycle lengths, if
a unit already exists with suitable reactors and compressors in place. In nearly all cases, vendors
supply the above test results for a range of RON's. For example, the lowest RON a refiner may
decide to produce might be 85 RON. A vendor could provide process design services to
determine the cycle length requested by that refiner for a set of specified equipment design
criteria. This, of course, is based on, among other criteria, the type of reformer.
Types of Reformers
There are two types of reformers in use today, the semi-regenerative reformer, and the
continuous reformer. The predominant operating differences between the two are the pressure
and the means for regenerating the catalyst.
6-20
-------
Final Regulatory Impact Analysis
The semi-regenerative reformer gets its name from the need to periodically shut down the
unit to regenerate and reactivate the catalyst. The catalyst, usually carrying a specific weight
percent platinum and rhenium on a common base material, is loaded in a series of down-flow
reactors. The process pressure is higher in this type of reformer, at around 200 psi to 350 psi.
Reactor inlet temperatures begin at around 930° F. This start-of-run inlet temperature may vary
from process to process, as will the final end-of-run temperature. A delta temperature from start
to end of about 40° F is common. Over time, as a result of some of the reforming/hydrocracking
reactions, coke builds up on the surface and the catalyst deactivates. As coke is gradually
deposited on the catalyst, the reforming reactions slow down somewhat and the reformate or
product octane begins to drop a little below the desired set point. To compensate, the feed
temperature is raised until the desired octane is reached again. These steps are repeated
periodically over the cycle length of the particular catalyst. Contaminants such as sulfur can
speed up the deactivation, as can other problems. When the maximum allowable feed
temperature is reached, the refiner must shut the unit down and regenerate the catalyst.
Regeneration may take place "in situ" or the catalyst may be removed from the unit and
sent to a regeneration contractor for regeneration. Briefly, regeneration involves carefully
burning the coke off of the catalyst surface, and then chemically treating the clean catalyst to
reactivate it. Regeneration is a fairly delicate operation, in that, for example, if too much oxygen
is allowed into the process, the temperature may get high enough to damage the catalyst and
prevent it from being reused. Regeneration, whether in situ or away from the refinery, is
generally done the same way. The one significant difference is that the catalyst is not reduced
with hydrogen directly following the burn phase at the off site plant. If carried out in situ, the
process can go forward without interruption. Some refiners insist on burning in situ.
Regardless, the catalyst still must periodically be dumped, screened to remove fines, and
reloaded. The burn phase also usually takes place before the unit is shutdown for other
maintenance. Startup following a regeneration period also requires patience and may take
several days before a specified product octane can be reached. An important step is to dry out
the catalyst before attempting to raise the reactor inlet temperatures to achieve the desired
octane. As the catalyst "life" shortens, the start-of-run temperature will gradually increase, so
that the usual delta T will gradually become narrower and eventually the catalyst cycle length
becomes too short to be economical.
This regeneration process can be burdensome on refiners. For this reason, refiners
choose to operate this unit at a higher operating pressure to reduce the frequency of regeneration
cycles. The higher operating pressure reduces the formation of coke on the catalyst which
extends the cycles between regeneration. Higher pressure also reduces hydrogen make and
increases the cracking of heavier aromatics to benzene.
The second type of reformer uses continuous catalyst regeneration, wherein the catalyst
is continuously withdrawn from the process, the coke burned off, the catalyst is reduced, and fed
back into the process without shutting the unit down for long operating periods. In some ways,
the process is similar to the FCC. The reactors are stacked rather than lined up separately in
series so that the catalyst can flow under gravity. From the bottom of the reactor stack, the
'spent' catalyst is 'lifted' by nitrogen to the top of the regenerator stack. In the regenerator, the
above mentioned "regeneration" steps of coke burning, chlorination and drying are done in
6-21
-------
Final Regulatory Impact Analysis
different sections, separated by a system of valves, screens, and other equipment. From the
bottom of the regenerator stack, catalyst is lifted by hydrogen to the top of the reactor stack, in a
special area called the reduction zone, where once heated is brought into contact with hydrogen,
which reduces (changes the valence) the catalyst surface to restore its activity. A continuous
regeneration process can be maintained without unit shutdown for run lengths of about 4 to 5
years.
The continuous reformer's regeneration process is much more streamlined than the semi-
regenerative reformer. For this reason, the continuous reformers are operated at a considerably
lower pressure, from as low as 90 to 120 psi, than the semi-regen process and the hydrogen make
is considerably higher. For the same reason, the severity of continuous reformers can be higher
and product octane in the range of 104 RON is not uncommon. The lower pressure of the
continuous reformer also causes less benzene make from the cracking of heavy aromatic
compounds.
6.3.2 How Can Benzene Levels be Reduced in Gasoline?
There are several ways available to refiners to reduce the benzene in their finished
gasoline.8 One way is to pre-fractionate the feed, and prevent the benzene precursors from
entering the reformer. The other way is to post-fractionate reformate into light and heavy cuts,
and either saturate the benzene in the light cut or extract it for sale in the chemical feed market.
6.3.2.1 Pre-Fractionation to Reroute Benzene Precursors
The heavy straight run naphtha can be cut differently to reduce gasoline benzene levels.
As discussed earlier, the heavy straight run naphtha is cut to prevent the C5s from being sent to
the reformer. This means that most of the C6s are sent to the reformer along with the C7s, C8s
and sometimes the C9s. The cut-point could be changed from between the Cs's and Ce's to
between the Ce's and Cy's; in so doing the benzene precursors are also cut out of the reformer.
To assure that most of the C6's are cut out of the reformer feed, the initial boiling point of the
feed would need to be raised from 180° F to around 215° F to 220° F by changing the draw
temperatures on the units. The cut adjustments can be made in the pre-flash column (a simple
unit before the crude tower which removes the lightest compounds before entering the crude
tower), the crude tower overhead, or the naphtha splitter. These various distillation columns are
usually designed to make a fairly imprecise cut between the C6s and C7s, which would also cut
some Cy's out of the reformer feed. Cutting some of the C7s out of the heavy straight run going
to the reformer would, of course, reduce the production of C7 aromatics (toluene), and further
reduce the make of hydrogen. This would be costly to the refiner, so the refiner pursuing this
strategy would be expected to increase the ability to make a sharper cut between the C6s and
B The benzene reduction technologies are discussed here in the context of the feasibility for
reducing the benzene levels of gasoline to meet a gasoline benzene content standard. However,
this section could also substitute for a feasibility discussion of complying with a total air toxics
standard since benzene control would be the means refiners would choose for complying with
such a standard.
6-22
-------
Final Regulatory Impact Analysis
C7s. They would accomplish this by adding a naphtha splitter column, or adding height or
adding trays to their existing naphtha splitter. In many cases, the refinery would replace the
existing naphtha splitter with a new taller tower. The naphtha splitter in some refineries would
already be outfitted to make such a cut.
Refiners have recently routed a gasoline sub stream differently that will affect the content
of their heavy straight run naphtha and ability to reduce their benzene levels. Many U.S.
refiners, especially in PADDs 4 and 5, and to a lesser extent in PADDs 2 and 3, blend some light
gasoline-like material, which is a by-product of natural gas wells, into their gasoline.
Previously, natural gasoline was almost exclusively blended directly into the gasoline pool by
each refinery in each PADD where natural gasoline is a feedstock for refineries. The benzene
concentration in this stream is estimated to be 1.3 volume percent which, because it is not high,
would be costly to treat by itself for reducing its benzene content. However, we believe that
refiners will already be routing natural gasoline differently in their refinery for other reasons. To
comply with the 30 ppm Tier 2 sulfur standard, refiners may be treating this stream in a way to
reduce its sulfur. Because natural gasoline is fairly low in octane, most refiners will be blending
it into crude oil where it would be distilled so that the heavy portion of it will go to the straight
run hydrotreater and then sent to the reformer. This will lower the sulfur in the heavy portion of
the natural gasoline and improve its octane. Also, as the naphtha streams are routed to reduce
benzene levels, the natural gasoline benzene will be treated along with the rest of naturally
occurring benzene.
A few other concerns would need to be addressed as a result of removing the benzene
precursors. Benzene has a fairly high octane blending value; well in excess of 100 RON.
Simple arithmetic demonstrates that for each one-percent benzene removed, the reformate octane
is reduced by at least one number. Most refiners can't tolerate this, particularly if other high
octane blendstocks are not readily available. An obvious means to recover the lost octane would
be to increase reformer severity; while this seems reasonable, there are generally additional
consequences. Increased severity will likely convert more of the Cy's, Cg's, and Cg's into
compounds that could finally end up as benzene. For example, methyheptane can also be
converted into benzene, through paraffin dehydrocyclizaion (the methylated paraffin is
converted into a cycloparaffm and dehydrogenated) and demethylization (the methyl group is
removed) the possibility of which is more likely in semi-regen reformers. Similar reactions can
be predicted for other Cg and Cg alkanes, all of which reduces the net effect of the original
reduction. Even so, the benzene content will be lower than prior to pre-fractionation.
Addressing the octane loss due to benzene precursor rerouting can be addressed through other
means described below in Section 6.6. Other potential problems are that hydrogen production
will be reduced and that the increased severity naturally shortens the catalyst cycle length; this is
particularly important for semi-regeneration units, but also affects the continuous regeneration
units.
Cutting the benzene precursors out of the reformer feed would definitely reduce the
benzene content in gasoline, but it would not completely eliminate it. As discussed above, some
of the benzene in reformate is formed by the cracking of heavy aromatics, thus some benzene
would remain in reformate. Also the naturally occurring benzene present with the benzene
precursors would still be present in the rerouted C6 stream.
6-23
-------
Final Regulatory Impact Analysis
6.3.2.2 Benzene Saturation via Isomerization
The rerouted benzene precursor stream contains the naturally occurring benzene from
crude oil. An existing isomerization unit could be used to saturate this naturally occurring
benzene in the rerouted C6 stream. The role of the isomerization unit is to convert straight chain
compounds to branched chain compounds using a catalyst and in the presence of hydrogen,
which increases the octane of the treated stream. The isomerization reactor saturates benzene
using the hydrogen present in the reactor for the isomerization reactions. However, isomerate
has a fairly high RVP (in the range of 13 psi to 15 psi) which could make it difficult for the
refiner to add more isomerization capacity in that refinery while still meeting the RVP
requirement that applies to its gasoline. As such, a safe assumption could be made that the
refinery would be capable to use the existing isomerization unit up to the listed capacity of the
unit. The refiner presumably sized the isomerization unit to be able to use that capacity in the
first place. Treating the benzene in the rerouted benzene precursor stream could be
accomplished by giving a higher priority to treating the rerouted C6 stream in the isomerization
unit. If the isomerization unit's capacity is reached before it can treat all the C5 and C6s, then
the original C5 stream could be backed out until all the C6s are treated. Even so, adding an
isomerization unit may be possible, which also may require the refiner to add some RVP
reduction capacity elsewhere in the refinery to compensate for increased isomerate.
A potential drawback to isomerization is that as benzene is saturated, it produces heat
(exothermic reaction). Isomerization reactions are all equilibrium reactions. As such, as the
temperature in the reactor increases, it changes the equilibrium and shifts the isomerization
reactions back, which could lower the product octane. The licenser of the Penex isomerization
process has provided a recommendation that the isomerization unit be limited to 6 volume
percent benzene in the feed for this reason. The refinery could still treat this C6 stream using
this means, it would, however, need an additional reactor installed before the isomerization
reactor solely designed for saturating the benzene in this stream. The combined benzene
saturation reactor with the isomerization reactor is called a Penex Plus unit.
Another potential drawback to the benzene saturation option is that it requires at least
three moles of hydrogen (as H^) per mole of benzene saturated; this of course would require
additional hydrogen production. Providing additional hydrogen would add additional operating
cost to supply this hydrogen and could require capital investment.
The naphtha splitter overhead (typically light straight run gasoline, LSR, most of which
is C5's with some C6's) is routinely fed to an isomerization unit (otherwise it is blended directly
into gasoline). Most refiners run the feed through a deisopentanizer to remove isopentane, since
it won't need to be treated (it is already a branched chain compound and would only use up
existing capacity). The deisopentanizer bottoms are mixed with hydrogen, which helps
minimize coke formation on the catalyst; hydrogen is neither generated nor consumed in the
isomerization reactions.
The reactor effluent, known as unstabilized isomerate, is fed to a stabilizer where the
vapor pressure is controlled. Any light gas produced by minor cracking reactions is typically
6-24
-------
Final Regulatory Impact Analysis
scrubbed and blended into the refinery fuel gas system. Isomerate, at this point, would probably
have a clear octane number 10 points higher than the LSR feed; perhaps 80 to 82 RON.
The overall severity of isomerization process conditions is relative low; the temperature,
and the total and hydrogen partial pressures are all relatively low, compared with, say, reforming
or some other refinery processes. Isomerization is a vapor-phase process which uses hydrogen
to suppress dehydrogenation and coking. The catalyst is ordinarily an alumina type onto which
organic chlorides have been deposited. In that the chlorides are sensitive to moisture, the feed
must be very dry. Some organic chloride is added to the feed in order to maintain catalyst
activity.
Increasing the severity of the isomerization unit will likely increase the product octane
but may likewise produce more light ends. Yields are highly dependent on feedstock
characteristics, which naturally are closely related to the characteristics of the original crude;
paraffmicity, aromaticity, etc. Poor feed quality will usually yield net liquid percent recovered
in the mid-80's or less, while good feed quality may yield net liquid percent recovered in the
mid- to upper 90's (the rest being cracked to gaseous hydrocarbons). The key control variable is
probably the process temperature, in that raising it increases severity and promotes
hydrocracking side reactions. Raising the process pressure may increase catalyst life but will
also likely promote hydrocracking reactions, which reduce the net liquid yield. While increased
hydrogen partial pressure may extend catalyst life, it nevertheless promotes hydrocracking side-
reactions that reduce net liquid yield. Run lengths can be extended using as low temperature as
possible with moderate hydrogen partial pressure and reduced space velocity. This may or may
not seem obvious, but extending run lengths this way has drawbacks as far as product quality
and net yield of octane-barrels is concerned.13
6.3.2.3 Reformate Post-Fractionation with Benzene Saturation
Another method for reducing reformate benzene is to post-fractionate reformate into
heavy and light cuts; the light, C6, cut would contain the reformate benzene which could be
treated to remove benzene, while the C7+ stream would be blended directly into gasoline. An
important question associated with this methodology is the efficiency that the benzene could be
removed from the rest of the reformate, preserving the C7s. Based on vendor information, a
typical reformate splitter would be designed to capture about 96 percent of the benzene while
only capturing 1 percent of the toluene in the C6 stream. The refinery would design this unit as
appropriate for the refinery considering their particular economics and refinery situation. The
C6 stream would then be sent to a benzene saturation unit to saturate the benzene into
cyclohexane. There are two technologies for doing this. One is named Bensat and is licensed by
UOP. The other is named CDHYDRO and is licensed by CDTech,
Bensat
UOP has put their Bensat™ process forward as a way to reduce the benzene content of
gasoline. The process was originally developed to reduce to below six percent the benzene
concentration in the feedstock to their Penex™ isomerization unit (the Penex unit is capable of
6-25
-------
Final Regulatory Impact Analysis
saturating the rest). The process saturates the benzene converting it into cyclohexane, which can
then be fed to the Penex™ unit.
Although the process was originally designed for Penex™ feed, the vendor has modified
it to be used to saturate the benzene in a light reformate cut. UOP reported in a bulletin
published on one of their websites14 that a Bensat™ unit can be designed to handle from 5% to
30% benzene in the feed. Although not stated, it was implied that the benzene content could be
reduced to below six percent. We have received personal communications indicating that while
the benzene content of light reformate will normally vary, an average range would be about 15%
to!8%.
The process is carried out in a standalone reactor and according to UOP the process uses
a commercially proven noble metal catalyst that is benzene-selective with no side reactions.
Since there is essentially no cracking there is also essentially no coke lay-down on the catalyst to
cause deactivation. Sulfur in the feed can deactivate the catalyst, but activity can be restored by
removing the sulfur. Of course, light reformate would be very low in sulfur; other feedstocks
may need to be hydrotreated.
During start-up, hydrogen is mixed with the feed and pumped through feed/effluent
exchangers and a start-up preheater. Once the unit is up and running, the heat generated by the
process provides heat to the feed via the feed/effluent exchangers. Benzene saturation requires
three moles of H2 per mole of benzene, so makeup hydrogen is continually added to the reactor
feed. The reactor effluent is routed to a stabilizer to remove light ends. As noted previously,
some octane loss due to benzene saturation can be regained by feeding the resulting cyclohexane
to an isomerization unit.15
CDHYDRO
Catalytic Distillation Technologies (CDTECH®) has two processes for reducing the
benzene content of gasoline by converting it into cyclohexane. Both are referred to as
CDHYDRO™ technologies, but one is actually specified for the selective hydrogenation of
benzene in the entire reformate to cyclohexane in a catalytic distillation column, while the other
is advertised to hydrogenate a benzene-only stream to cyclohexane in a catalytic distillation
column.
They advertise both processes online; we note that if a refiner finds it necessary to extract
the benzene from his reformate to saturate it, the process advertised to convert benzene to
cyclohexane may be of interest16. However, we will focus on the process they put forward for
reducing the benzene content of reformate, in that they claim it is possible to do without
fractionating the reformate prior to the saturation step17. This has a clear advantage by
combining a splitting column with a benzene saturation reactor which would be expected to
reduce the capital cost for this technology.
According to CDTECH® in excess of 90% of the benzene in reformate can be hydrated
and the treated Ce's removed from the final product, all in a single catalytic distillation tower;
the tower they recommend is a benzene-toluene splitter, either refitted or new. The feed appears
6-26
-------
Final Regulatory Impact Analysis
to be a mixture of low pressure hydrogen and reformate. The feed is sent to the column and the
benzene saturation reaction occurs in the reactor. The overhead stream is condensed, cooled,
and collected in a reflux or overhead accumulator drum. The accumulator off gas, mainly
unreacted hydrogen, is recycled to feed. There also appears to be an off-gas purge stream. The
reflux drum liquid is said to be primarily treated Ce's. Part of the overhead is used for tower
reflux while the balance is pumped back into the C?+ treated reformate tower bottoms. Since
this reaction process takes place in a conventionally designed C6/C7 splitter column, this column
could presumably be designed to treat the same benzene/ toluene split that a Bensat unit would
be designed for.
6.3.2.4 Benzene Extraction
The extraction of benzene from reformate for use as a petrochemical feed can be a useful
way to remove the benzene from the gasoline pool. This method is more attractive when the
refinery is located near to petrochemical complexes which use benzene as a feedstock.
Benzene extraction involves three different steps. The first step is to separate a C6
stream from the rest of reformate using a reformate splitter. This C6, benzene-rich stream is sent
to a liquid/liquid extraction unit where the benzene and any other aromatic compounds, such as
any toluene which may captured along with the benzene in the reformate, are extracted from the
rest of the hydrocarbons. This aromatic stream is then sent to a very robust distillation process
for concentrating the benzene for sale into the chemicals market.
The reformate would be split to separate the C6s from the rest of reformate. This cut
would likely be made similar to the splitter unit used for the benzene saturation unit, although
since the toluene would only be separated and not be chemically treated, refiners would have
more leeway to capture more of the benzene in this case with less effect on the rest of the stream
then with benzene saturation.
After separation, the C6 light reformate cut, containing a fairly complex mixture of
paraffins, isoparaffin, and benzene, would be fed to an extraction unit. This type of operation,
commonly known as liquid-liquid extraction is one variation on a whole host of extraction
processes used in the petrochemical industry.
The essence of the benzene extraction process is to bring the light-reformate cut into
intimate contact with a slightly miscible to completely immiscible solvent, into which the
benzene may be selectively transferred (absorbed or dissolved) from the light-reformate. Liquid-
liquid extraction is applied by several industries, including the pharmaceutical and perfume
businesses, in a variety of vessels, such as stirred mixer-settlers, high-speed rotary centrifugal
extractors, and various columns, each of which is designed for a particular type of extraction.
There are several column types from which an engineer could choose, such as static or agitated,
along with spray, sieve plate, and packed columns. For the purposes of this discussion, we will
be referring to a static column.
For our general case, the extraction column has essentially two inlet streams and two
outlet streams. One inlet stream, fed at the top of the column is the light-reformate from which
6-27
-------
Final Regulatory Impact Analysis
the benzene aromatic components are to be extracted. The other inlet stream is the lean solvent
(solvent with no aromatics in solution) which will extract the aromatics from the light-reformate.
The solvent flows upward, while the light-reformate flows downward, during which time the
two streams come into intimate contact on the surface of the tower internals.
As designed, the solvent, containing the extracted aromatics, leaves the top of the column
as the extract or "aromatic-rich" stream. The light-reformate leaves the column bottom with
only a small residual volume of aromatics remaining and may be referred to as the raffinate. It
will consist mostly of paraffins and isoparaffins that can be sent to the gasoline blending pool.
The aromatic-rich stream is then separated from the solvent, after which the solvent is
recycled back to the extractor for reuse. The benzene, subsequently separated from the other
aromatics, can be sold into the chemicals market. The benzene-free aromatics, consisting of
toluene and in some cases xylene, which have high octane blending values, can be sent to
gasoline blending or to the chemicals market as well.
Despite only being regulated to reduce the benzene content of gasoline, the refiner may
choose to also extract toluene and xylenes. Taking such a step would cause a much larger
impact on the octane level of the refinery's gasoline and this octane loss would have to be
recovered. This may be possible using the octane recovery technologies summarized below.
This may improve the economics for reducing benzene levels, particularly because xylenes are
valued more than benzene. Extracting the C6 - C8 aromatics may allow omitting the reformate
splitter since refineries omitting the heavy straight run naphtha from the reformer feed (omitting
the C9+ fraction) could send all the reformate to the extraction unit. The extraction unit would
have to be designed to be much larger and of course the downstream distillation unit would have
to be much larger as well.
There are three proprietary extraction processes available. They are the Udex, the
Sulfolane, and the Carom processes. The di-, tri-, and tetra-ethylene glycol isomers are used as
solvents.
Extractive distillation provides what appears to be a very reasonable alternative to full
liquid-liquid aromatics extraction. According to one source, "Liquid-liquid extraction (LLE)
was for many years the primary choice for aromatics recovery, because the solvents available
during that time were not suitable for separating a wide-boiling range feedstock in the extractive
distillation mode of operation. To do so required making narrow boiling feed fractions sent to
separate extractive distillation units." "However, solvent technology has improved, and the
availability of new solvent blends makes it feasible and more profitable to employ extractive
distillation to aromatics separation."18
In short, when certain mixtures cannot be easily separated by ordinary distillation, either
because of low relative volatility or the presence of a homogeneous azeotrope, it may be possible
to effect a separation by the use of extractive distillation. According to Perry's "In extractive
distillation, the agent or 'solvent' is considerably less volatile than the regular feed components
and is added near the top of the column. Because of its low volatility, the agent behaves as a
typical heavier-than-heavy key component and is also readily separated from the product
6-28
-------
Final Regulatory Impact Analysis
streams. A typical extractive distillation might be a unit for separating benzene and cyclohexane
using phenol as the separating agent. "Benzene and cyclohexane have nearly identical boiling
points and form a homogeneous azeotrope containing about 45 wt.% cyclohexane. However,
with the phenol present, the cyclohexane volatility is nearly twice that of benzene."19 The
benzene/cyclohexane mixture is fed at or near the center of the distillation column, while the
phenol separating agent is fed into the tower a few trays below the top. The phenol remains in
the liquid phase and flow downward over the trays and out the bottom. The overhead vapor is
essentially pure cyclohexane. The bottom phenol/benzene stream is sent to a second tower for
separation. Another source suggested using aniline for the benzene/cyclohexane separating
agent.20 A full-boiling range light reformate may be more complicated, but the principles are
essentially the same. It appears that the choice of separating agent is critical. As demonstrated
by the benzene/cyclohexane example we just described, using two different solvents, it should be
clear that there will likely be more than one choice available for any given system. An economic
argument for using extractive distillation as opposed to liquid-liquid extraction is that fewer
pieces of processing equipment are usually required.
We identified another possible means to remove benzene from reformate which also
creates a concentrated benzene stream for sale to the petrochemical market. This process uses
steam extraction instead of extractive distillation as the primary unit operation. The first step in
this process is similar to conventional benzene extraction - the reformate is distilled to
concentrate benzene in a six carbon hydrocarbon stream. However, instead of sending this
material to an extraction facility, this six carbon hydrocarbon stream is fed to a stream cracker.
The very stable benzene is not cracked in the steam cracker, while other hydrocarbons in that
same stream are nearly completely cracked to light olefins, including ethylene, propylene,
butylene and butadiene. After the steam cracker, the relative heavy benzene molecules are easily
separated from the much lighter cracked olefins using simple distillation. This process creates a
benzene stream which is 98% concentrated, as opposed to benzene extraction which creates a
benzene stream that is nearly 100% pure. However, many petrochemical manufacturers are
satisfied with benzene that is 98% pure. The potential advantage for this process is that the rich
benzene stream is created at lower cost, requiring less capital and consuming less in utilities.
There has not been any long term commercial demonstrations of this technology, however, six
carbon, benzene-rich reformate has temporarily been sent to a steam cracker and it has been
demonstrated in practice over the short term.2122
6.3.2.5 Low-Pressure Reformer Operation
Lowering the pressure at which the reformer operates is another means of controlling the
benzene content. Lower pressure operation would provide some benzene reduction by reducing
the benzene formed from the hydrodealkylation (cracking) of heavier aromatics to benzene.
Beyond retarding the hydrodealkylation reaction, low pressure is an effective means of
increasing hydrogen and liquid yields, but can hurt catalyst cycle lengths. Lowering process
pressure in a semi-regen unit is reported to provide from 50% to 70% benefits of a continuous
catalyst regeneration reformer.
However, it is somewhat difficult to lower the pressure of an early-design semi-regen unit
6-29
-------
Final Regulatory Impact Analysis
below a certain level. The early generations of reformers were designed for pressures in the
range of 350 psi (as an example). Higher pressure usually allowed design engineers to specify
small diameter pipe. Lowering the pressure changes the hydraulics, restricts flow, and the
reformer simply won't operate. The recycle compressors would also likely need to be changed
in order to reduce the pressure. In short, it is not a simple fix to change a unit from high-pressure
to low-pressure. Continuous regen reformers already operate at pressures considerably lower
than semi-regen units, in the range of say, 90 psi and therefore have little room for improvement.
6.3.2.6 Pre-fractionation Combined with Low-Pressure Reformer Operation
Pre-fractionation of benzene precursors combined with low pressure reformer operation
(< 100 psi) will usually produce less than 1 vol% benzene in the reformate regardless of the feed
composition. If octane can be obtained through other means, this appears to be a useful
approach.
6.4 Experience Using Benzene Control Technologies
All these benzene reduction technologies and octane generating technologies described
above have been demonstrated in refineries in the U.S. and abroad. Each of these technologies
have been used for compliance purposes for the federal Reformulated Gasoline program, which
requires that benzene levels be reduced to an average of 0.95 volume percent or lower starting in
1995. The two primary means used by refiners to produce low benzene gasoline for the RFG
program is routing benzene precursors around the reformer and benzene extraction. Benzene
saturation is another technology used to achieve benzene reductions for the reformulated
gasoline program on a limited basis.
According to the Oil and Gas Journal's worldwide refining capacity report for 2003,
there are 27 refineries in the U.S. with extraction units. Those refineries which chose extraction
often reduced their benzene to levels well below 0.95 volume percent because the value of
benzene as a chemical feedstock is high. The reformulated gasoline program also caused the
installation of a couple of benzene saturation units. There are two benzene saturation units in the
Midwest installed in refineries there to produce RFG for the markets there. California has its
own reformulated gasoline program which also put into place a stringent benzene standard for
the gasoline sold there. The Oil and Gas Journal's Worldwide Refining Report shows that four
California refineries have benzene saturation units. If we assume that those refineries producing
RFG that do not have extraction or saturation units are routing their precursors around their
reformer, then there are 28 refineries using benzene precursor rerouting as their means to reduce
benzene levels. Personal conversations with several refiners have revealed that some of the
refineries which are routing the benzene precursors around the reformer are sending that rerouted
stream to their isomerization unit for saturating the benzene and recovering lost octane. Thus,
these four technologies have been demonstrated in many refineries since the mid-90s in the U.S.
and should be considered by the refining community as commercially proven technologies.
A vendor of benzene control technology has shared with us how the refining companies
in other countries have controlled the benzene levels of their gasoline in response to the benzene
standards put in place there. In Europe, benzene control is achieved by routing the benzene
6-30
-------
Final Regulatory Impact Analysis
precursors around the reformer and feeding that rerouted stream to an isomerization unit. In
Japan, much of the benzene is extracted from gasoline and sold to the chemicals market. Finally,
in Australia and New Zealand, refiners use benzene saturation to reduce the benzene levels in
their gasoline.
6.4.1 Benzene Levels Achievable through Reformate Benzene Control
We evaluated the benzene levels achievable by refineries applying benzene control in
two different ways. One way was to evaluate the benzene levels of refineries in 2003 which are
producing low benzene gasoline to comply with the RFG requirements. The second way was to
use the refinery-by-refinery cost model to evaluate the benzene levels achievable by the various
benzene control technologies.
Refiners today are producing gasoline with low benzene levels for sale into the RFG
market. The RFG program requires that gasoline must meet a 0.95 benzene control standard.
While the benzene standard is much less stringent than the benzene control standard, many
refiners comply at a much lower level probably because they are using benzene extraction to
comply. When extracting benzene from gasoline, the high capital costs associated with
extraction provides a strong incentive to maximize the extraction of as much benzene as
possible. The low benzene levels achieved by today's refineries provide an indication of the
feasibility of complying with the benzene standard. RFG averages 0.62 volume percent benzene
- the same level as the average benzene standard.
There are 17 refineries today producing gasoline which currently averaged 0.62 volume
percent benzene or below. Of these 17 refineries with very low benzene levels, 11 are located in
PADD 3, four are located in PADD 1, and one each are located in PADDs 2 and 4. The benzene
levels for these refineries range from 0.29 to 0.62 volume percent and average of 0.51 volume
percent. The average benzene level for these refineries is well below the benzene standard. We
reviewed the list of refinery unit capacities from EIA and the Oil and Gas Journal to determine if
these refineries have benzene saturation or extraction benzene control technologies. Of the 17
refineries with benzene levels at or below 0.62 volume percent, 14 of these have benzene
extraction or saturation units, while two more are assumed to be selling reformate to other
refineries with extraction units. While this demonstrates that achieving the benzene standard is
feasible for a portion of U.S. refiners, this does not indicate that all U.S. refiners are capable of
achieving a 0.62 volume percent benzene level.
To assess the ability for the rest of the refineries to achieve a benzene level of 0.62 or
below, we used the refinery-by-refinery model. For each benzene control technology, we
assessed its ability to achieve benzene reductions. Routing the benzene precursors around the
reformer is the least severe benzene control technology. The refinery by refinery cost model
shows that refineries using this technology can reduce their gasoline benzene levels from an
average of about 1.6 volume percent to 1.1 volume percent, a 30 percent reduction. The
refinery-by-refinery cost model shows that only two refineries would be able to meet or exceed
the new benzene standard using this technology. This technology is clearly insufficient for
achieving the required benzene control by itself.
6-31
-------
Final Regulatory Impact Analysis
Those refineries with isomerization units would be able to route their rerouted benzene
precursors to this unit further reducing their benzene levels by saturating the naturally occurring
benzene in this stream. The refinery-by-refinery cost model shows that on average these
refineries would be able to reduce their gasoline benzene levels to 0.75 volume percent using this
technology combined with benzene precursor rerouting. Of these refineries, 9 would be able to
achieve the benzene standard. Averaged across the U.S. refineries, benzene precursor rerouting
can achieve about a 60 percent reduction in reformate benzene levels. When benzene precursor
rerouting is combined with isomerization, about an 80 percent reduction in reformate benzene
levels is possible. While this benzene precursor rerouting combined with isomerization can
achieve a significant reduction in refinery benzene levels, the application of further benzene
control technologies is still required to enable the U.S. refining industry to achieve the benzene
control standard. The reason why these combined benzene control technologies are incapable of
achieving a significant enough benzene reduction is because they do not address the benzene
formed from reforming the heavy part of reformate.
We assessed the benzene reduction capacity of benzene saturation and benzene
extraction. These two technologies are able to achieve a deeper reduction in gasoline benzene
levels because they treat all the benzene in reformate - that formed from the six carbon
hydrocarbons, that formed from the cracking of heavier aromatics to benzene in heavy reformate,
and the naturally occurring benzene which is in the feed to the reformer. Our analysis of these
benzene control technologies reveals that they are able to reduce reformate benzene levels by 96
percent. The refinery-by-refinery model shows that for those refineries that were found eligible
for using benzene saturation, they were able to reduce their gasoline benzene levels from about
1.6 volume percent to 0.5 volume percent, a 60 percent reduction. For refineries identified as
eligible as using benzene extraction, the refinery-by-refinery cost model estimates that they are
capable of reducing their gasoline benzene levels from 0.9 volume percent to 0.5 volume
percent, a 40 percent reduction. The refineries eligible for benzene extraction are already low in
benzene because many of them are using extraction today, or they are selling a benzene-rich
reformate stream to a neighboring refinery which is extracting the benzene from this stream.
However, the refinery-by-refinery cost model estimates that they are able to achieve further
benzene reduction, by revamping their benzene extraction units to do so. While the use of
benzene extraction is limited to refineries on the East and Gulf Coasts, where they have access to
the petrochemical markets, the use of benzene saturation is not limited. Therefore, each refinery
in the U.S. is able to install one of these two benzene control technologies. We assessed the
benzene reduction capacity of using these two maximum reformate control technologies.
We found that, on average, U.S. refineries could achieve a benzene level of 0.50 volume
percent based on the maximum level of benzene control from reformate, assuming that benzene
saturation or extraction was applied in each refinery in the country. However, this average was
obtained by averaging refineries with benzene levels both above and below 0.50 volume percent
ranging between 0.19 to 0.85 volume percent benzene. To illustrate the benzene levels
achievable by the application of benzene extraction and benzene saturation in each refinery in
the U.S., we plotted the estimated final benzene level for each refinery against their cumulative
gasoline volume from low to highest benzene level in Figure 6.4-1. To provide a perspective for
how the gasoline benzene levels for U.S. refineries compare to the benzene standard, we
provided a line at 0.62 volume percent benzene.
6-32
-------
Final Regulatory Impact Analysis
Figure 6.4-1. Benzene Levels Achievable by U.S. Refineries Applying
Benzene Extraction and Saturation
0.00
1000 2000 3000 4000 5000 6000
Cumulative Volume (thousand barrels/day)
7000
8000
As shown in Figure 6.4-1, the refinery-by-refinery cost model estimates that if reformate
were treated with benzene saturation and benzene extraction, 8 refineries would continue to have
benzene levels above 0.62 volume percent benzene. Under the ABT program, this would not be
an issue since those refineries with benzene levels above 0.62 could purchase credits from
refineries with benzene levels below the 0.62 benzene standard. However, credits must always
be available for these refineries to show compliance with the new benzene program. While we
believe that credits would be available, it is still possible to show that each refinery could attain
the benzene standard with additional benzene control options available to them.
6.4.2 Other Benzene Controls
We have identified other means that could be used to reduce gasoline benzene levels in
addition to the technologies discussed above and modeled in the refinery-by-refinery cost
model.2324 Although we have not quantified their costs, they could be more expensive and
therefore less attractive for achieving benzene reductions than the reformer-based treating
technologies identified above.
We believe that four light gasoline streams are possible candidates for benzene reduction.
At some point in most modern refineries, at least one and sometimes all four of these streams
can be found. They are light-straight run (LSR) naphtha, light coker naphtha (LCN), light
6-3
-------
Final Regulatory Impact Analysis
hydrocrackate (LHC), and light cracked gasoline (LCG). The actual distillation composition of
each stream varies somewhat from refinery to refinery, and can vary within the same refinery,
usually as a function of seasonal changes and crude compositional variations. Upon enquiring of
just a few refiners as to an approximate boiling range, we found that currently light naphtha
streams vary from a C5 (80 °F-90 °F) initial boiling point (IBP) to as high as 340 °F final boiling
point (FBP). The range for most of the streams was around Cs-200 °F. We believe this reflects
post-MSAT I operations; a pre-MSAT I nominal boiling range for these streams was around C5-
180 °F. The benzene concentration in each of these light streams is, typically: LSR may range
from 0.5 vol% to 2.5 vol% (typically 1.1 vol%); LHC from 0.1 vol% to 5.5 vol% (typically 2.4
vol%); and LCN from 0.2 vol% to 2 vol% (typically 2.0 vol%). These may seem quite high, but
the relative volume of these streams is quite low.
The following includes a brief description of the units that produce these streams as well
as a brief summary of their current disposition (post Tier II) with regard to how they are cut,
processed, and blended. We don't intend to discuss all of the operating conditions or product
streams associated with the units. Rather, we will focus mainly on the streams we've
highlighted and on process conditions in the units or tower sections from which they flow. We
then suggest ways refiners may be able to modify the boiling ranges of these streams and
perhaps install additional equipment to reduce the overall benzene concentration of their
gasoline pool sufficiently to comply with this rule.
Light Straight Run Naphtha (LSR)
LSR is derived from crude oil. Although most crude oils contain at least some benzene,
it is seldom reported as a separate crude component. In the past, naturally occurring benzene,
regardless of its concentration, was a desirable component, of otherwise little concern, and
usually ended up in gasoline. Nevertheless, we believe that in order to comply with this rule, a
few refiners may need to consider removing the benzene that comes in with their crude.
In a common crude unit configuration, a preflash tower overhead/topped crude cut point
of about 180 °F separates the LSR (consisting of mostly €4 and Cs isomers) from the whole
crude feed. This cut point also fixed the IBP of the topped crude (and subsequently the HSR) at
about the same 180 °F25. A stabilizer or stripper take the C4's and lighter, overhead, for feed to
the saturated gas plant. The stripper bottoms, or Cs's, are either isomerized or blended directly
into gasoline. As previously mentioned, the 180 °F cut point, leaves most of the benzene and
benzene precursors in the topped crude.
Subsequently, the topped crude was fed to the main crude fractionator, from which the
HSR, with the benzene and benzene-precursors, are taken overhead, fed to a naphtha
hydrotreater, and then to a reformer. If the stabilized LSR requires desulfurization, it will be
hydrotreated with the HSR, following which they were split out for isom feed.
As described above, refiners can comply with the MS ATI benzene restrictions by
shifting the preflash LSR/topped crude cut from 180 °F to somewhere around 200 °F to 210 °F,
to keep the benzene and benzene precursors in the LSR and out of the reformer. The
debutanized LSR, consisting of Cs's and Ce's, could then be blended directly into gasoline, or
fed to an isom unit to saturate the benzene and to convert the straight-chained Cs/Ce paraffins
into isoparaffms, in order to recover some of the octane lost to benzene removal.
6-34
-------
Final Regulatory Impact Analysis
Alternatively, if refiners are putting in a reformer post-treat benzene treatment unit, either
a benzene saturation unit or an extraction unit, they can adjust the endpoint of their LSR higher
to route the small amount of benzene in LSR into the heavy straight run naphtha so that it would
be sent to the benzene posttreaters. The stabilizer or stripper that most refiners use to separate
the LSR from the rest of the naphtha is likely not capable of making a sufficiently hard cut to cut
the benzene in LSR into the heavy straight run naphtha without cutting some C5s into heavy
straight run as well. Thus refiners would likely have to install a light naphtha splitter to
accomplish this.
Light Hydrocrackate (LHC)
Hydrocrackers are designed to use high temperature and high hydrogen partial pressure,
in the presence of hydrocracking catalyst, to convert low-value heavy oil into a variety of light
products including diesel, jet fuel or kerosene, and gasoline. If process conditions are
sufficiently severe, such as when producing primarily diesel, benzene formed during
hydrocracking will likely be saturated. Under less severe conditions, such as when producing
mostly gasoline, benzene likely won't be saturated and will end up in the naphtha; olefms are
usually saturated under all hydrocracking conditions. In that the hydrocracking process
ultimately saturates any olefms produced during cracking, LHC is actually somewhat similar to
LSR.
LHC has a nominal boiling range of Cs-180 °F, while heavy hydrocrackate (HHC) has a
boiling range from around 180 °F-390 °F. Because the HHC normally has low octane, it is
usually mixed with heavy straight run naphtha and fed to a naphtha hydrotreater and reformer.
The cut between LHC and HHC is made with a main fractionator unit which also makes the cuts
between the HHC and the heavier compounds exiting the hydrocracker unit. There are two
means for further reducing the benzene levels of the LHC. A refiner could shift the
aforementioned LHC-FBP from 180 °F to around 200 °F to keep any benzene or benzene
precursors in the LHC. The LHC could then be fed with the similar Cs/Ce-LSR stream from the
crude unit to an isom unit for benzene saturation and octane improvement. If the refiner does not
have an isomerization unit, or if it is of insufficient capacity to treat both the LSR and the LHC,
then the refiner would not be able to treat the LHC that way. Alternatively, the refinery could
cut the LHC lighter so that all the benzene would end up in the HHC and be treated with the rest
of the reformate. However, the fractionation column is not designed to make fine adjustments in
distillation temperature, nor is it capable of making hard cuts to cut the benzene into the HHC
without also cutting the lighter hydrocarbons into the HHC, which is undesirable for feed to the
reformer. Thus, it would likely be necessary to add a naphtha splitter to make the appropriate
distillation cut the benzene into the HHC.
Light Cracked (LCG) Gasoline and Heavy Cracked (HCG) Gasoline
To produce gasoline, most fully integrated refineries have FCC's to catalytically crack
heavy atmospheric and vacuum gasoil from the crude and vacuum units. The volume of benzene
produced by an average FCC is ordinarily quite low when compared with other "cracking" type
units, but can be somewhat higher in severe, high-conversion operations. Prior to Tier II,
debutanized or depentanized, full-range FCC cracked gasoline was usually sent directly to
gasoline blending. To comply with Tier II sulfur restrictions, many refiners were able to split the
6-35
-------
Final Regulatory Impact Analysis
full-range stream into LCG, the cut with most of the olefins, and HCG, the cut in with most of
the sulfur. The LCG is usually caustic washed (with either a Merox or Merichem unit) to
remove mercaptans and sent directly to gasoline blending. Only the HCG was desulfurized, to
avoid LCG olefin saturation and the consequent octane loss.
Controlling the benzene in the FCC cracked naphtha presents a different set of issues. If
the benzene is cut into the LCG, it would need to be severely hydrotreated to saturate the
benzene. This could be quite costly, since under these conditions the olefins would also
undoubtedly be saturated, ultimately reducing the finished octane. Many refiners would find this
unacceptable, given the contribution LCG usually makes to the gasoline blending pool.
Although, currently, there doesn't appear to be an easy, inexpensive way to remove benzene
form LCG, without some reduction in octane, there are a few vendors that claim they can
minimize the loss. In some cases, the capital costs are a little higher than those for hydrotreating
or isomerization units, but they are reported to be offset by significantly lower operating costs.
The HCG is usually hydrotreated and sent to gasoline blending, once the LCG has been
removed. If the benzene is cut into the HCG and it is severely hydrotreated to saturate the
benzene, the product would be quite similar to HHC and would likely need to be routed to a
reformer. Reformer capacity could easily become an issue. While olefin levels are much lower
in HCG, there still are enough olefins in this refinery stream to cause higher octane losses than
the straight run naphtha streams.
A possible means for reducing the benzene in FCC naphtha has been hypothesized
through the alkylation of the benzene. As proposed, this process would first separate the
benzene and other six carbon compounds from the rest of the FCC naphtha. The five carbon and
seven carbon and heavier compounds in the rest of the FCC naphtha would continue to be
blended into gasoline. This six carbon stream, which is estimated to contain 2 to 5 percent
benzene, would be reacted over the appropriate catalysts with FCC offgas, which contains
hydrogen, methane, ethane, and ethylene, propane and propylene. The benzene would react with
the olefins, which are mainly ethylene and propylene, creating ethylbenzene and cumene
(propylbenzene). Since these alkylated benzene compounds are no longer benzene, they are
blended into the gasoline pool where they have increased the octane of gasoline slightly over the
benzene that they replaced. There are several unknowns with this concept. One unknown is
what catalyst would be best for catalyzing this reaction quickly, with few side reactions, in the
presence of some residual sulfur and nitrogen containing compounds. The second is identifying
the operating conditions that would be best for this reaction. The third is to determine the
operation run lengths for this process with the identified catalysts operating conditions. Since
these basic processing elements have not yet been answered, this potential FCC unit benzene
control technology must be further developed before it is ready for installation in refineries.26
Light Coker Naphtha (LCN)
Cokers thermally crack low API Gravity, high-sulfur asphaltic crude, vacuum unit
residuum (also usually asphaltic), and, in a few cases, FCC decant or heavy cycle oil to produce,
among several valuable products, coker naphtha. Other than having more sulfur, fewer olefins,
and a few other contaminants, it also contains some benzene. The LCN cut is ordinarily quite
6-36
-------
Final Regulatory Impact Analysis
low-volume; thus, prior to Tier II, regardless of sulfur content or the presence of other minor
contaminants, it could actually be sent directly to gasoline blending or mixed with isom unit
feed, with no real negative effects. The heavy coker naphtha (HCN) is usually sent to a naphtha
hydrotreater and, subsequently, to a reformer.
To comply with Tier II, refiners set the LCN-FBP to around 190 °F-200 °F to capture the
thiophenes (along with the benzene and benzene precursors), and sent it to the FCC naphtha
hydrotreater. The relatively mild FCC-hydrotreater conditions allowed the benzene to pass
through, unsaturated, into the gasoline blending pool. We also note that while a few olefins may
be present, the volume is quite low compared with LCG and they will obviously be saturated in
the naphtha hydrotreater.
MSAT II Compliance
Perhaps the single most important factor for this discussion is that the relative volumes of
these light naphtha streams is low. On average, the plants size to handle each of these streams
separately would be relatively small and consequently capital and operating costs on a per-barrel
basis of either feed or benzene produced would most likely be inordinately high. This might not
be the case for large refiners though.
The refiners that choose to comply with this rule using the benzene/benzene precursor
rerouting and isomerization unit benzene saturation schemes might be able to add one or more of
these additional light naphtha streams along with their LSR to feed of the isomerization unit. A
potential critical problem is that isomerization unit capacity limitations may become a problem.
We acknowledge that the fractionating towers in the other four units we've identified, may be
able to more efficiently cut the Ce's from the Cy's and other heavy ends of the various streams
we been discussing, thus reducing the effects of limited isomerization capacity.
The economics of rerouting these light naphtha streams to the isomerization unit to
saturate benzene are not favorable, especially given the high cost of building small units as well
as the prospects of overall system octane reduction, and hydrogen consumption in the
isomerization unit. We estimate that it could cost from $100 to $135 per barrel of benzene to
control the benzene in LHC; for LSR, we estimate it could cost from $45 to $222 per barrel of
benzene. These costs are at middle and the high end of the marginal costs that would compete
with the technologies that our model shows would be used to comply with the final rule benzene
control program. These costs would likely be much more attractive for a large refinery with high
benzene levels in their LSR and LHC.
For the LSR, LHC, and LCN, we suggest that perhaps the best pathway to compliance
may be to return the benzene to the reformer. Then, depending on the specific refinery
economics, the benzene could be either saturated and sent to the gasoline pool or extracted for
sale in the chemical market. The cut point between each of the light and heavy streams would be
set at or even slightly lower than 180 °F; basically, the opposite of what we previously discussed.
While we acknowledge that keeping Cs's out of the reformer is desirable, depending on the
stage efficiencies of the various fractionating towers, some Cs's may find their way into the feed.
If some C5s are sent to the reformer they can be tolerated, and in any case, there is a good
chance the Cs's can be recovered from the naphtha hydrotreater stabilizer overhead, upstream of
the reformer. The net stabilizer overhead, usually a gaseous Cs-Cs stream, could be sent to the
6-37
-------
Final Regulatory Impact Analysis
gas concentration unit for Cs recovery, if such isn't already the case. We estimate that benzene
controlled by saturation could cost, very roughly, from $70 to $350 per barrel of benzene. To
control by extraction could cost from $30 to $900 per barrel of benzene. The re-cut LSR, LHC,
and LCN could be sent to isomerization for octane improvement. The great variance in costs is
due to the range in light naphtha stream volume and benzene level.
While the cost analysis we conducted for reducing the benzene levels of these light
naphtha streams was only preliminary, the cost analysis suggests that the treatment of benzene in
LSR, LHC, and LCN could be cost-effective. If and when we reconsider setting more stringent
toxics control standards for gasoline, we should revisit whether the benzene standards we set
could be more stringent considering the treatment of benzene in these light naphtha streams.
For our feasibility analysis, we discovered that 8 refineries would not be able to comply
with the 0.62 average benzene control standard, even when applying maximum reformate
benzene control, unless if credits were available. Each refinery should be able to achieve the
average standard without relying on credits. Therefore we assessed the benzene levels
achievable by applying light naphtha benzene control technologies, as discussed above, or other
benzene control means that we identified that would be available to them.
One of these other benzene control opportunities would apply for those refineries using
benzene saturation or extraction. They could achieve additional benzene reduction with these
units by capturing more of the benzene in the reformate splitter and sending this additional
benzene to their saturation or extraction unit. Refiners attempt to optimize the capital and
operating costs with the amount of benzene removed when splitting a benzene-rich stream out of
the reformate stream for treating in a benzene saturation or extraction unit. To do this, they
optimize the distillation cut between benzene and toluene, thus achieving a benzene reduction of
about 96 percent in the reformate while preserving all but about 1 percent of the high-octane
toluene. However, if a refiner was to be faced with the need for additional benzene reductions, it
could change the distillation cut in their existing reformate splitter to send the last 4 percent of
the benzene to the saturation or extraction units. This action though would also capture more of
the seven carbon hydrocarbons, resulting in the saturation of the toluene contained in the seven
carbon hydrocarbons. Refiners using this strategy to capture more of the benzene in the
reformate splitter would have to have sufficient capacity downstream in the saturation or
extraction units to process this additional volume, although refiners normally design their units
with some excess capacity. They could design either their reformate splitter, or their benzene
saturation or extraction units with this end in mind. On the one hand, they could design their
reformate splitter to be larger to make a "hard cut" thus capturing virtually all the benzene and
rejecting virtually all the toluene; sending only the additional volume of benzene to their
downstream saturation or extraction unit. This option would entail increased capital and
operating costs for their reformate splitter, but only a very slight increase in capital and operating
costs for the benzene saturation or extraction unit.
Another means for further reducing the benzene levels for 6 of these 8 refineries is to
reduce the benzene content of the LSR naphtha stream. Refiners could use additional distillation
equipment to cut the benzene in the LSR naphtha into the heavy straight run naphtha where it
6-38
-------
Final Regulatory Impact Analysis
would be treated along with the rest of the reformate using benzene saturation or extraction. For
each of the 6 refineries which the refinery-by-refinery cost model shows could not achieve 0.62
volume percent benzene, we estimate the extent that benzene levels could be further reduced by
addressing the benzene in light straight run naphtha summarize this in Table 6.4-1.
Another means for further reducing the benzene levels for 4 of these 8 refineries which
have a hydrocracker is to reduce the benzene content of the LHC and LCN naphtha streams. For
each of the 6 refineries with a hydrocracker or coker which the refinery-by-refinery cost model
shows could not achieve 0.62 volume percent benzene, we estimate the extent that benzene
levels could be further reduced by addressing the benzene in light hydrocrackate and summarize
this in Table 6.4-1.
Another possible option for these refineries to further control benzene might be to control
the benzene content in naphtha from the fluidized catalytic cracker, or FCC unit. As we
discussed above, segregating a benzene-rich stream from FCC naphtha for sending to a benzene
saturation unit would saturate the olefms in this stream, in addition to the benzene, causing an
unacceptable loss in octane value. We learned that one refinery is operating their FCC unit very
severely to produce a high octane (92 octane number) gasoline blendstock. This resulted in this
particular FCC naphtha having a benzene content of 1.2 volume percent. This refiner could
change the operations of their FCC unit (change the catalyst and operating characteristics) to
reduce the severity and produce slightly less benzene and make up the octane loss in other ways,
such as blending in ethanol.27 We do not know if any of the refineries which the refmery-by-
refmery cost model has identified as not being able to achieve the 0.62 benzene standard using
reformate benzene control are operating their FCC units this way. Thus, we cannot estimate that
any of these refineries could reduce their gasoline benzene levels by reducing the severity of
their FCC units. Our conclusion after carefully considering treating this stream is that we cannot
assume that LCN nor HCN can be treated to lower the benzene contained in this stream.
For each of the 8 refineries which the refinery-by-refinery model shows could not
achieve 0.62 vol% benzene using maximum reformate control, we estimate the extent that
benzene levels could be further reduced based on the discussion above. Table 6.4-1 summarizes
the benzene levels achievable by each of these refineries by capturing some of the remaining
benzene and treating it in a saturation unit or extracting it from gasoline.
6-39
-------
Final Regulatory Impact Analysis
Table 6.4-1. Additional Benzene Reduction Achievable by non-Reformate Means of
Control for Refineries Unable to Achieve the Average Standard using Reformate Control
Refinery Number
1
2
3
4
5
6
7
8
Gasoline Benzene
Level after
Reformate Benzene
Control
0.66
0.69
0.68
0.67
0.85
0.71
0.75
0.67
Treating last 4% of
Reformate Benzene
-0.04
-0.07
-0.02
-0.01
-0.09
-0.06
-0.09
-0.04
Treating 99.5% of
Light Straight Run
Naphtha Benzene
-0.07
N/A
-0.18
-0.09
N/A
-0.06
-0.24
-0.16
Treating 99.6% of
Light Hydrocrackate
Benzene
-0.14
N/A
N/A
-0.20
-0.71
N/A
-0.41
N/A
6.5 Averaging, Banking, and Trading (ABT) Program
We are finalizing a program where refiners and importers can use benzene credits
generated or obtained under the averaging, banking, and trading (ABT) program to meet the 0.62
vol% annual average standard in 2011 and beyond (2015 and beyond for small refiners). We are
also finalizing a 1.3 vol% maximum average standard which takes effect in July 2012 (July 2016
for small refiners). The maximum average standard must be met based on actual refinery
benzene levels and may not be met through the use of credits.
This regulatory impact analysis begins with a discussion of today's gasoline benzene
production levels. From there, we outline the conclusions of the refinery-by-refinery cost model
(described in more detail in Chapter 9) including a summary of refiners' projected compliance
strategies for meeting the 0.62 and 1.3 vol% gasoline benzene standards. For the ABT analysis,
we focus on when the benzene reductions would occur (some likely to occur early while others
could lag) and the resulting credit generation/demand scheme. We also describe the gradual
phase-in of the 0.62 vol% standard as a result of early credit use and demonstrate how such a
program is more cost effective than a program lacking an early credit program or ABT program
all together. We provide explanation on how early and standard credits are generated as well as
how a refinery would compute their credit demand, if they should choose to rely on benzene
credits. Finally, we present our predictions on how the credit generation/trading scheme would
work via company to highlight our certainty that credits will likely be available to those in need.
6.5.1 Starting Gasoline Benzene Levels
To begin the ABT analysis, we started by examining current gasoline benzene levels. In
2004, the benzene content of gasoline produced by 113 U.S. refineries located outside of
California ranged from 0.34 to 4.04 vol% with an overall volume-weighted average of 1.00 vol%
6-40
-------
Final Regulatory Impact Analysis
as shown in Table 6.5-1 .
Table 6.5-1. 2004 Gasoline Benzene Production Levels
PADD1
PADD2
PADD3
PADD4
PADD 5 - CA
Total
No. of Refineries by Gasoline Benzene Level (vol%)
<0.5
3
1
4
0
1
9
0.5 to <1
4
6
20
1
0
31
1 to<1.5
2
8
9
4
2
25
1.5to<2
1
9
6
7
2
25
2 to <2.5
2
1
1
2
2
8
2.5+
0
1
1
2
1
5
Benzene Level (vol%)
Min
0.39
0.41
0.34
0.88
0.39
0.34
Max
2.26
2.86
2.86
4.04
3.66
4.04
Range
1.87
2.46
2.52
3.15
3.27
3.69
Avg
0.67
1.26
0.85
1.56
1.80
1.00
This data, as well as all the data presented from this point forward, includes 16 U.S.
refineries that we project will meet the small refiner criteria in § 80.1338°. This data includes
both reformulated gasoline (RFG) and conventional gasoline (CG), but excludes gasoline
produced by terminals as well as gasoline produced by California refineries for use outside of
California. It is also worth emphasizing that this data represents gasoline benzene production
levels by region. This is not necessarily the same as in-use gasoline benzene levels by region
due to the movement of gasoline across the country. For a more detailed discussion on projected
in-use levels considering gasoline distribution patterns, refer to section 6.10.
As shown above in Table 6.5-1, there is currently a wide variation in gasoline benzene
levels throughout the county. The variation (explained in more detail above in 6.2) is primarily
attributed to differences in crude oil quality, use of low-benzene blendstocks, benzene control
technology, and refinery operating procedures. PADDs 1 and 3 have the lowest average benzene
levels in the country. Refineries in these regions are located in close proximity to the
petrochemicals market making benzene extraction a viable strategy for reducing gasoline
benzene. Refineries in PADD 2 have the next lowest benzene levels primarily due to the
availability of low-benzene blendstocks, i.e. ethanol. PADDs 4 and 5 currently have the highest
benzene levels based on the benzene-rich Alaskan crude they process and their distance from the
petrochemicals market.
6.5.2 Model-Predicted Refinery Compliance Strategies
To determine how each refinery would behave under the MSAT2 program, we relied on a
linear programming (LP) cost model (discussed in more detail in Chapter 9). The LP model
considered starting benzene levels, existing benzene-control technology as well as cost and
predicted a compliance strategy for each U.S. gasoline refinery. The model assumed that
refineries would choose the most economical strategy for complying with the 0.62 and 1.3 vol%
standards. The model predicts that 77 of the 103 refineries would make technological
c 2004 gasoline benzene production levels based on batch reports received by EPA under the RFG / Anti-Dumping
requirements.
D EPA's current assessment is that 14 refiners (owning 16 refineries) meet the small refiner criterion under § 80.1338
of having 1,500 employees or less and a crude capacity of less than or equal to 155,000 bpcd. It should be noted that
because of the dynamics in the refining industry (i.e., mergers and acquisitions) and decisions by some refiners to
enter or leave the gasoline market, the actual number of refiners that ultimately qualify for small refiner status under
the MSAT2 program could be different than these estimates.
6-41
-------
Final Regulatory Impact Analysis
improvements to reduce gasoline benzene levels. For some of these refineries, it was
economical to reduce benzene levels to < 0.62 vol%, while for others it was more economical to
reduce benzene levels to < 1.3 vol% (to meet the maximum average standard) and rely on credits
to meet the annual average standard. The model shows that the remaining 26 refineries would
simply maintain their current benzene levels - which are < 1.3 vol% on average, or in some
cases < 0.62 vol%. A summary the model-predicted refinery compliance strategies is presented
in Table 6.5-2.
Table 6.5-2. Predicted Refinery Compliance Strategies
Refinery Compliance Strategy
Make process improvement to reduce
gasoline benzene levels?
Yes, reduce Bz levels to 0.62 - 1 .3 vol%
Yes, reduce Bz levels to <= 0.62 vol%
No, Bz levels already 0.62 - 1 .3 vol%
No, Bz levels already <= 0.62 vol%
Rely on
Credits?
Yes
No
Yes
No
Total Number of Refineries
No. of Refineries by PADD
PADD 1
3
4
1
4
12
PADD 2
12
12
1
1
26
PADD 3
9
18
6
8
41
PADD 4
12
1
3
0
16
PADD 5a
4
2
1
1
8
Total
40
37
12
14
103
aPADD 5 excluding California refineries
Most refiners planning on reducing gasoline benzene levels will focus on reformate
control, since the majority of the benzene found in gasoline comes from the reformer as
explained in 6.3.1. We predict that most refiners would choose this strategy since it is capable of
getting the greatest benzene reductions and the technology is known and readily available. The
refinery cost model and this subsequent ABT analysis focuses specifically on the following
forms of reformate control: light naphtha splitting, isomerization, benzene extraction and
benzene saturation. These technologies are discussed in more detail above in section 6.3.2.
As mentioned above, the refinery cost model predicts which benzene-reducing steps each
refinery would take to meet the 0.62 and 1.3 vol% standards at the lowest possible cost. The
strategy that a refinery selects will depend on existing equipment, proximity to the
petrochemicals market, and technology costs compared to the cost of buying credits. Of the 77
refineries predicted to make technological improvements (from Table 6.5-2), 17 would pursue
light naphtha splitting, 4 would pursue isomerization, 24 would implement a combination of
light naphtha splitting and isomerization, 16 invest in benzene extraction, and the remaining 16
would invest in benzene saturation. A summary of the predicted benzene reduction strategies by
PADD is found below in Table 6.5-3.
6-42
-------
Final Regulatory Impact Analysis
Table 6.5-3. Predicted Benzene Reduction Strategies
Ultimate Benzene Reduction Strategy
Light naphtha splitting
Isomerization
Light naphtha splitting & isomerization
Benzene extraction
Benzene saturation
Total Number of Refineries
No. of Refineries by PADD
PADD1
2
0
0
3
2
7
PADD 2
5
1
14
1
3
21
PADD 3
3
3
6
12
3
27
PADD 4
6
0
4
0
3
10
PADD 5a
1
0
0
0
5
1
Total
17
4
24
16
16
77
aPADD 5 excluding California refineries
The strategies listed above in Table 6.5-3 are ultimate benzene control strategies.
However, refineries may also make additional operational changes (requiring zero cost) that are
not necessarily captured in Table 6.5-3. For example, a refinery ultimately pursuing benzene
extraction may also make early operational changes (e.g., LNS, isomerization or both) to reduce
gasoline benzene levels prior to making their final investment. In this case, only their final
control strategy (benzene extraction) has been reflected in Table 6.5-3. Likewise, refineries may
complete their process improvement as a series of small steps. For example, a refinery pursuing
light naphtha splitting may make early operational changes and postpone their final investment
until later. In this case, LNS (the overall strategy) would only be listed once in Table 6.5-3. A
refinery's ability to implement their benzene control technology sooner than required is a
function of cost and lead time. A more detailed discussion on the implementation of benzene
control technologies and the resulting phase-in of the benzene standards is found below.
6.5.3 Predicted Reductions in Gasoline Benzene
Refineries will need lead time to complete refinery modifications and/or invest in new
technology for meeting the 0.62 and 1.3 vol% standards. The rule we are finalizing provides
nearly four years of lead time for this to occur. However, in many cases there are incremental
benzene reductions that can be made earlier than required. To encourage early introduction of
benzene control technology, refiners can generate early benzene credits from June 1, 2007 to
December 31, 2010 (December 31, 2015 for small refiners) by making qualifying reductions
from their 2004-2005 individual refinery baselines. A discussion of how refinery baselines are
established and what constitutes a qualifying benzene reduction is found below in section
6.5.4.2.
The early reductions we are predicting to occur would be consistent with each refinery's
ultimate benzene control strategy but simply completed sooner than required. The early credits
generated can be used to provide the refining industry with additional lead time to make their
final (more expensive) investments in benzene control technology. As a result, some benzene
reductions will occur prior to the start of the program while others will lag (within the limits of
the credit life provisions described below). We anticipate that there will be enough early credits
generated to allow refiners to postpone their final investments by three years - the maximum
6-43
-------
Final Regulatory Impact Analysis
time afforded by the early credit life provisions. In addition, we predict that standard credits
generated during this early credit lag period (January 1, 2011 through December 31, 2013) will
allow for an additional 16 months of lead time. The result is a gradual phase-in of the 0.62
vol% benzene standard beginning in June 2007 and ending in July 2016 as shown below in
Figure 6.5-1.
Figure 6.5-1. Benzene Level vs. Time
1.05
1.00 -
0.95 -
0.90 H
> 0.85
01
N 08°
C
01
m
0) 0.75
0)
01
0.70 -
0.65 -
0.60 -
0.55 -
0.50
1.00
Jun-2007
Early
Operational
Changes
0.83
Jan-2010
Early Small
Capital
Investments
Total Credit Lag
^.76
Jul-2012 \_
Non-Small Refiner
Compliance with
1.3vol%Std.
0.74
May-2015
Non-Small Refiner
Final Investments at
End of Credit Lag
0.625
0.62
Jul-2016
Small Refiner
Compliance with
1.3vol% Std.
Jan- Jul- Jan- Jul- Jan- Jul- Jan- Jul- Jan- Jul- Jan- Jul- Jan- Jul- Jan- Jul- Jan- Jul- Jan- Jul- Jan- Jul- Jan- Jul-
06 06 07 07 08 08 09 09 10 10 11 11 12 12 13 13 14 14 15 15 16 16 17 17
Time
As shown in Figure 6.5-1 (and described in more detail below), our modeling assumes a
stepwise reduction in gasoline benzene levels over time. However, due to the inputs of many
different individual refinery decisions over time, we anticipate that a more continuous benzene
reduction pattern will actually occur, but follow the same trend.
The ABT analysis assumed that small refiners would comply with the 1.3 vol%
maximum average standard in January 2015 at the same time as the 0.62 vol% annual average
standard. However, in actuality, we are finalizing a later maximum average standard
implementation date (July 2016) for small refiners. We anticipate that this will have very little
effect on the overall credit generation/use picture and therefore have elected not to change our
ABT analysis. As a result, the phase-in of benzene control technologies (presented below) and
the subsequent credit and cost savings calculations (presented in section 6.5.4) are based on
small refiners complying with the 1.3 vol% maximum average standard in January 2015 (instead
of July 2016).
6-44
-------
Final Regulatory Impact Analysis
6.5.3.1 Early Operational Changes in Benzene Control Technology
We estimate that the first phase of early benzene reductions would occur as early as June
1, 2007. These refinery modifications would consist of operational changes made to the
reformer that could be implemented without making a capital investment. The early operational
changes we predict to occur are light naphtha splitting and isomerization. For refineries that are
already splitting light naphtha in their crude distillation columns (or have the potential to), we
assume that operational changes could be made to re-route up to 75% of the benzene precursors
around the reformer. If the refinery is equipped with an isomerization unit, we predict that this
re-routed light naphtha would also be isomerized. If no isomerization unit exists, we predict that
the light naphtha would simply be combined with the light straight run to make gasoline.
Based on the refinery cost model findings, we predict that 46 of the 103 refineries would
take advantage of the early credit generation opportunities and make early operational changes.
More specifically, 18 refineries would implement light naphtha splitting, 4 would implement
isomerization, and 24 would pursue a combination of both. These operational changes would
result in a 17% reduction in average gasoline benzene level from 1.00 vol% to 0.83 vol%. The
changes would also result in an overall 29% reduction in maximum benzene level from benzene
level variation from 4.04 vol% to 2.85 vol%. A summary of these reductions and resulting
benzene levels by PADD is found in Table 6.5-4.
Table 6.5-4. Benzene Levels after Early Operational Changes
PADD1
PADD 2
PADD 3
PADD 4
PADD 5 - CA
Total
No. of Refineries by Gasoline Benzene Level (vol%)
<0.5
3
1
4
0
1
9
0.5 to <1
4
14
23
1
1
43
1 to<1.5
3
9
10
12
3
37
1.5to<2
0
0
3
2
1
6
2 to <2.5
2
2
0
0
2
6
2.5+
0
0
1
1
0
2
Benzene Level (vol%)
Min
0.39
0.44
0.35
0.90
0.39
0.35
Max
2.17
2.49
2.85
2.59
2.10
2.85
Range
1.78
2.05
2.50
1.69
1.70
2.50
Avg
0.65
0.91
0.77
1.26
1.21
0.83
6.5.3.2 Early Small Capital Investments in Benzene Control Technology
We estimate that a second round of early benzene reductions could occur by January
2010. These refinery modifications would consist of upgrades in reformate benzene control
technology requiring a relatively small capital investment. For the purpose of this analysis, we
are defining a small capital investment as an investment in technology with an incremental cost
of < $60 per barrel of benzene reduced. The early technology changes we predict to occur
include light naphtha splitting, isomerization, and benzene extraction. We predict that refineries
could invest in dedicated columns for splitting light naphtha that would be capable of re-routing
100% of the benzene precursors around the reformer. As with the operational changes
mentioned above, if the refinery is equipped with an isomerization unit, we predict that the re-
routed light naphtha would also be isomerized. If no isomerization unit exists, the light naphtha
would be combined with the light straight run to make gasoline. .
At this time, we predict that 38 of the 103 refineries would make early technology
changes requiring a small capital investment. More specifically, 31 refineries would implement
6-45
-------
Final Regulatory Impact Analysis
light naphtha splitting and/or isomerization at this time. In addition, we predict that seven
refineries currently extracting benzene would make modifications to their existing extraction
units (costing up to $60/bbl Bz) to improve benzene separation and in turn reduce the benzene
concentration of their finished gasoline. Together these changes would result in an 8% reduction
in average gasoline benzene level from 0.83 vol% to 0.76 vol%. There would be no change in
the maximum benzene level as a result of this step. A summary of these reductions and resulting
benzene levels by PADD is found in Table 6.5-5.
1
PADD1
PADD 2
PADD 3
PADD 4
PADD 5 - CA
Total
"able 6.5-5. Benzene Levels after Early Small Capital Investments
No. of Refineries by Gasoline Benzene Level (vol%)
<0.5
3
4
5
0
1
13
0.5 to <1
5
16
24
4
2
51
1 to<1.5
2
4
10
9
2
27
1.5to<2
0
0
1
2
1
4
2 to <2.5
2
2
0
0
2
6
2.5+
0
0
1
1
0
2
Benzene Level (vol%)
Min
0.39
0.44
0.35
0.88
0.39
0.35
Max
2.17
2.49
2.85
2.59
2.10
2.85
Range
1.78
2.05
2.50
1.71
1.70
2.50
Avg
0.63
0.76
0.72
1.14
1.16
0.76
6.5.3.3 Compliance with the 1.3 vol% Maximum Average Standard
In January 2011, the 0.62 vol% standard becomes effective for refineries that do not meet
the small refiner criteria in § 80.1338. However, since these refineries will have a sufficient
amount of early credits available to them (described in more detail below in section 6.5.4.3), we
estimate that they could maintain their 2010 benzene levels until July 2012 when the 1.3 vol%
maximum average standard takes effect.
At this time, we predict that 7 of the 103 refineries would implement benzene saturation
to reduce their average benzene levels down to 1.3 vol% to comply with the maximum average
standard. At this point in the analysis we also incorporated any outstanding benzene reductions
associated with increased ethanol use in response to the Energy Policy Act of 2005.E Together
these changes would result in a 3% reduction in average gasoline benzene level from 0.76 vol%
to 0.74 vol%. The changes would also result in a 14% reduction in maximum benzene level
from 2.85 vol% to 2.45 vol%. A summary of these reductions and resulting benzene levels by
PADD is found in Table 6.5-6.
The Renewable Fuel Standard proposed on September 22, 2006 (71 FR 55552) would require 7.5 billion gallons of
renewable fuel to be blended into gasoline by 2012, the majority of which would likely be satisfied by ethanol use.
However, in AEO 2006, EIA projected that ethanol use would be 9.6 billion gallons by 2012, well exceeding the
RFS requirement. As a result, for this rulemaking we have elected to incorporate the impacts of blending 9.6 billion
gallons of ethanol into gasoline by 2012. For the ABT analysis, as refineries were predicted to make early benzene
reductions, the impacts of increased ethanol use were incorporated. For refineries not predicted to make any early
process changes, increased ethanol use was incorporated in the 2012 year.
6-46
-------
Final Regulatory Impact Analysis
Table 6.5-6. Benzene Levels after 1.3 vol% Standard Becomes Effective
PADD1
PADD2
PADD3
PADD4
PADD 5 - CA
Total
No. of Refineries by Gasoline Benzene Level (vol%)
<0.5
3
4
5
0
1
13
0.5 to <1
5
16
24
6
2
53
1 to<1.5
3
5
12
8
4
32
1.5to<2
0
0
0
1
1
2
2 to <2.5
1
1
0
1
0
3
2.5+
0
0
0
0
0
0
Benzene Level (vol%)
Min
0.39
0.45
0.34
0.81
0.34
0.34
Max
2.11
2.17
1.30
2.45
1.75
2.45
Range
1.72
1.72
0.96
1.64
1.41
2.11
Avg
0.61
0.75
0.70
1.05
1.07
0.74
Based on credit availability and the relatively high operational costs associated with
benzene saturation, we predict that the seven refineries implementing benzene saturation at this
time would postpone running the units to their maximum capacity until May 2015 (end of the
credit lag, described in more detail below in section 6.5.4.6). In the interim, these refineries
would produce gasoline with 1.3 vol% benzene on average and rely on credits to meet the 0.62
vol% annual average standard.
6.5.3.4 Small Refiner Compliance with the Benzene Standards
As mentioned above, we assumed that in January 2015, both the 0.62 vol% annual
average standard and the 1.3 vol% maximum average standard would become effective for
refineries meeting the small refiner criteria in § 80.1338. At this time, we predict that two small
refineries would implement light naphtha splitting and isomerization to reduce their benzene
levels to the maximum extent possible. Additionally, we predict that four small refineries would
implement benzene saturation to reduce their average benzene levels to 1.3 vol%. Together
these changes would result in a 1% reduction in average gasoline benzene level from 0.74 vol%
to 0.73 vol%. These changes would also result in a 47% reduction in maximum benzene level
from 2.45 vol% down to the maximum average standard of 1.30 vol%. A summary of these
reductions and resulting benzene levels by PADD is found in Table 6.5-7.
Table 6.5-7. Benzene Levels after the 1.3 vol% Standard Becomes Effective for Smalls
PADD1
PADD 2
PADD 3
PADD 4
PADD 5 - CA
Total
No. of Refineries by Gasoline Benzene Level (vol%)
<0.5
3
4
5
0
1
13
0.5 to <1
5
17
24
7
2
55
1 to<1.5
4
5
12
9
5
35
1.5to<2
0
0
0
0
0
0
2 to <2.5
0
0
0
0
0
0
2.5+
0
0
0
0
0
0
Benzene Level (vol%)
Min
0.39
0.45
0.34
0.81
0.34
0.34
Max
1.30
1.30
1.30
1.30
1.30
1.30
Range
0.91
0.85
0.96
0.49
0.96
0.96
Avg
0.61
0.74
0.70
1.03
1.06
0.73
Unlike the assumption made above for benzene saturation, we predict that the four small
refineries investing in benzene saturation will never run their units to their maximum capacity to
minimize operational costs. In the event that they did, the benzene levels in the future could be
slightly lower than what is reported here.
6.5.3.5 Full Program Implementation / Ultimate Compliance with the 0.62 vol% Standard
We estimate that the final phase of benzene reductions would occur in May 2015 at the
6-47
-------
Final Regulatory Impact Analysis
end of the early/standard credit lag (described in more detail below in section 6.5.4.6). At this
time we predict that 12 refineries would pursue benzene saturation, 9 refineries would pursue
benzene extraction, and 12 refineries would pursue light naphtha splitting and/or isomerization.
Of the 12 refineries predicted to pursue benzene saturation, five would be investing in
brand new saturation units and the other seven would be making operational changes to run their
new units (installed in July 2012) to their maximum benzene reduction potential. Of the nine
refineries predicted to pursue benzene extraction, three would be investing in brand new units
and the remaining six would be making modifications to their existing extraction units (costing
over $60/bbl Bz). Of the 12 refineries predicted to pursue light naphtha splitting and/or
isomerization, nine would be investing in new units and three would be making changes to
existing units - steps that could have been completed early but were postponed due to the early
credit trigger point (explained in more detail in section 6.5.4.1).
Together the 33 technology changes made at this time would result in a 15% reduction in
average gasoline benzene level from 0.73 vol% to 0.62 vol%. There would be no change in the
maximum benzene level as a result of this step. However, the program in its entirety would
result in a 68% reduction in maximum benzene level from 4.04 vol% to 1.30 vol%. Similarly,
the program overall would result in a 38% reduction in average gasoline benzene levels from
1.00 vol% to 0.62 vol%. A summary of resulting benzene levels by PADD is found below in
Table 6.5-8.
PADD1
PADD 2
PADD 3
PADD 4
PADD 5 - CA
Total
Table 6.5-8. Benzene Levels after Full Program Implementation
No. of Refineries by Gasoline Benzene Level (vol%)
<0.5
6
7
7
0
3
23
0.5 to <1
5
15
29
12
4
65
1 to<1.5
1
4
5
4
1
15
1.5to<2
0
0
0
0
0
0
2 to <2.5
0
0
0
0
0
0
2.5+
0
0
0
0
0
0
Benzene Level (vol%)
Min
0.39
0.41
0.34
0.60
0.34
0.34
Max
1.30
1.30
1.18
1.30
1.30
1.30
Range
0.91
0.89
0.84
0.70
0.96
0.96
Avg
0.52
0.63
0.61
0.90
0.69
0.62
6.5.4 Credit Generation/Use Calculations & Considerations
6.5.4.1 What factors impact refiners' decisions to make early process changes?
As mentioned before, a refinery's ability to make early benzene reductions depends on
the nature of the improvement(s), required lead time, and associated capital costs. However, a
refinery's decision to make early improvements depends on several other factors.
First, an early reduction must be consistent with the refinery's ultimate compliance
strategy. Our analysis assumes that refineries will make all model-predicted operational changes
requiring zero capital to reduce starting benzene levels regardless of their ultimate strategy for
meeting the 0.62 and 1.3 vol% standards. However, we assume that they will only make early
technology changes requiring a small capital investment if these changes are consistent with
their final control strategy. For example, a refinery would not invest capital in a dedicated light
naphtha splitting column (even if it was < $60/bbl Bz to incrementally reduce benzene) to reduce
6-48
-------
Final Regulatory Impact Analysis
benzene and generate credits if its ultimate strategy for complying with the 1.3 vol% maximum
average standard involved investing in a benzene saturation unit.
Second, a refinery would only make a model-predicted early change if the benzene
reduction was significant enough to allow them to generate early credits. In other words, a
refinery would not make an early benzene reduction if it did not satisfy the 10% reduction trigger
point derived in the proposal (EPA420-D-06-004, February 2006). Applying this assumption
reduced the number of predicted early refineries predicted to make operational changes from 52
to 47 and the number of refineries predicted to make early small capital investments from 40 to
39.
Third, a refinery would only make a model-predicted early change if the company had a
need for early credits, i.e., the company's average starting benzene level was higher than the 0.62
vol% standard. If a company's average benzene level was at or below the standard to begin
with, they would not have a need to generate early credits to postpone compliance since they
could do nothing and still comply with the standard in 2011 via company averaging. Applying
this assumption, one refinery which the model predicted to make both operational and small
capital investments was assumed not to do so early. This further reduced the number of
refineries predicted to make early operational changes from 47 to 46 and the number of refineries
predicted to make early small capital investments from 39 to 38.
It is worth noting that refineries constrained by these last two conditions would go on to
make the outlined model-predicted changes, just not earlier than required.
6.5.4.2 How are early credits calculated?
Before we estimate early credit generation, we must first explain how early credit
baselines and annual average benzene levels are computed and briefly how the trigger point
impacts early credit generation.
As mentioned earlier, refiners are eligible to generate early credits for making qualifying
benzene reductions prior to the start of the program. Refiners must first establish individual
benzene baselines for each refinery planning on generating early credits. Per § 80.1280, benzene
baselines are defined as the annualized volume-weighted benzene content of gasoline produced
at a refinery from January 1, 2004 through December 31, 2005. To qualify to generate early
credits, refineries must make operational changes and/or improvements in benzene control
technology to reduce gasoline benzene levels in accordance with § 80.1275.
Additionally, a refinery must produce gasoline with at least ten percent less benzene (on
a volume-weighted annual average basis) than its 2004-2005 baseline. The purpose of setting an
early credit generation trigger point is to ensure that changes in benzene level are representative
of real process improvements. Without a trigger point, refineries could generate credits based on
operational fluctuations in benzene level from year to year. This would compromise the
environmental benefits of an ABT program because the early credits generated would have no
associated benzene emission reduction value. A more detailed discussion on how we arrived at a
10 percent reduction trigger point is found in the proposal (EPA420-D-06-004, February 2006).
6-49
-------
Final Regulatory Impact Analysis
Once the 10% trigger point is met, refineries can generate early credits based on the
entire benzene reduction. For example, if in 2008 a refinery reduced its annual benzene level
from a baseline of 2.00 vol% to 1.50 vol% (below the trigger of 0.90 x 2.00 = 1.80 vol%), its
benzene credits would be determined based on the difference in annual benzene content (2.00 -
1.50 = 0.50 vol%) divided by 100 and multiplied by the gallons of gasoline produced in 2008
(credits expressed in gallons of benzene).
Under the ABT program, the first early credit generation period is from June 1, 2007
through December 31, 2007, and subsequent early credit generation periods are the 2008, 2009,
and 2010 calendar years (2008 through 2014 calendar years for small refiners). To estimate the
number of early credits that would be generated during these years, we used the 2004 refinery
model baseline (derived from benzene levels summarized in Table 6.5-1) to represent early
credit baselines. The benzene level from which early credits are calculated is the volume-
weighted average benzene concentration of all batches of gasoline produced during a given
averaging period. This is referred to as the annual average benzene concentration. To estimate
early credits, we used the benzene levels predicted by the refinery cost model to represent annual
average benzene levels. For 2007, 2008, and 2009, we have used the post-operational change
benzene levels reflected in Table 6.5-4. For 2010, we have used the benzene levels following
the early small capital investments reflected in Table 6.5-5.
6.5.4.3 How many early credits do we predict will be generated?
By applying these criteria to the refinery cost model, we estimate that refineries making
early operational changes and small capital investments in reformate technology from June 1,
2007 to December 31, 2010 could generate over 765 million gallons (MMgal) of benzene
credits. A breakdown of the early credit generation by PADD is found below in Table 6.5-9.
Table 6.5-9. Early Credits Generated by PADD (gal Bz)
PADD 1
PADD 2
PADD 3
PADD 4
PADD 5 - CA
Total
2007
1,387,041
59,878,797
24,796,242
9,601,712
11,484,773
107,148,564
2008
2,399,049
103,978,138
42,909,137
16,726,807
20,019,372
186,032,503
2009
2,420,505
105,326,076
43,314,833
16,998,147
20,356,434
188,415,995
2010
5,932,981
154,049,197
77,511,287
22,877,834
23,278,019
283,649,318
Total
12,139,576
423,232,208
188,531,499
66,204,500
75,138,597
765,246,381
In addition to the above-referenced early credits, small refiners are predicted to generate
an additional 110 MMgal of credits from January 1, 2011 through December 31, 2014, bringing
the total early credit generation to 875 MMgal. These additional early credits generated by small
refiners have not been included in Table 6.5-9 to preserve confidential business information.
6.5.4.4 How many early credits will be demanded?
Early credits can be applied to the first three years of the program to comply with the
0.62 vol% annual average standard. This is governed by the three-year early credit life provision
6-50
-------
Final Regulatory Impact Analysis
described in more detail in section 6.5.4.10. However, credits may not be used to meet the 1.3
vol% maximum average standard which begins July 1, 2012. In other words, refineries whose
benzene levels are at or below 1.3 vol% following their early technology changes in 2010 can
choose to use early credits to comply with the 0.62 vol% standard and postpone their final
investment for up to three years. Refineries predicted to be above the maximum average
standard in 2010 will not be able to rely exclusively on early credits. After July 1, 2012, these
refineries will need to reduce benzene levels to meet the 1.3 vol% annual average standard.
Once this hurdle has been met, they can choose to rely on early credits to meet the 0.62 vol%
standard.
For example, consider a refinery whose annual average benzene level was 0.80 vol% in
2010. If the refinery did not make any additional benzene reductions in the first three years of
the program, its early credit demand would be computed as follows. In 2011, its early credit
demand (expressed in gallons of benzene) would be determined based on the difference between
its annual average benzene level and the standard (0.80 - 0.62 = 0.18 vol%) divided by 100 and
multiplied by its annual gasoline production volume. The early credit demand would be the
same in 2012 and 2013, provided gasoline production did not change.
As another example, consider a refinery whose average benzene concentration was 1.60
vol% until July 1, 2012 when it was reduced to 0.80 vol% to meet the 1.3 vol% maximum
average standard. If the refinery did not make any additional reductions in the first three years of
the program, its early credit demand would be calculated as follows. In 2011, its early credit
demand would be determined based on the difference between its starting annual average
benzene level and the standard (1.6 - 0.62 = 0.98 vol%) divided by 100 and multiplied by its
annual gasoline production volume. In 2012, its early credit demand would be the difference
between the first half of the year's average benzene level and the standard (1.6 - 0.62 = 0.98
vol%) divided by 100 and multiplied by the first half of the year's gasoline production volume
plus the difference between the second half of the year's average benzene level and the standard
(0.80 - 0.62 = 0.18 vol%) divided by 100 and multiplied by the second half of the year's
gasoline production volume.F In 2013, its early credit demand would be determined based on
the difference between its final annual average benzene level and the standard (0.80 - 0.62 =
0.18 vol%) divided by 100 and multiplied by its annual gasoline production volume.
Applying this methodology to all 103 refineries, we anticipate that 579 million gallons of
early benzene credits would be demanded from January 1, 2011 through December 31, 2013 as
shown below in Table 6.5-10. In addition, we predict that small refiners would demand an
additional 39 MMgal of credits from January 1, 2015 through December 31, 2017, bringing the
total early credit demand to 618 MMgal. These additional early credits demanded by small
refiners have not been included in Table 6.5-10 to preserve confidential business information.
F This is equivalent to computing the volume-weighted annual average benzene concentration in the second year
(e.g., 1.2 vol%) and calculating the credit demand based on this value. However it's worth noting that since 2012 is
a transitional year, a refinery's computed annual average benzene concentration could feasibly be above 1.3 vol%
yet the refinery could still be on track for complying with the 1.3 vol% maximum average standard during the first
compliance period (July 1, 2012 through December 31, 2013). The first compliance period is 18 months and
subsequent compliance periods are the calendar years beginning with 2014.
6-51
-------
Final Regulatory Impact Analysis
Table 6.5-10. Early Credits Demanded by PADD (gal Bz)
PADD 1
PADD 2
PADD 3
PADD 4
PADD 5 - CA
Total
2011
13,647,236
48,090,307
86,828,577
22,394,715
30,479,498
201,440,332
2012
12,992,329
46,403,262
84,978,468
20,380,618
28,322,111
193,076,789
2013
12,412,384
44,756,871
83,351,100
18,259,995
26,074,485
184,854,835
Total
39,051,950
139,250,440
255,158,145
61,035,327
84,876,094
579,371,956
As outlined above, we predict that there will be enough early credits generated to allow
for refiners to postpone their final investments for up to three years or through January 2014 -
the maximum time afforded by the early credit life provision. In additional, we predict that there
will be a 40 percent surplus in early credits (total early credit generation is 875 MMGal, total
early credit demand over the first three compliance years is only is 618 MMGal). To the extent
that the predictions from the refinery cost model are directionally accurate, there would be a
built-in early credit compliance margin which would essentially increase the certainty that early
credits would be available to those in need, including small refiners.
6.5.4.5 How are standard credits calculated?
Beginning January 1, 2011, standard benzene credits can be generated by any refinery or
importer that overcomplies with the 0.62 vol% gasoline benzene standard on an annual volume-
weighted basis in 2011 and beyond. For example, if in 2011 a refinery's annual average benzene
level was 0.52, its benzene credits (expressed in gallons of benzene) would be determined based
on the margin of overcompliance with the standard (0.62 - 0.52 = 0.10 vol%) divided by 100 and
multiplied by the its annual gasoline production volume. Likewise, if in 2012 the same refinery
produced the same amount of gasoline with the same benzene content they would earn the same
amount of credits. The credit generation opportunities for overcomplying with the standard
continue indefinitely.
6.5.4.6 How much additional lead time would be generated by standard credits generated
during the early credit "lag"?
From January 1, 2011 through December 31, 2013 while early credits are being used, we
predict that standard credits will be generated by refineries that are already below the 0.62 vol%
standard or plan to get there by making early technology changes. A summary of the predicted
standard credit generation is found below in Table 6.5-11.
6-52
-------
Final Regulatory Impact Analysis
Table 6.5-11. Standard Credits Generated during 3-Year Early Credit Lag (gal Bz)
PADD 1
PADD2
PADD 3
PADD 4
PADD 5 - CA
Total
2011
12,548,070
7,064,755
34,125,185
0
653,573
54,391,583
2012
13,149,182
6,862,297
35,584,771
0
748,092
56,344,342
2013
13,866,802
6,656,029
37,202,521
0
836,160
58,561,511
Total
39,564,053
20,583,080
106,912,477
0
2,237,825
169,297,436
We calculate that enough standard credits will be generated during this period to extend
the credit lag by another 16 months, or through May 2015. This will essentially allow refineries
to maintain their 2010 post-operational change benzene levels a little longer following a similar
credit demand scheme to that described above in Table 6.5-10.
For the above credit generation/demand calculations as well as those presented below, we
have made a simplifying assumption that importers will play a negligible role in the overall ABT
program. In other words, that beginning in 2011 importers will bring in gasoline that is
compliant gasoline with the 0.62 vol% standard and thus will neither generate nor demand
credits.
6.5.4.7 How do we estimate ongoing standard credit generation/demand?
Once refineries make their final investments in benzene control technology in (described
above in section 6.5.3.5), nationwide gasoline benzene levels will be at 0.62 vol% on average.
We predict that this will occur by May 2015 at the end of the total credit lag. At this point, the
refinery cost model predicts that 50 refineries will be below the 0.62 vol% standard (generating
standard credits) and 53 will be above (demanding standard credits). A summary of the resulting
standard credit generation/demand scheme is found below in Table 6.5-12. We have chosen to
present 2016 standard generation/demand (based on projected gasoline production levels) since
2015 is a transitional year with two sets of predicted benzene reductions.
Table 6.5-12. Standard Credits Generated/Demanded in 2016 & Beyond i
PADD1
PADD 2
PADD 3
PADD 4
PADD 5 - CA
Total
Credits
Generated by
Refineries
<0.62 vol%
20,197,659
20,423,752
48,151,821
55,477
4,478,444
93,307,153
Credits
Demanded by
Refineries
>0.62 vol%
3,859,615
22,768,665
42,522,657
15,457,960
8,698,256
93,307,153
Net Credit
Generation (+) or
Demand (-)
16,338,044
-2,344,913
5,629,164
-15,402,483
-4,219,812
0
gal/yr)
6-53
-------
Final Regulatory Impact Analysis
Although, the above table shows credit generation and demand balancing in 2016 and
beyond, our refinery cost model actually predicts that there will be a small surplus of standard
credits if small refineries rely on early credits (as opposed to standard credits) for the first three
years of their program (2015-2017). To the extent that the predictions from the refinery cost
model are direct!onally accurate, there would be a built-in 39 MMgal standard credit surplus that
would essentially increase the certainty that standard credits would be available to those in need.
This would be an ongoing compliance margin that could effectively carry over from year-to-
year (within the 5-year standard credit life provision) provided credits were used in the order
they were generated.
6.5.4.8 What are the credit use provisions?
Refineries and importers can use benzene credits generated or purchased under the
provisions of the ABT program to comply with the 0.62 vol% gasoline benzene standard in 2011
and beyond. As mentioned earlier, credits may not be used to demonstrate compliance with the
1.3 vol% maximum average standard beginning in July 2012 (July 2016 for small refiners).
Refineries must reduce gasoline benzene levels to < 1.3 vol% on average, essentially placing a
ceiling on credit use.
All benzene credits are to be used towards compliance on a one-for-one basis, applying
each benzene gallon credit to offset the same volume of benzene produced in gasoline above the
standard. For example, if in 2011 a refinery's annual average benzene level was 0.72 vol%, the
number of benzene credits needed to comply (expressed in gallons of benzene) would be
determined based on the margin of under-compliance with the standard (0.72 - 0.62 = 0.10 vol%)
divided by 100 and multiplied by the annual gasoline production volume.
Early credits may be used equally and interchangeably with standard credits to comply
with the 0.62 vol% benzene standard in 2011 and beyond. However, based on the credit life
provisions described below, we predict that refiners would choose to use early credits first before
relying on standard credits. Likewise, we expect that refineries would choose to use standard
credits in the order in which they were generated (first in, first out) to avoid the likelihood that
they would expire and become worthless.
6.5.4.9 Are there any geographic restrictions on credit trading?
We are not placing any geographic restrictions on where credits may or may not be
traded and thus are finalizing a nationwide ABT program. If PADD restrictions were placed on
credit trading, there would be an imbalance between the supply and demand of credits. As
shown in Table 6.5-12, if there was no inter-PADD trading allowed, PADDs 1 and 3 would have
a surplus of standard credits while PADDs 2, 4, and 5 would have a shortage of credits. This
would result in surplus credits expiring and becoming worthless in PADDs 1 and 3 while at the
same time PADDs 2, 4, and 5 would experience insufficient credit availability. This would force
refineries with more expensive benzene technology costs in PADDs 2, 4, and 5 to comply
increasing the total compliance costs. Overall, restricting credit trading by PADD would result
in a more expensive, less flexible, and less efficient program.
6-54
-------
Final Regulatory Impact Analysis
Additionally, we believe that restricting credit trading could reduce refiners' incentive to
generate credits and hinder trading essential to this program. In other fuel standard ABT
programs (e.g., the highway diesel sulfur program), fuel credit trading restrictions were
necessary to ensure there was adequate low-sulfur fuel available in each geographic area to meet
the corresponding vehicle standard. Since there is no vehicle emission standard associated with
this rulemaking that is dependent on gasoline benzene content, we do not believe there is a
crucial need for geographic trading restrictions. We project that under the proposed nationwide
ABT program, all areas of the country would still experience large reductions in gasoline
benzene levels as shown in Table 6.5-13.
Table 6.5-13. Predicted Reductions in Benzene Level by PADD
PADD 1
PADD 2
PADD 3
PADD 4
PADD 5 - CA
Total
Starting Bz
Levels (vol%)a
0.67
1.26
0.85
1.56
1.80
1.00
Ending Bz
Levels (vol%)b
0.52
0.63
0.61
0.90
0.69
0.62
Overall %
Bz Reduction
22%
50%
28%
42%
62%
38%
6.5.4.10
aBased on 2004 gasoline benzene production levels
bBased on model-predicted benzene reductions
What are the credit life provisions?
Early credits must be used towards compliance within three years of the start of the
program; otherwise they will expire and become invalid. In other words, early credits generated
or obtained under the ABT program must be applied to the 2011, 2012, or 2013 compliance
years. Similarly, early credits generated/obtained and ultimately used by small refiners must be
applied to the 2015, 2016, or 2017 compliance years. No early credits may be used towards
compliance with the 2014 year. Our intent is that a break in the early credit application period
will funnel surplus early credits facing expiration to small refiners in need.
Standard credits must be used within five years from the year they were generated
(regardless of when/if they are traded). For example, standard credits generated in 2011 would
have to be applied towards the 2012 through 2016 compliance year(s); otherwise they would
expire and become invalid. To encourage trading to small refiners, there is a credit life extension
for standard credits traded to and ultimately used by small refiners. These credits may be used
towards compliance for an additional two years, giving standard credits a maximum seven-year
life. For example, the same above-mentioned standard credits generated in 2011, if traded to and
ultimately used by a small refiner, would have until 2018 to be applied towards compliance
before they would expire.
6.5.4.11 Consideration of credit availability
Our ABT analysis presented here assumes perfect nationwide credit trading. In reality,
6-55
-------
Final Regulatory Impact Analysis
we recognize that not all credits generated may necessarily be available for sale. Since EPA is
not managing the credit market, credit trading will be at the generating parties' discretion. With
such a program, there are usually concerns that credits may not be made available on the market,
especially among single refinery refiners. To determine the likelihood of credit availability, we
have assessed the model-predicted credit generation and trading by company. To preserve
confidentiality, each of the 39 refining companies have been assigned a random/discrete
company ID. The resulting early and standard credit generation by company is found in Tables
6.5-14 and 6.5-15, respectively.
6-56
-------
Final Regulatory Impact Analysis
Table 6.5-14. Early Credit Trading by Company
Company
Company 1
Company 2
Company 3
Company 4
Company 5
Company 6
Company 7
Company 8
Company 9
Company 10
Company 11
Company 12
Company 13
Company 14
Company 15
Company 16
Company 17
Company 18
Company 19
Company 20
Company 21
Company 22
Company 23
Company 24
Company 25
Company 26
Company 27
Company 28
Company 29
Company 30
Company 31
Company 32
Company 33
Company 34
Company 35
Company 36
Company 37
Company 38
Company 39
Total
Early Credits
Generated (gal
Bz)
0
0
50,206,864
8,048,513
865,453
48,098,896
89,419,215
35,628,541
6,627,618
0
34,272,947
0
3,173,008
7,072,043
0
48,059,499
5,554,977
410,372
0
0
0
125,647,950
0
27,472,537
19,718,663
146,615,646
14,140,554
32,608,280
69,312,293
3,492,799
0
9,666,313
0
0
16,199,400
53,749,916
0
12,754,685
6,580,236
875,397,218
Early Credits
Demanded (gal
Bz)
0
3,539,225
3,867,817
1,095,769
187,023
41,666,480
69,297,769
59,287,855
975,466
352,305
184,192
555,401
12,199,184
1,579,656
1,115,973
43,424,323
10,157,863
2,167,872
5,752,804
73,894,178
5,505,778
38,587,398
18,800,732
13,929,960
6,747,108
105,384,519
18,007,249
4,440,272
20,330,411
25,103,447
4,792,226
3,053,908
5,214,858
615,214
6,648,814
616,417
5,980,295
0
3,516,739
618,576,501
Surplus /
Shortage
(gal Bz)
0
-3,539,225
46,339,047
6,952,744
678,430
6,432,416
20,121,446
-23,659,314
5,652,152
-352,305
34,088,755
-555,401
-9,026,177
5,492,387
-1,115,973
4,635,176
-4,602,886
-1,757,500
-5,752,804
-73,894,178
-5,505,778
87,060,552
-18,800,732
13,542,577
12,971,555
41,231,126
-3,866,695
28,168,008
48,981,882
-21,610,648
-4,792,226
6,612,405
-5,214,858
-615,214
9,550,586
53,133,499
-5,980,295
12,754,685
3,063,497
256,820,716
% of Net Early
Credit Supply
10.36%
1 .55%
0.15%
1 .44%
4.50%
1 .26%
7.62%
1 .23%
1 .04%
19.46%
3.03%
2.90%
9.21%
6.30%
10.95%
1 .48%
2.13%
1 1 .87%
2.85%
0.68%
100.00%
% of Net Early
Credit
Demand
1 .86%
12.41%
0.18%
0.29%
4.73%
0.59%
2.41%
0.92%
3.02%
38.76%
2.89%
9.86%
2.03%
1 1 .34%
2.51%
2.74%
0.32%
3.14%
100.00%
As shown above in Table 6.5-14, 20 of the 39 companies have the potential to generate
more early credits than they could possibly use during the 2011-2013 early credit generation
period (or 2015-2017 time frame for small refiners). The refinery concentration of early credits
ranges from <1% to 19%. Since there does not appear to be substantial credit market
6-57
-------
Final Regulatory Impact Analysis
concentration, there should be significant potential for the 18 refiners seeking early credits to
postpone future investments to find them. Additionally, 60% of the early credits are anticipated
to be used by the companies which generated them. Because these internal company trades are
the easiest to plan and carry out, there is a high likelihood that the predicted early credit reliance
would actually occur.
6-58
-------
Final Regulatory Impact Analysis
Table 6.5-15. Standard Credit Trading by Company
Company
Company 1
Company 2
Company 3
Company 4
Company 5
Company 6
Company 7
Company 8
Company 9
Company 10
Company 11
Company 12
Company 13
Company 14
Company 15
Company 16
Company 17
Company 18
Company 19
Company 20
Company 21
Company 22
Company 23
Company 24
Company 25
Company 26
Company 27
Company 28
Company 29
Company 30
Company 31
Company 32
Company 33
Company 34
Company 35
Company 36
Company 37
Company 38
Company 39
Total
Std Credits
Generated (gal
Bz/yr)
6,812,377
0
2,005,577
0
0
1,837,570
15,354,274
11,052,495
0
0
0
0
11,785,789
474,273
0
2,796,506
0
0
0
724,306
889,237
6,639,988
56,834
0
0
6,342,861
1,505,238
548,378
12,113,619
10,958,768
147,283
0
0
1,233,715
0
591,320
0
1,718,955
0
95,589,360
Std Credits
Demanded (gal
Bz/yr)
0
1,208,597
1,320,807
374,190
43,765
4,093,155
10,653,361
7,156,828
333,108
120,307
62,899
189,662
0
0
304,052
0
2,491,856
740,299
1,964,504
14,072,746
1,551,206
4,070,613
477,093
4,756,891
2,418,278
18,239,546
121,503
1,598,961
6,008,460
3,811,154
0
1,095,767
1,831,627
0
2,113,754
107,556
2,042,189
0
214,625
95,589,360
Surplus /
Shortage
(gal Bz/yr)
6,812,377
-1,208,597
684,770
-374,190
-43,765
-2,255,585
4,700,913
3,895,667
-333,108
-120,307
-62,899
-189,662
11,785,789
474,273
-304,052
2,796,506
-2,491,856
-740,299
-1,964,504
-13,348,441
-661 ,970
2,569,375
-420,259
-4,756,891
-2,418,278
-11,896,686
1,383,736
-1,050,583
6,105,159
7,147,614
147,283
-1,095,767
-1,831,627
1,233,715
-2,113,754
483,764
-2,042,189
1,718,955
-214,625
0
% of Net Std
Credit Supply
13.12%
1 .32%
9.05%
7.50%
22.69%
0.91%
5.38%
4.95%
2.66%
1 1 .75%
13.76%
0.28%
2.38%
0.93%
3.31%
100.00%
% of Net Std
Credit
Demand
2.33%
0.72%
0.08%
4.34%
0.64%
0.23%
0.12%
0.37%
0.59%
4.80%
1 .43%
3.78%
25.70%
1 .27%
0.81%
9.16%
4.66%
22.90%
2.02%
2.11%
3.53%
4.07%
3.93%
0.41%
100.00%
As shown above in Table 6.5-15, 15 of the 39 companies have the potential to generate
more standard credits than they could use up in a given year. The refinery concentration of
standard credits ranges from <1% to 23%. Since there does not appear to be substantial credit
market concentration, there should be significant potential for the 24 refiners predicted to rely
6-59
-------
Final Regulatory Impact Analysis
upon credits for compliance with the 0.62 vol% standard to find them. Additionally, 35% of the
standard credits are anticipated to be used by the companies which generated them. Because
these internal company trades are the easiest to plan and carry out, there is a high likelihood that
the predicted reliance on standard credits would actually occur.
6.5.4.12 What is the economic value of the ABT program?
In addition to earlier benzene reductions and a more gradual phase-in of the 0.62/1.3
vol% standards (as shown above in Figure 6.5-1), the ABT program results in a more cost-
effective program for the refining industry. Our modeling shows that allowing refiners to
average benzene levels nationwide to meet the 0.62 vol% standard reduces ongoing compliance
costs by about 50% - from 0.51 to 0.27 cents per gallon (as explained in section 9.6.2).
Our modeling further suggest that the early credit program we are finalizing results in the
lowest possible compliance costs during the phase-in period (represented as the area under the
curve in Figure 6.5-2). Without an early credit program, the total cost incurred by the refining
industry from June 1, 2007 through December 31, 2015 is estimated to be $905 million (2003
dollars). With an early credit program, the total amortized capital and operating costs incurred
during the same phase-in period is reduced to $608 million, providing about $300 million in
savings. In the absence of an ABT program altogether, the total cost incurred during the phase-
in period would be $1.7 billion. As a result, the ABT program we are finalizing could save the
refining industry up to $1.1 billion in compliance costs from 2007-2015. For a more detailed
discussion on compliance costs, refer to section 9.6.2.
6-60
-------
Final Regulatory Impact Analysis
450
Figure 6.5-2. Costs Savings Associated with ABT Program
Annualized Compliance Costs vs. Time
400 -
350 -
300 -
250 -
E 200 -
o
o
s
i 150 A
100 -
50 -
0
Jan-07 Jan-08
Jan-0
Jan-10 Jan-11 Jan-12
Time
Jan-13 Jan-14
Jan-15
ABT w/Early Credit Program (MM$) — - ABT w/o Early Credit Program (MM$) No ABT Program (MM$)
The aforementioned program costs and resulting cost savings were estimated based on
compliance costs presented in section 9.6.2 and adjusted back to 2007 to account for the time-
value of money based on a 7% average rate of return. The computed annual compliance costs
for this ABT analysis also consider the projected growth in gasoline production. Gasoline
growth rates from 2004-2012 were estimated by the refinery cost model and future growth rates
were obtained from EIA's AEO 2006. A summary of the semi-annual gasoline inputs and
respective compliance costs is found below in Table 6.5-16.
6-61
-------
Final Regulatory Impact Analysis
Table 6.5-16. AST Program Cost Comparison
Jan-07
Jul-07
Jan-08
Jul-08
Jan-09
Jul-09
Jan-10
Jul-10
Jan-11
Jul-11
Jan-12
Jul-12
Jan-13
Jul-13
Jan-14
Jul-14
Jan-15
Jul-15
Total
Total
Gasoline
Production
(MMbbl)
1,371
1,385
1,385
1,400
1,400
1,416
1,416
1,431
1,431
1,446
1,446
1,466
1,466
1,485
1,485
1,503
1,503
24,436
ABTw/
Early Credit
Program
(MM$)
0
22
21
20
19
18
40
38
36
35
33
46
44
42
41
39
40
74
608
ABT w/o
Early Credit
Program
(MM$)
0
0
0
0
0
0
0
0
109
104
100
95
92
88
84
80
77
74
905
No ABT
Program
(MM$)
0
0
0
0
0
0
0
0
207
197
189
180
174
165
160
152
146
139
1,709
6.6 Feasibility for Recovering Octane
The use of the various benzene control technologies modeled would affect each
refinery's octane in various ways. Rerouting the benzene precursors, adding a benzene
saturation unit, adding a new extraction unit, or revamping an existing one, all would reduce the
octane of gasoline. In the case that the rerouted benzene precursors are sent to an isomerization
unit, there would be a slight increase in octane for the rerouted stream. We evaluated the
average octane impacts of each of these technologies on reformate and on the gasoline pool for
those refineries assumed to be taking action under the benzene control standard. As we
compiled these figures, we observed that there is a large variance in octane impacts for these
technologies. The reason for much of the variance in octane impacts is that many refineries are
estimated to be using benzene precursor rerouting or some benzene extraction today. These
technologies reduce the octane of reformate today. Thus when the reformate treating
technologies are applied the octane loss is smaller than if the refinery is not already using
benzene precursor rerouting or benzene extraction. Since the refineries with large octane
impacts would need to recover all of their octane loss caused by benzene controls, we provide
the maximum octane impacts in addition to the average octane impacts. The average and
maximum octane impacts on gasoline for each benzene control technology are summarized in
Table 6.6-1.
6-62
-------
Final Regulatory Impact Analysis
Table 6.6-1. Octane Impacts for Control Technologies
Expected to Be Used to Meet the Benzene Standards ((R+M)/2)
Average Octane Impacts
Maximum Octane Impacts
Estimated Number of
Benzene Control
Technologies under the
MSAT2 Program
Benzene
Precursor
Rerouting
0.13
0.35
26
Benzene Precursor Rerouting
followed by Isomerization of
Benzene Precursors
0.12
0.34
28
Benzene
Saturation
0.25
0.40
11
Extraction
0.13
0.20
23
We assessed the extent to which various means for recovering octane would have to be
applied to recover the octane reduced by the application of benzene control technologies. The
various octane recovery means we evaluated included revamping certain octane producing units
to produce more of that blendstock, purchasing and blending in high octane blendstocks, and
reducing the production of premium gasoline. As shown in Table 6.6-1, depending on a refiner's
benzene control technology selection, the volume-weighted average octane impact for those
refineries which take steps to reduce their benzene levels averaged 0.13 octane numbers. When
weighted across the entire gasoline pool, this decreases to 0.08 octane numbers. The maximum
octane loss that we observed over all the technologies is a loss of 0.40 octane numbers. We
assess below the ability for differing octane recovery means to recover 0.13, and 0.40 octane
number reductions in the gasoline pool, which represents the average and maximum reduction in
octane numbers.
Alkylate averages about 93 octane numbers and because it is very low in benzene it is an
ideal blendstock for recovering lost octane. Alkylate can be produced within a refinery or it
could be purchased from outside sources. Other blendstocks similar to alkylate are isooctane
and isooctene. Depending on the feedstocks, isooctane and isooctene can have an octane as high
as 100. Along with alkylate, isooctane and isooctene are likely replacements for the phase-out of
MTBE by reusing the MTBE feedstocks. Because isooctane and isooctene will largely be
produced when MTBE is phased out of gasoline and used to explicitly replace MTBE, it will not
be considered in this analysis, although it could still play a marginal role for octane recovery. In
Table 6.6-2 below, we estimate the amount of alkylate which would have to be blended into a
refiner's gasoline pool to recover the various octane losses described above.
Isomerization converts straight chain hydrocarbons into branched chain hydrocarbons
and can also saturate benzene. The isomerization unit increases the octane of light straight run, a
gasoline blendstock which averages an octane number of 70, into a gasoline blendstock with an
average octane number of about 80. While isomerate is not a high octane blendstock and is
generally not sold as one, it is very useful for increasing the octane of a refiner's gasoline while
saturating benzene at the same time. In Table 6.6-2, we estimate the volume of light straight run
that would have to be isomerized to recover the various octane losses described above.
Ethanol's very high octane number of 115 allows making up the octane loss using a
smaller volume than the other blendstocks. Ethanol is an economical source of octane in part
6-63
-------
Final Regulatory Impact Analysis
due to the federal 51 cents per gallon subsidy. It contains a very small amount of benzene
(benzene is present in ethanol only because gasoline is added as a denaturant). The Energy
Policy Act of 2005 (EPAct) recently established a renewable fuels requirement that is expected
to predominantly be met with the addition of ethanol into gasoline. An estimated 4 billion
gallons of ethanol was blended into gasoline nationwide in 2005. By 2012, the EPAct calls for
7.5 billion gallons of renewable fuels to be blended into gasoline and actual ethanol use is
anticipated to be considerably higher due to market forces. The increased use of ethanol
provides a synergistic match with the octane needs of the benzene standard. In Table 6.6-2 we
summarize the volume of ethanol that would have to be blended into gasoline to recover a range
of octane losses.
Finally premium gasoline usually meets either a 91 or 93 octane number rating, while
regular grade gasoline must meet an 87 octane number requirement, although for high altitude
areas the requirement is relaxed to an 85 octane number requirement. The recent increase in
energy prices resulted in a reduced demand for premium grade gasoline. From 2000 to 2005, the
fraction that premium gasoline comprises of total gasoline consumed in the U.S. decreased from
20 percent to 12 percent. Considering that this reduced demand for premium grade gasoline may
continue, we evaluated the extent that the demand in premium grade gasoline would have to
continue to be supplanted by regular grade gasoline to make up for the projected loss of octane
due to benzene reduction in gasoline (this supplanted octane production means that these
refineries producing less premium gasoline would have the potential to increase their octane
production potential by this same amount). This shift in premium gasoline demand to regular
grade demand to recover the range of octane losses is described in Table 6.6-2.
Table 6.6-2. Percent Changes in Gasoline Content for Recovering Octane Shortfalls
(volume percent of gasoline)
Isomerizing Light Straight
Run Naphtha
Blending in Alky late
Blending in Ethanol
Reduced 9 lor 93 ON
Premium Grade Gasoline
0.13 Octane Number Loss
1
2
0.5
3
0.40 Octane Number Loss
4
7
2
10
Isomerizing a refinery's gasoline blendstocks is effective because in addition to
addressing octane, it can also treat the benzene normally found in gasoline. It would not be an
available technology in those refineries that are already fully using isomerization. The refinery -
by-refmery cost model estimates that light straight run feedstock to the isomerization unit
typically comprises about 7 percent of each refinery's gasoline pool so it potentially could meet
the octane needs of even the greatest octane needs caused by this rulemaking if isomerization is
not already being used. Even those refineries that will be isomerizing all their light straight run
prior to the implementation of the benzene standard could reroute the six carbon hydrocarbons
around the reformer and send this stream to an isomerization unit to recover at least a part of the
octane loss associated with the benzene reduction. An average octane loss of 0.14 octane
numbers and the refinery-specific maximum 0.40 octane numbers would require an additional 1
6-64
-------
Final Regulatory Impact Analysis
volume percent and 4 percent of the light straight run currently blended into gasoline to be
isomerized, respectively.
Alkylate's moderate octane value requires a relatively large volume to make up for the
octane losses associated with the removal of benzene. At the estimated highest octane loss, the
volume of alkylate would have to nearly double relative to the 12 percent typically blended into
gasoline in 2003. Additional alkylate may be able to be produced by increasing the severity of
the FCC unit, if there is capacity to do so, that would increase the production of feedstocks to the
alkylate unit. Alkylate's very desirable gasoline blending properties, which is high octane, low
RVP and sulfur and very low benzene, encourages its use. To replace an average octane loss of
0.14 octane numbers and the refinery-specific maximum 0.40 octane numbers, a refinery would
need to produce or purchase and blend in an additional 2 volume percent and 7 percent of
alkylate into their gasoline, respectively.
Ethanol is very high in octane which allows the recovery of lost octane caused by the
treating of benzene with a smaller volume than the other octane recovery means considered. The
additional volume of ethanol expected to be blended into gasoline under EPAct makes it a likely
candidate for an octane replacement for a benzene standard. If all of EPAct's renewable
requirement is met with the blending of ethanol into gasoline, the 31A> additional billion gallons
of ethanol that would be blended into gasoline between today and 2012 would increase ethanol's
content in gasoline from 2.8 to 4.7 volume percent, a 1.9 volume percent increase in all U.S.
gasoline. To replace an average octane loss of 0.14 octane numbers and the refinery-specific
maximum 0.40 octane numbers, a refiner would need to blend in an additional 0.5 volume
percent and 2 percent of ethanol in their gasoline, respectively. This provides far more than the
octane number increase needed to recover the average octane loss of refineries that reduce their
benzene levels to comply with the benzene standard, and even more ethanol use is expected.
The phasing-in, under the ABT program, of the benzene standard and its associated octane loss
would coincide with the period that EP Act's renewable requirement phases in and ethanol's use
expands.
The decreasing demand for premium grade gasoline would provide another means for
refiners to recover the octane lost from benzene control. The demand for premium has been
supplanted by a higher demand for lower octane regular, freeing up octane producing potential in
refineries. Between 2000 and 2005, premium gasoline demand decreased by 8 volume percent.
This decrease represents nearly a 0.4 octane number decrease in the gasoline pool. To replace an
average octane loss of 0.14 octane numbers and the maximum refinery-specific 0.40 octane
numbers, a refiner would need to have shifted 3 volume percent and 10 percent of their gasoline
production from premium grade to regular grade, respectively. This indicates that there may be
more than enough excess octane producing potential already to satisfy a loss in octane that
would be expected to begin in 2007 under the benzene control program.
6.7 Will the Benzene Standard Result in Any New Challenges to the Fuel
Distribution System or End-Users?
There are two potential concerns regarding whether the implementation of the benzene
standards would adversely impact the fuel distribution system and end-users of gasoline. The
6-65
-------
Final Regulatory Impact Analysis
first potential concern relates to whether additional product segregation would be needed. The
small refiner and ABT provisions in today's notice would result in some refiners producing
gasoline with benzene levels below the standard while other refiners would meet the standard
through the use of credits or under the small refiner provisions. Thus, gasoline benzene levels
would vary on a refinery by refinery basis, much as they always have. Today's proposal would
not result in the need for the segregation of additional grades of gasoline in the distribution
system. Consequently, we do not the MSAT2 program to require construction of new storage
tanks in the fuel distribution system or result in other facility or procedural changes to the
gasoline distribution system.
The second potential concern relates to whether the gasoline property changes that might
result from the benzene standard could adversely impact the equipment in the fuel distribution
system or end-user vehicles. We are aware that a stringent benzene standard is associated with a
potential need to make up for some loss of octane. If octane replacement is warranted, we
anticipate that refiners accomplish this by blending ethanol or other suitable octane-rich
blendstocks, or in some cases by increasing the production of other octane rich refinery streams.
Consequently, we expect that there would be no net change in gasoline octane levels as a result
of the benzene standards, and no impact on equipment in the distribution system.
We are aware of no other gasoline property changes that might be of potential concern to
the distribution system.
6.8 Impacts on the Engineering and Construction Industry
An important aspect of the feasibility of a fuel program is the ability of the refining
industry to design and construct any new equipment required to meet the new fuel quality
standard. In this section we assess the impact of the gasoline benzene program on demand for
engineering design and construction personnel. We will focus on three types of workers that are
needed to design and build new equipment involved in benzene reduction: front-end designers,
detailed designers, and construction workers. This analysis builds on those done for the 2007
heavy-duty highway and nonroad diesel sulfur rulemakings, and will include the impacts of these
programs on the industry's ability to comply with the new benzene standard. We compare the
overall need for these workers to estimates of total employment in these trades. In general, it
would also be useful to expand this assessment to specific types of construction workers which
might be in especially high demand, such as pipe-fitters and welders. However, we are not
aware of appropriate estimates of the number of people currently employed in these job
categories. Thus, it is not possible to determine how implementing these programs might stress
the number of personnel needed in these types of specific job categories.
To carry out this analysis we first estimated the level of design and construction
resources required for new and revamped benzene reduction equipment. We next projected the
number of these units which would be needed under the gasoline benzene program and how the
projects might be spread out over time. We then developed a schedule for when the various
resources would be needed throughout each project. Finally, we projected the level of design
and construction resources needed in each month and year from 2000 through 2015 and
compared this to the number of people employed in each job category.
6-66
-------
Final Regulatory Impact Analysis
6.8.1 Design and Construction Resources Related to Benzene Reduction Equipment
The calculation of job-hours necessary to design and build individual pieces of
equipment and the number of pieces of equipment per project mirrors the analysis done for the
nonroad diesel rulemaking promulgated in 2004. The methodology was originally based on a
technical paper authored by Moncrief and Ragsdale28 in support of a National Petroleum Council
study of gasoline and diesel fuel desulfurization and other potential fuel quality changes. Unit
types we considered for construction to meet the new standard are light naphtha splitters (LNS)
for routing benzene pre-cursors around the reformer unit, benzene saturation units, and benzene
extraction units.0 We assumed that benzene saturation equipment projects were of the same
scale as described for a hydrotreater project, while LNS units were 50% smaller projects and
benzene extraction units were conservatively 50% larger projects. Consistent with Moncrief and
Ragsdale, revamps were assumed to use fewer resources than a new unit. All benzene saturation
and LNS units are expected to be new installations, while work on benzene extraction units is
split between new and revamped units. Estimated resource needs for these projects are
summarized in Table 6.8-1.
Table 6.8-1. Design and construction factors for benzene reduction equipment.
Project type
Number of pieces of equipment
LNS
New
30
Revamp*
15
Saturation
New
60
Extraction
New
90
Revamp*
30
Job-hours per piece of equipment
Front-end design
Detailed design
Direct and indirect construction
300
1200
9150
150
600
4575
300
1200
9150
300
1200
9150
150
600
4575
*Equipment revamps were assumed to use half the usual job-hours per piece of equipment
6.8.2 Number and Timing of Benzene Reduction Units
The next step was to estimate the types of equipment modifications necessary to meet the
benzene standard. This was a complex task due to the ABT program, which allows refiners the
flexibility to balance their own benzene reductions with purchase of credits from reductions
elsewhere, resulting in different types of equipment projects being chosen depending on what is
most economical for a particular refinery. Detailed analysis of equipment choices was carried
out in our assessment of the costs of the fuel program. H Those results provide inputs for this
G These technologies are discussed in detail in Section 6.3.2 of this RIA.
H Equipment choice and project timing is covered in more detail in discussions of the averaging, banking
and trading analyses (see Section 6.5 of this RIA).
6-67
-------
Final Regulatory Impact Analysis
analysis, shown in Table 6.8-2.
Once equipment types were tabulated, timing of projects had to be considered. Worst-
case scenarios of unit startup dates of January 1, 2011 are unlikely for a number of reasons.
First, the early credit program is expected to encourage refiners planning relatively simple
process modifications, such as revamping or de-bottlenecking of equipment for light naphtha
splitting, to take these actions shortly after fmalization of the standards. Furthermore, given the
flexibility of ABT and the different approaches available for benzene reduction, projects will
differ in complexity and scope. Expected project timing, assuming some early compliance, is
summarized in Table 6.8-2.1
Table 6.8-2. Number and timing of startup for benzene reduction projects.
Start-up date
LNS: New
Saturation: New
Extraction: New
Revamp
2010
31
0
0
7
2012
0
7
0
0
2015 (Jan)
2
4
0
0
2015 (May)
8
5
3
6
6.8.3 Timing of Projects Starting Up in the Same Year
Even if refiners all desired to complete their project on the same date, their projects
would begin over a range of months. Thus, two projects scheduled to start up at exactly the
same time are not likely to proceed through each step of the design and construction process at
the same time. In addition, it is reasonable to assume design and construction of units will be
spread out over the calendar year. We assumed 25 percent of the units would initiate design and
thus, startup, each quarter leading up to the date upon which they had to be operational.
6.8.4 Timing of Design and Construction Resources Within a Project
The next step in this analysis was to estimate how the engineering and construction
resources are spread out during a project. For the nonroad diesel rulemaking we developed a
distribution of each type of resource across the duration of a project for the 2007 heavy-duty
highway and nonroad diesel sulfur programs, and this methodology was extended for this
rulemaking. The fractions of total hours expended each month were derived as follows.
6-68
-------
Final Regulatory Impact Analysis
Per Moncrief and Ragsdale, front end design typically takes six months to complete. If
25 percent of the refineries scheduled to start up in a given year start their projects every quarter,
each subsequent group of the refineries starts when the previous group is halfway through their
front end design. Overall, front end design for the four groups covers a period of 15 months, or
six months for the first group plus six months for each of the three subsequent groups. In
spreading this work out over the 15 months, we assumed that the total engineering effort would
be roughly equal over the middle nine months. The effort during the first and last three month
periods would be roughly two-thirds of that during the peak middle months. The same process
was applied to the other two job categories.J The distribution of resources is summarized in
Tables 6.8-3 and 6.8-4.
In the case of projects to be completed for 2010, front end design schedules were
compressed to half. This seemed reasonable, given that these early projects are expected to
either be installation of LNS units or revamps of other units, which do not require extensive
design work.
Table 6.8-3. Duration of pro.
Duration per project
Total duration for
projects starting up in
a given calendar year
Front-end
design
(2010)
3 months
7 months
Front-end
design
(2012+)
6 months
15 months
ect phases.
Detailed
engineering
(All years)
11 months
20 months
Construction
(All years)
14 months
23 months
1 The reader is referred to the Final Regulatory Impact Analyses for the 2007 Heavy Duty Highway Diesel
rulemaking (EPA420-R-00-026, Chapter IV Section B. 1) and the Nonroad Diesel rulemaking (EPA420-R-04-007,
Chapter 5.7) for more detailed description of the methodology used.
6-69
-------
Final Regulatory Impact Analysis
Table 6.8-4. Distribution of personnel requirements throughout project.
Month
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Fraction of total hours expended by month for completion years shown
Front-end
design
(2010)
0.100
0.100
0.200
0.200
0.200
0.100
0.100
Front-end
design
(2012+)
0.050
0.050
0.050
0.078
0.078
0.078
0.078
0.078
0.078
0.078
0.078
0.078
0.050
0.050
0.050
Detailed
engineering
(All years)
0.020
0.030
0.040
0.040
0.040
0.050
0.050
0.060
0.065
0.075
0.075
0.075
0.060
0.060
0.050
0.050
0.040
0.040
0.030
0.020
Construction
(All years)
0.030
0.030
0.030
0.040
0.040
0.040
0.040
0.050
0.050
0.055
0.055
0.060
0.060
0.055
0.055
0.050
0.050
0.040
0.040
0.040
0.030
0.030
0.030
6.8.5 Projected Levels of Design and Construction Resources
We calculated the number of workers in each of the three categories required in each
month by applying the distributions of the various resources per project (Table 6.8-4) to the
number of new and revamped units projected to start up in each calendar year (Table 6.8-2) and
the number of person-hours required per project (Table 6.8-1). We converted hours of work into
person-years by assuming that personnel were able to actively work 1877 hours per year, or at 90
percent of capacity assuming a 40-hour work week. We then determined the maximum number
of personnel needed in any specific month over the years 2007-2015 for each job category both
with and without the new benzene control program. The results are shown in Table 6.8-5.
In addition to total personnel required, the corresponding percentage of the relevant U.S.
workforce is also shown. These percentages were based on estimates of recently available U.S.
employment levels for the three job categories given in Moncrief and Ragsdale: 1920 front end
design personnel, 9585 detailed engineering personnel, and roughly 160,000 construction
workers. The figure for construction workers was given as 80,000 specifically for the Gulf
Coast, where it is estimated that half of refining projects will take place. Based on this, we
estimated the available pool of construction personnel nationwide at twice that figure, or
6-70
-------
Final Regulatory Impact Analysis
160,000, under the assumption that construction personnel would be distributed proportional to
refining capacity on a geographical basis.
Table 6.8-5. Maximum monthly personnel demand.
Program
Tier 2 gasoline sulfur,
Highway and nonroad diesel
sulfur
Gasoline benzene
Parameter
Max. number
of workers
Current
workforce*
Max. number
of workers
Current
workforce*
Front-end design
758
(Mar '03)
40%
763
(Apr '07)
40%
Detailed
Engineering
2,720
(Mar '04)
28%
2,720
(Mar '04)
28%
Construction
17,646
(November '04)
11%
17,646
(November '04)
11%
*Based on recent U.S. employment in trades listed. Year and month of maximum personnel demand is
shown in parentheses.
Shown in Table 6.8-5, the gasoline benzene program has a projected maximum monthly
requirement for front end design personnel equivalent to the level seen in 2003 for previous
programs. Peaks in the other two job categories' monthly personnel demand projected for this
program remain below levels previously seen for prior programs. Based on this analysis,
projected demand levels represent less than half of the estimated front-end design workforce, and
less than one third of the estimated workforce in the detailed design and construction trades
Figures 6.8-1 through 6.8-3 illustrate that average monthly personnel demand trends for
the gasoline benzene program, based on annual workload, generally occur after significant peaks
related to other programs have passed. Given these results, we believe that the E&C industry is
capable of supplying the refining industry with the personnel necessary to comply with the
gasoline benzene program.
6-71
-------
Final Regulatory Impact Analysis
Figure 6.8-1. Projected Average Monthly Front-End Engineering Personnel Demand
Trends 2000-2015.
c
(0
(I)
Q
"3
c
c
o
S2
o
0)
800
600
Front-End
Engineering
I
Existing Programs
With Gasoline Benzene Program
Year
6-72
-------
Final Regulatory Impact Analysis
Figure 6.8-2. Projected Average Monthly Detailed Engineering Personnel Demand Trends
2000-2015.
C
n
I
0)
C
o
S2
0)
0.
>
0)
G)
5
0)
3,000
2,500
1,500
,000
500
Detailed Engineering
,
Existing Programs
With Gasoline Benzene Program
Year
VA
,
6-73
-------
Final Regulatory Impact Analysis
Figure 6.8-3. Projected Average Monthly Construction Personnel Demand Trends
2000-2015.
-o 20,000
Ł
Q
o 15,000
°- 10,000
^^
C
O
-------
Final Regulatory Impact Analysis
Table 6.9-1. Lead Time Required Between Promulgation of the Final Rule and
Implementation of the Gasoline Sulfur Standard (years)
Scoping Studies
Process Design
Permitting
Detailed Engineering
Field Construction
Start-up/Shakedown
Naphtha/Gasoline Hydrotreating
Time for Individual Step
0.5-1.0*
0.5
0.25-1.0
0.5-0.75
0.75-1.0
0.25
Cumulative Time
0.5
1.0
1.25-2.0
1.5-2.25
2.0-3.0
2.25-3.25
* Can begin before FRM
Table 6.9-1 shows that 2 1A to 3 1A years is estimated to be needed to install a naphtha
hydrotreater. The naphtha hydrotreater investments are significant, costing refiners tens of
millions of dollars per refinery and requiring the installation of many pieces of equipment. Some
of the equipment needed for a FCC naphtha hydrotreater includes high pressure reactors and
hydrogen compressors, that generally require a long purchase lead time, as well as heat
exchangers and a furnace. The associated octane loss and hydrogen use could also require the
installation of additional hydrogen and octane production capacity.
The benzene control technologies projected to be installed to reduce gasoline benzene
levels are typically much less involved and can therefore be installed in the same or less time
than the FCC naphtha hydrotreaters. The rerouting of benzene precursors requires that the
naphtha splitter distillation column be revamped to provide a better split between the six and
seven carbon hydrocarbons to allow the bypassing of the six carbon hydrocarbons around the
reformer. In some cases this revamping only requires the addition of some trays or packing in
the existing naphtha splitter. However, in other cases, the revamp would require the complete
replacement of the existing naphtha splitter. These changed can take up to 1 to 2 years. If the
refinery has an isomerization unit, it could further reduce its gasoline benzene level by feeding
the rerouted benzene precursor stream to this unit. This additional step can occur with no
additional investment by the refinery and therefore takes no appreciable amount of time to
implement.
Additional benzene reduction is projected to occur by revamping existing extraction
units. The revamp can occur by further reducing the benzene level of the refinery with the
extraction unit, or by treating a benzene rich reformate stream of a neighboring refinery. The
revamp could occur in one or more places, including the reformate splitter to capture more of its
own benzene, expanding the extraction unit, or expanding the distillation towers after the
extraction unit. Each of these possible revamp opportunities are similar in nature to those for
revamping a light straight run splitter. Thus they can also occur in 1 to 2 years.
The other two means for benzene control are grassroots extraction and benzene saturation
units. As grassroots units they both require the installation of numerous pieces of equipment,
including furnaces, heat exchangers, the distillation towers, and extraction and saturation
reactors, and instrumentation. Grassroots extraction units also require the installation of benzene
storage vessels and loading equipment. The design and construction of all these pieces of
6-75
-------
Final Regulatory Impact Analysis
equipment is why grassroots benzene saturation and extraction units are expected to need a lead
time more in line with naphtha hydrotreaters, which is 2 Va to 3 l/2 years.
Refiners would also need to recover lost octane. The octane can be recovered by
purchasing high octane blendstocks, such as alkylate, ethanol or isooctane, or by revamping
existing octane producing units or installing new units, including alkylate and isomerization
units. Revamping existing alkylate or isomerization units is expected to require 1 to 2 years to
complete. Installing new octane generating units would likely take no more time than the 2 1A to
3 1A years estimated for grassroots benzene saturation and extraction units.
Some revamped or new capital may be needed for providing the hydrogen needed to
saturate the benzene in isomerization and saturation units, or to make up hydrogen lost by
routing the benzene precursors around the reformer. For most refineries we expect that they can
use excess hydrogen production capacity or could purchase the needed hydrogen from a third
party provider. A few refineries will have to modify their hydrogen plant which would only take
1-2 years. Should the refinery be in the position to have to install a new hydrogen plant, it
could do so in no more time than the 2 1A to 3 1A years estimated for grassroots benzene
saturation and extraction units.
The 2!/4 to 3!/4 years identified above for installing the benzene control technologies, and
potentially for installing octane recovery and hydrogen production facilities, could allow starting
the program after 3 years, in 2010, instead of four years. However, in our assessment of the
impacts of the benzene control program on the engineering and construction industry, we
identified that an earlier start date would overlap the engineering and construction (E&C)
demands of this program with other fuel control programs. The last of the investments being
made for the Tier 2 gasoline sulfur control program are occurring in 2010. The 15 ppm sulfur
standard mandated by the Nonroad Diesel Fuel program applies to nonroad diesel fuel in 2010
and to locomotive and marine diesel fuel in 2012. Finally, the last of the 15 ppm highway diesel
fuel sulfur standard applies in 2010. Implementing this benzene control program in 2010 would
result in an overlap of the E&C demands with the various other fuel programs phasing in that
year.
Phasing in this benzene fuel control program in 2011 instead would slightly stagger the
start year of this benzene fuel program with the start years for the Tier 2, Nonroad and Highway
Diesel Fuel sulfur programs. Staggering the start dates may also help refiners seeking funding to
make the capital investments.
6.10 Will the Benzene Standards Be More Protective Than Current
Programs?
Three fuels programs (RFG, Anti-dumping and MSAT1) currently contain direct controls
on the toxics emissions performance of gasoline.k The RFG program, promulgated in 1994,
contains a fuel benzene standard which requires a refinery's or importer's RFG to average no
k Other gasoline fuel controls, such as sulfur, RVP or VOC performance standards, indirectly control toxics
performance by reducing overall emissions of VOCs.
6-76
-------
Final Regulatory Impact Analysis
greater than 0.95 vol% benzene annually, with a per-gallon cap of 1.3 vol%.29'' Each refinery's
or importer's RFG must also achieve at least a 21.5% reduction in total toxics emissions (as
determined by the Complex Model) compared to 1990 baseline gasoline. The Anti-dumping
regulations require that a refinery's or importer's CG produce no more exhaust toxics emissions
(also using the Complex Model) than its 1990 gasoline. 30'31 This was intended to keep refiners
from complying with RFG by simply shifting fuel components responsible for elevated toxics
emissions into CG.
The MS ATI program, promulgated in 2001, was overlaid onto the RFG and Anti-
dumping programs.32 It was not designed to further reduce MSAT emissions, but to lock in
overcompliance on toxics performance that was being achieved by that time in RFG and CG
under the RFG and Anti-dumping programs. The MS ATI rule required the annual average
toxics performance of a refinery's or importer's gasoline to be at least as clean as the average
performance of its gasoline during the three-year baseline period 1998-2000. Compliance with
MS ATI is determined separately for each refinery's or importer's RFG and CG.
The new benzene content standard will apply to all of a refinery's or importer's gasoline,
that is, the total of its RFG and CG production or imports. This level of benzene control far
exceeds RFG's statutory standard, and puts in place a benzene content standard for CG for the
first time. An analysis was carried out to determine how the overall toxics performance of
gasoline vehicle emissions under the new standard compares to performance under the relevant
pre-existing standards.
6.10.1 Modeling Approach
Two levels of analysis were carried out to address this question. The first was an
examination of the relationship between toxics performance of individual gasoline refiners (or
other producers) under the new benzene program and their historical or required performance.
This analysis was quantitative where changes in fuel parameters were known or could be
projected with some confidence, followed by further qualitative examination where changes in
other fuel parameters (such as oxygenate blending) could only be projected direct!onally.
We also undertook a second level of analysis with the aim of producing quantitative
results more likely to represent reality at the time of phase-in of the new standard, accounting for
the complexities of oxygenate changes as well as sulfur reductions, projected benzene
reductions, and changes in the mix of new technology vehicles in future year fleets. This
analysis was done on a regional basis, which allowed aggregation of fuel parameters, increasing
our confidence in the projection of future trends.
The refinery-by-refinery analysis of toxics emissions performance was conducted using
the Complex Model (the same model used for determining compliance with these programs).
We used 2004 exhaust toxics performance for CG and 2004 total toxics performance for RFG as
benchmarks, which are at least as stringent as the relevant toxics performance baselines. We
applied changes to each refiner's fuel parameters for the new benzene standard and the gasoline
1 Refiners also have the option of meeting a per gallon limit of 1.0 vol%.
6-77
-------
Final Regulatory Impact Analysis
sulfur standard (30 ppm average, 80 ppm max, fully implemented in 2006). The results indicate
that all refineries maintained or reduced their emissions of toxics over 2004 due to both sulfur
and benzene reductions. Large reductions in sulfur have occurred in almost all refineries under
the gasoline sulfur program. We do not expect backsliding in sulfur levels by the few refiners
previously below 30 ppm because they had been producing ultra-low sulfur gasoline for reasons
related to refinery configuration. We project large reductions in CG benzene levels will also
occur along with modest reductions in RFG benzene levels. Because of its petrochemical value
and the credit market, we do not expect any refiners to increase benzene content in their
gasoline.
In addition, we expect significant changes in oxygenate blending over the next several
years, but these are very difficult predict on a refinery-by-refinery basis. Regardless of how
individual refineries choose to blend oxygenates in the future, we believe their gasoline will
continue to comply with baseline requirements. This is because all RFG is currently
overcomplying with the statutory requirement of 21.5% annual average toxics reductions by a
significant margin. Similarly, most CG is overcomplying with its 1990 baselines by a significant
margin. Furthermore, we believe most refiners currently blending oxygenates will continue to
do so at the same or greater level into the future.
The second level of analysis employed MOBILE6.2 to estimate emission rates (mg/mi)
for air toxics under a number of existing and projected fuel control cases, and is the subject of
the rest of this section. This modeling included evaluation of toxics emissions on a regional
level for baseline and future year scenarios. Five regions of the country were examined, divided
according to PADDs (defined in 40 CFR 80.41), using PADD-aggregate fuel parameters. In
looking ahead to the phase-in period of the gasoline benzene standard, this work accounted for
significant changes in gasoline properties since the MS ATI baseline period. The Tier 2
program, currently phasing in, brings together very low gasoline sulfur standards and stringent
vehicle standards that will reduce emissions significantly. In addition, over the next several
years, fuel qualities will change in many regions of the country as ethanol blending increases as
described in the Renewable Fuels Standard rulemaking. 33
6.10.1.1 Choice of Analysis Cases and Data Sources
The Energy Policy Act of 2005 requires that toxics emissions baselines for RFG be
adjusted to reflect 2001-2002 performance, which would make them slightly more stringent than
the 1998-2000 baselines used in the MSAT1 program.34 However, as provided for in the Act,
this action becomes unnecessary and can be avoided if this benzene control program can be
shown to bring greater reductions of toxics emissions from vehicles in RFG areas than would be
achieved by this baseline adjustment. Therefore, in addition to comparing the gasoline benzene
standard to the current MS ATI program, we also compared it to standards as they would change
under EPAct. In addition, we compared projected emissions in 2011 with and without the
MSAT2 standards.
For this analysis, MOBILE inputs included fuel parameters and the fleet year being
examined, as well as an average daily temperature profile for each region and season. Separate
aggregate fuel parameter sets were generated for each PADD for CG and RFG, summer and
6-78
-------
Final Regulatory Impact Analysis
winter. Model outputs for various compounds and emission types were then aggregated into
annualized mg/mi total toxics emissions for an average vehicle in each PADD by RFG and CG.
An MS ATI baseline case was run using 1998-2000 volume-weighted data aggregated
from batch reports submitted to EPA by refiners under the reporting requirements of existing
programs. A second set of baseline figures were generated using 2001-2002 batch reports for
RFG, based on the requirements of EPAct. It should be noted that the baseline toxics emissions
figures generated in this analysis are different from those used to determine compliance with the
MSAT1 program. MSAT1 compliance baseline figures are generated by the Complex Model,
which includes emissions of POM but not acrolein, and does not account for effects of changes
in vehicle technology or fleet mix.
Future cases chosen for comparison included year 2011 without the MSAT2 program,
under the MSAT2 fuel program only, and under both the MSAT2 fuel and vehicle programs. An
additional case was run for year 2025 including effects of both vehicle and fuel standards. A
summary of the cases and datasets examined is given in Table 6.10-1. The future year 2011 was
chosen because of the effective date of this standard.™ The future year 2025 was chosen based
on a significant phase-in of vehicles (> 80% of the fleet) produced under the new vehicle
standard. Fuel parameter data for the 2011 and 2025 cases were generated by taking 2004 data
and making adjustments to account for changes expected due to regulatory programs and
projected oxygenate blending trends.
m This analysis assumes a simplified phase-in of the standard. Details of projected phase-in period are
covered in Section 6.5 of this RIA.
6-79
-------
Final Regulatory Impact Analysis
Table 6.10-1. Choice of Analysis Cases and Data Sources
Case
MS ATI Baseline
MSAT1 Baseline as
Modified by EP Act
EPAct Baseline,
2011
MS AT2, 20 11 (Fuel
standard only)
MS AT2, 20 11 (Fuel
+ vehicle standards)
MSAT2, 2025 (Fuel
+ vehicle standards)
RFG fuel parameter dataset
1998-2000
2001-2002
2004 plus adjustments for:
- Low-sulfur gasoline
- Increased ethanol blending
- MTBE & other ethers phased out
2004 plus adjustments for:
- Low-sulfur gasoline
- Increased ethanol blending
- MTBE & other ethers phased out
- 0.62% benzene std
2004 plus adjustments for:
- Low-sulfur gasoline
- Increased ethanol blending
- MTBE & other ethers phased out
-0.62% benzene std
- 20°F vehicle HC std
2004 plus adjustments for:
- Low-sulfur gasoline
- Increased ethanol blending
- MTBE & other ethers phased out
-0.62% benzene std
- 20°F vehicle HC std
CG fuel parameter dataset
1998-2000
1998-2000
2004 plus adjustments for:
- Low-sulfur gasoline
- Increased ethanol blending
- MTBE & other ethers phased out
2004 plus adjustments for:
- Low-sulfur gasoline
- Increased ethanol blending
- MTBE & other ethers phased out
-0.62% benzene std
2004 plus adjustments for:
- Low-sulfur gasoline
- Increased ethanol blending
- MTBE & other ethers phased out
-0.62% benzene std
- 20°F vehicle HC std
2004 plus adjustments for:
- Low-sulfur gasoline
- Increased ethanol blending
- MTBE & other ethers phased out
-0.62% benzene std
- 20°F vehicle HC std
Fleet year
2002
2002
2011
2011
2011
2025
6.10.1.2
Adjustment of Fuel Parameters for Future Years
In order to carry out the analysis as realistically as possible, adjustments were applied to
fuel parameters when running future year cases. Starting from 2004 gasoline data (the most
recent available at the time of the analysis), the changes accounted for in this analysis were
sulfur reduction related to the gasoline sulfur program, increased ethanol blending to 9.6 billion
gallons per year nationwide as described in the Renewable Fuels Standard rulemaking (9.6 Max-
RFG case), phase-out of MTBE and other ethers, and reduction of gasoline benzene levels under
the new program. Some of these changes are expected to have predictable secondary effects on
non-target fuel parameters that were also considered.
Reduction of Gasoline Sulfur
Under the recent gasoline sulfur rulemaking, as of January 1, 2006 all gasoline (except
gasoline produced by small refiners and those covered by the geographic phase-in provisions) is
required to meet an average standard of 30 ppm sulfur (80 ppm per-gallon cap). Therefore,
6-80
-------
Final Regulatory Impact Analysis
MOBILE inputs for gasoline sulfur levels were set to 30 ppm average and 80 ppm max for all
PADDs and seasons. No adjustments to other fuel parameters were made as a result of sulfur
reductions.
Increased Blending ofEthanol
Under the Energy Policy Act of 2005, EPA was charged with putting in place a
regulatory system to ensure that renewable fuels are used in the national fuel pool at an
increasing rate through the year 2012, as well as evaluating the air quality, energy supply, and
economic impacts of these changes.35 Part of this work involved projecting corresponding
changes to gasoline qualities, the results of which were also used in this analysis. This analysis
is described in detail in Chapter 2 of the draft RIA of the proposed Renewable Fuels Standard
(RFS); the major points are summarized below.36
This analysis used the ethanol blending volumes projected for the scenario of 9.6 billion
gallons per year in 2012 with maximum use in RFG, as developed in Section 2.1.4.6 of the RFS
draft RIA (in this analysis we did not attempt to adjust ethanol blending for any difference
between 2011, the fleet year of the analysis, and 2012). Differences in market share of ethanol
and MTBE blending between 2004 and 2012 were used to adjust 2004 fuel parameters. Summer
and winter blending ratios were assumed to be equal, and market shares for 2012 were also used
in 2025. These figures are shown here in Tables 6.10-2 and 6.10-3.
Table 6.10-2. Projected Changes in Ethanol Use in Gasoline (% volume).
2004 2012
PADD CG RFG CG RFG
I
II
III
IV
V (ex/CA)
ALL
0.0%
3.2%
0.3%
1.8%
2.6%
3.5%
10.0%
0.5%
-
-
2.3%
9.7%
0.5%
6.9%
5.1%
10.0%
10.0%
10.0%
-
-
Table 6.10-3. Projected Changes in MTBE Use in Gasoline (% volume).
2004
PADD
I
II
III
IV
V (ex/CA)
ALL
CG
0.0%
0.0%
0.0%
0.0%
0.2%
RFG
7.2%
0.0%
10.4%
-
-
2012
CG
0.0%
0.0%
0.0%
0.0%
0.0%
RFG
0.0%
0.0%
0.0%
-
-
The secondary fuel parameters adjusted were aromatics, olefms, E200, E300 and vapor
pressure (MTBE, ethanol, sulfur, and benzene content were already being changed as a direct
result of regulatory or other actions). The impact on each of these parameters was calculated
6-81
-------
Final Regulatory Impact Analysis
separately for each PADD by CG and RFG, according to the factors in Table 6.10-4. In PADDs
where MTBE use was reduced, the MTBE factors shown were applied in a negative direction,
meanwhile the ethanol factors were applied in a positive direction where ethanol use was
increasing. These factors were developed as part of the RFS proposal.
Table 6.10-4. Fuel Parameter Adjustment Factors for Oxygenates.
E200 (%)
E300 (%)
Aromatics (Vol%)
Olefins (Vol%)
RVP (psi)
Conventional Gasoline
Ethanol
MTBE
+1.0
+0.52
+0.24
+0.17
-0.5
-0.59
-0.16
0
+0.1
0
Reformulated Gasoline
Ethanol
MTBE
0
0.1
0
0.1
0
0
0
0
0
0
Phase-out of Ether Blending
Use of MTBE and other ethers has been outlawed by several states, including California,
New York, and Connecticut. All refiners we have spoken with are phasing out production and
blending of these at their facilities regardless of such prohibitions, mainly for reasons of
potential environmental liability, uncertainties of future markets, and related costs. Furthermore,
with the renewable fuels mandate in EPAct, essentially all gasoline oxygenate use has shifted to
ethanol. Given these facts, ether content was assumed to be zero in all regions for future year
cases.
Reduction of Benzene Content
The final step of fuel quality adjustment for future year cases was to incorporate the
gasoline benzene standard. Modeling done to evaluate the cost of the program resulted in
projected benzene levels for each PADD. These figures are given in Section 6.5.4 above, and
were used as the final benzene levels as summarized in Table 6.10-8 below. Analysis of trends
in fuel property data suggested that this reduction of benzene content is expected to be
accompanied by an equal reduction in total aromatics content. Therefore, both benzene and
aromatics levels were adjusted in this final step.
6.10.1.3
Conversion of Production Properties to In-Use Properties
To analyze the impacts of gasoline quality on vehicle emissions on a large scale, it is
important to know the properties of the gasoline consumed in a given state or region of the
country as opposed to the gasoline produced there. Some information on point-of-use quality is
available through gasoline quality surveys conducted by the Alliance of Automobile
Manufacturers and TRW, but these surveys are too limited to use for a detailed national analysis.
Very comprehensive data on gasoline production is available through the reporting requirements
of other regulatory programs, whereby refiners report gasoline batch volumes and quality
6-82
-------
Final Regulatory Impact Analysis
information to EPA. However, following production, gasoline is often shipped long distances.
Due to the complex nature of the gasoline distribution system and the intentional fungibility of
the product, there is no straightforward way to track the vast majority of gasoline after it leaves
the refinery. Thus, there is no accurate way to relate gasoline production properties to
consumption properties for a state or region of the country.
We assessed whether to attempt to use the very limited survey data or work through the
complications of adapting production data for this purpose, and eventually decided that
production data would lead to a better overall estimate of fuel quality estimates for broad regions
of the country. We estimated the qualities for gasoline as consumed in each of the five PADDs,
based on qualities of gasoline produced in each PADD and its movement to other PADDs. EIA
collects and reports to the public a variety of data on gasoline production, movement, and
consumption. Included in their analyses are quantities of gasoline moved between PADDs,
broken down by RFG, CG, and oxygenated CG. By linking this information with gasoline
volume and property information from EPA's database, we developed weighted average fuel
parameters for gasoline as consumed in each PADD.
Generally speaking, we weighted together the properties of gasoline produced in a PADD
with those of gasoline transported into that PADD. Using data from 2004 refiner compliance
reports submitted to EPA, gasoline property figures were aggregated into volume-weighted
PADD averages. Separate aggregates were made for domestic RFG and CG, as well as imports.
Meanwhile, volumes for production, movement, and imports were taken from the EIA
Petroleum Supply Annual 2004 and Petroleum Marketing Annual 2004 reports, available from
the EIA website.37 Gasoline volumes used were for 'Finished Motor Gasoline' and were
reported by EIA as 'Reformulated,' 'Oxy' and 'Other.' For purposes of this analysis, the 'Oxy'
and 'Other' volumes were aggregated together as CG.
Due to differences in the sources of data for gasoline properties and volume figures, some
assumptions had to be made to complete the analysis. Major assumptions and their rationale are
as follows.
First, gasoline transported into one PADD from another has the weighted average
gasoline properties of the gasoline produced in the source PADD. While it is possible that
gasoline transported into a PADD is then transported out to another PADD, this information
cannot be known given the available data.
Second, when we estimate the properties for gasoline consumed in future years, we
assume that that the ratios between flows are the same as in the 2004 data, since future
consumption patterns are not known.
Third, because EIA does not supply data on flows between California and the rest of
PADD V, some assumptions were required to separate gasoline properties in these areas. The
volume of California RFG produced beyond what was consumed (a relatively small quantity)
was assumed to be transported into the rest of PADD V, as was any non-RFG gasoline produced
in California. Imports reported for PADD V as a whole were apportioned between California
and the rest of PADD V based on import data tables available on the EIA website. Furthermore,
6-83
-------
Final Regulatory Impact Analysis
California RFG transferred into the rest of PADD V, as well as RFG imports into PADDs IV and
V, are counted as CG at the point of consumption since there are no federal RFG areas in
PADDs IV and V outside of California.
Table 6.10-5 shows a summary of the input figures for gasoline volumes and benzene
content in 2004 and Table 6.10-6 shows the benzene levels after the modeled reduction to meet
the new benzene standard. Volumes shown would be the same if consumption values were
being estimated for another gasoline parameter. Table 6.10-7 shows the estimated benzene
levels for gasoline consumed in each PADD and Table 6.10-8 shows the benzene values after the
modeled reduction to meet the new benzene standard. Differences between production and
consumption volume totals for CG and RFG result from the assumption that all gasoline being
consumed in PADDs IV and V is counted as CG, regardless of designation at production. This
assumption doesn't make a difference for the final value of the gasoline parameter as consumed
in that PADD, only in attribution of the volumes. Table 6.10-9 shows the PADD transfer
volumes taken from the EIA data and used in the analysis. Figure 6.10-1 gives a conceptual
view of gasoline flows between PADDs with production and consumption benzene levels for
2004; the relative size of the arrows indicates approximately the relative volumes of the
transfers.
Table 6.10-5. Inputs to In-Use Analysis based on 2004 Gasoline Benzene.
Production + Imports
PADD
I
II
III
IV
V (ex/CA)
CA
ALL
Total
vol (MMgal)
26,253
32,016
55,822
4,389
4,613
18,618
141,712
bz v%
0.72
1.24
0.87
1.55
1.75
0.62
0.94
CG
vol (MMgal)
11,414
26,513
45,452
4,389
4,613
2,379
94,760
bz v%
0.84
1.33
0.94
1.55
1.75
0.61
1.10
RFG
vol (MMgal)
14,839
5,503
10,370
0
0
16,239
46,952
bz v%
0.63
0.81
0.54
0.00
0.00
0.62
0.63
*This volume of gasoline is likely for the Phoenix area, which has a state fuels program with
requirements similar to federal RFG.
Table 6.10-6. Estimated Benzene Levels After Benzene Control
(vol% in 2011) Production + Imports
PADD
I
II
III
IV
V (ex/CA)
ALL
CG
0.53
0.63
0.63
0.90
0.67
0.63
RFG
0.52
0.61
0.55
-
0.58
6-84
-------
Final Regulatory Impact Analysis
Table 6.10-7. Outputs From In-Use Analysis Based on 2004
Gasoline Benzene and Movement.
Consumption
PADD Total CG RFG
vol(MMgal) bzv% vol (MMgal) bz v% vol (MMgal)
I 50,125 0.59 30,902 0.61 19,222
II 40,166 0.62 34,543 0.62 5,623
III 22,480 0.61 16,978 0.63 5,501
IV 4,387 0.85 4,387 0.85 0
V(ex/CA) 9,709 0.65 9,709 0.65 0
CA 14,846 0.62 0 0.62 14,846
ALL 141,712 0.62 96,519 0.63 45,192
Table 6.10-8. Estimated Benzene Levels after Benzene Control
(vol% in 2011) Consumption
PADD CG RFG
I 0.61 0.54
II 0.62 0.60
III 0.63 0.55
IV 0.85
V (ex/CA) 0.65
ALL 0.63 0.58
Table 6.10-9. Gasoline Flows Between PADDs (MMgal in 2004).
bz v%
0.54
0.60
0.55
0.00
0.00
0.62
0.58
Destination
CG
RFG
I
o
u
g
I
n
m
IV
V
CA
268
22,483
0
2
0
=
tt
§
i
n
m
IV
V
CA
0
0
0
0
0
0
II
3,265
5,361
315
0
0
0
0
0
0
0
0
m
0
323
0
0
0
0
0
0
0
0
0
IV
0
319
428
0
0
0
0
0
0
0
0
V
0
0
525
435
2,295
0
0
0
0
0
1,393
I
0
0
0
0
0
0
0
4,383
0
0
0
II
0
0
0
0
0
0
0
354
0
0
0
III
0
0
0
0
0
0
0
235
0
0
0
IV
0
0
0
0
0
0
0
0
0
0
0
V
0
0
0
0
0
0
0
0
0
0
0
6-85
-------
Final Regulatory Impact Analysis
Figure 6.10-1.
Conceptual View of Inter-PADD Transfers and In-Use Benzene Levels, 2004.
1.55%/1.47%
1.75% 71.21%
Figures listed as
Production/Consumption
These results illustrate a few predominant trends. In-use levels of benzene in gasoline in
PADDs II, IV, and V are depressed by lower-benzene gasoline transferred from PADD III.
Benzene levels in PADD V are further reduced due to transfers from California. Meanwhile,
fuel benzene levels in PADD I increase slightly as a result of imports and transfers from PADD
III.
6.10.1.4
Running the MOBILE Model
Version 6.2 of MOBILE was used for this analysis. To run the model and generate
meaningful outputs, several inputs were required for each case besides fuel parameters as
discussed above.
Temperature Profiles
MOBILE6.2 allows input of a daily temperature profile (24 hourly values) to increase the
fidelity of modeling temperature effects on emissions. Representative cities were chosen for CG
and RFG areas in each PADD, and their temperature profiles were pulled from the database used
in EPA's National Mobile Inventory Model (NMIM). Two profiles were used for each city, July
and January, for summer and winter seasons. These cities, listed in Table 6.10-10, were chosen
because they are relatively large population areas located near the north-south center of the area
associated with use of each fuel type in each PADD.
Note that this choice of representative cities can produce some artifacts in the modeling
results where CG and RFG within the same PADD are consumed in slightly different climates.
For instance, while RFG in PADD I is generally lower in fuel components like benzene and
aromatics than CG in PADD I, the toxics emissions appear lower for CG because it is modeled
as being consumed in Norfolk, which has a warmer climate than New York City where RFG
6-86
-------
Final Regulatory Impact Analysis
consumption is modeled. However, this artifact does not affect the comparisons being made
between the various regulatory scenarios in this analysis.
Table 6.10-10. Representative Cities for Temperature Profiles
by PADD and Fuel Type.
PADD
I
II
III
IV
V
RFC
New York City, NY
Chicago, IL
Dallas, TX
CG
Norfolk, VA
Indianapolis, IN
Austin, TX
Denver, CO
Reno, NV
Maximum Gasoline Sulfur Levels
The MOBILE6.2 command "FUEL PROGRAM : 4" was used, which allowed
specification of average and maximum sulfur levels for years between 2000 and 2015. Average
sulfur levels were calculated as part of the fuel parameter datasets, but maximum sulfur levels
needed to be generated for use in the baseline year cases. Due to the requirements of the recent
gasoline sulfur program, all cases other than the baselines were assumed to have average sulfur
content of 30 ppm with 80 ppm maximum.
For the baseline cases, one approach was to simply take the highest batch sulfur level
reported by a refinery in a given season. However, a few problems arise in doing this. First,
some of these values exceeded the upper limit on input value of 1,000 ppm imposed by
MOBILE6.2. Second, a single very high sulfur batch did not seem representative of maximum
sulfur levels to be seen by a significant number of vehicles in a PADD-wide analysis. Therefore,
after some review of the datasets, a factor of three times the average sulfur was chosen to
represent the maximum sulfur value for CG, while for RFG a factor of two was chosen. This
allowed straightforward calculation of a representative maximum that was generally tolerable by
MOBILE'S input requirements. In any case where MOBILE'S input limit of 1,000 ppm would
have been exceeded using this method (two cases in CG), the maximum sulfur value was simply
set to 1,000 ppm.
Conversion of Oxygenate Blending Percentage to MOBILE Input Values
The fuel parameter datasets used in this analysis do not give reliable information about
what the actual concentration of the oxygenate was in the vehicle fuel tank. For example, the
gasoline data may indicate that on average, gasoline in a certain area had ethanol blended at 5
vol%. However, this could mean that all of the gasoline had 5 vol% ethanol, or half of it had 10
vol% ethanol, each having a different effect on vehicle emissions. Therefore, oxygenate inputs
to MOBILE (using the OXYGENATE command) require two values: blending vol% and market
share. Converting the average blending percent calculated in the datasets to these values
required some assumptions about the blending ratio for each oxygenate type. The figures used
were 10.00 vol% for ethanol, 11.04 vol% for MTBE, 12.78 vol% for ETBE, and 12.41 vol% for
TAME, based on typical blending volumes for these compounds in RFG or gasohol in the case
6-87
-------
Final Regulatory Impact Analysis
of ethanol. From these values, appropriate market shares could be derived. MOBILE6.2 does
not allow modeling of a fuel with a mix of oxygenates, therefore, the sum of market shares for all
oxygenates used must not exceed one.
Start Emission Factor Parameters
Vehicle start emission factors in MOBILE6.2 were adjusted by temperature and vehicle
technology to better characterize cold temperature start emissions observed in recent test data for
Tier 1, LEV and Tier 2 vehicles. These adjustments are discussed in more detail in Chapter 2 of
the RIA. Using a data file set up for phase-in of the cold temperature VOC standards also part of
this program allowed modeling of scenarios with and without phase-in of vehicle controls.
Processing of Output from the MOBILE Model
For each case listed in Table 6.10-1, input scenarios were generated for each PADD, for
CG and RFG, summer and winter. Output values for 1,3-butadiene, acetaldehyde, acrolein,
benzene, and formaldehyde were summed to represent total toxics emissions for each scenario.
The summer and winter seasonal results were annualized (averaged) by weighting according to
the quantity of gasoline supplied in each season according to data taken from EIA. The resulting
figures are presented in Table 6.10-11.
6-8
-------
Final Regulatory Impact Analysis
Table 6.10-11. Estimated Annual Average Total Toxics Performance of Light Duty
Vehicles in mg/mi Under Current and Projected Scenarios.*
Regulatory Scenario
MS ATI Baseline**
(1998-2000)
EPAct Baseline**
(RFC: 2001-2002)
EPAct Baseline, 2011***
MSAT2 program, 2011***
(Fuel standard only)
MSAT2 program, 2011***
(Fuel + vehicle standards)
MSAT2 program, 2025***
(Fuel + vehicle standards)
Fleet
Year
2002
2002
2011
2011
2011
2025
RFC by PADD
I
112
104
67
66
64
39
II
129
121
78
76
72
45
III
97
87
52
52
48
31
CG by PADD
I
114
114
62
60
56
36
II
145
145
83
77
74
45
III
107
107
54
52
47
31
IV
145
145
82
74
70
44
V
156
156
88
81
78
48
* Total toxics performance for this analysis includes overall emissions of 1,3-butadiene, acetaldehyde,
acrolein, benzene and formaldehyde as calculated by MOBILE6.2. Although POM appears in the Complex Model,
it is not included here. However, it contributes a small and relatively constant mass to the total toxics figure (~4%),
and therefore doesn't make a significant difference in the comparisons.
** Baseline figures generated in this analysis were calculated differently from the regulatory baselines
determined as part of the MS ATI program, and are only intended to be a point of comparison for future year cases.
*** Future year scenarios include (in addition to the MSAT2 standards, where stated) effects of the Tier 2
vehicle and gasoline sulfur standards, and vehicle fleet turnover with time, as well as rough estimates of the effects
of increased ethanol blending and the phase-out of ether blending.
6.10.2
Interpretation of Results
The first row in Table 6.10-11 shows mg/mi air toxics emissions in 2000 under the
MS ATI refinery-specific baseline requirements. The second row shows how these would
change by updating the RFG baselines to 2001-02 as specified in EPAct. Since significant
changes are expected in the gasoline pool between 2002 and the projected implementation time
of the fuel benzene program, such as gasoline sulfur reductions and oxygenate changes, we
decided to model a 'future baseline' to allow comparison with the benzene program at the time it
becomes effective in 2011." As a result, the third row shows the projected mg/mi emissions in
2011 under the EPAct baseline adjustments, but without the benzene program. The large
reductions in air toxics emissions between the EPAct baseline and this 2011 baseline are
primarily due to nationwide reduction in gasoline sulfur content to 30 ppm average and
significant phase-in of Tier 2 vehicles across the national fleet.
An important comparison is made between rows three and four, where the estimated
"Ibid.
6-89
-------
Final Regulatory Impact Analysis
toxics emissions under the new gasoline benzene program only are compared to the projected
emissions without the new gasoline program. We also evaluated the effects from inclusion of
the new vehicle standard finalized in this rule on toxics emissions at two points in time, shown in
the last two rows of the table.
In this analysis, all three RFG areas show a slight improvement in 2011 as a result of the
gasoline benzene program in 2011. This is not surprising, since the level of the average benzene
standard, 0.62 vol%, is near the RFG benzene content. The effects of the program on CG are
larger, as expected given the higher levels of benzene in that gasoline pool. The vehicle standard
does not show much effect in 2011, since it is just starting to phase in at that time. By 2025
however, with the fuel benzene program in effect as well as a significant phase-in (estimated at
>80%) of the vehicle standards, a reduction in total toxics emissions of more than 60% from the
baseline is projected for both CG and RFG areas.
Projected emissions in 2011 are lower under the MSAT2 program than projected to occur
otherwise, and much lower than would be required by adjusting RFG baselines to 2001-2002
averages. Therefore, we conclude that adjustment of these baselines as described by EPAct
section 1504(b) will not be necessary.
6.10.3 Conclusions
When RFG and CG toxics emissions are evaluated at this new level of benzene control, it
is clear that the new gasoline benzene program will result in the RFG, Anti-dumping and
MS ATI emissions performance requirements being surpassed not only on average nationwide,
but for every PADD.
In summary, the new benzene program will fulfill several statutory and regulatory goals
related to gasoline mobile source air toxics emissions. The program will meet our commitment
in the MSAT1 rulemaking to consider further MSAT control. It will also bring emission
reductions greater than required under all pre-existing gasoline toxics programs, as well as under
the baseline adjustments specified by the Energy Policy Act.
6.11 MSAT Fuel Effects Test Program
6.11.1 Overview of Test Program
We have recently completed a small fuel effects test program in cooperation with several
automakers to further evaluate the impacts of fuel property changes on emissions from the latest
technology vehicles.0 This study examined exhaust emissions of regulated pollutants (NMHC,
CO, NOX) and several unregulated compounds of interest (1,3-butadiene, acetaldehyde, acrolein,
benzene, ethylbenzene, formaldehyde, n-hexane, naphthalene, toluene, xylene). The fuel
parameters being controlled were benzene, sulfur, and volatility.
0 Participating manufacturers were DaimlerChrysler, Ford Motor Company, General Motors, Honda,
Mitsubishi, and Toyota. Some of these companies are represented by the Alliance of Automobile Manufacturers.
6-90
-------
Final Regulatory Impact Analysis
Nine Tier 2 compliant production vehicles ranging in model year from 2004-2007 and
meeting the Tier 2 Bin 5 or Bin 8 emission standards were tested on chassis dynamometers at
three industry labs and NVFEL over the course of several months. The vehicles were fitted with
catalysts that were laboratory-aged to simulate a service life of approximately 120,000 miles.
Before testing began, a correlation vehicle was circulated to verify that lab-to-lab measurement
variation for all pollutants was within acceptable limits.
Each vehicle was tested three times on five fuels, with a repeat of the first fuel at the end
of the sequence. Four of the test fuels were intended to allow comparisons of the effects of the
three parameters of primary interest, and consisted of abase fuel to which butanes, benzene, and
sulfur were added sequentially to create three additional fuels. In addition to these four fuels,
non-oxygenated Phase 3 California RFG was also tested as an independent baseline. Fuel
property data for the five test fuels is given in Table 6.11-1. In this table, the Fuel ID is
shorthand for how the fuel was made; for instance, BASE is the blending base, while BASERB
has butanes (RVP) and benzene added. This is denoted in the second row below the Fuel ID.
Table 6.11-1. Test fuel properties.
Fuel ID
Description
RVP, psi
T10, °F
T50, °F
T90, °F
Aromatics, vol%
Olefins, vol%
Benzene, vol%
Sulfur, ppm
Density, g/ml
Octane, R
Octane, M
Octane, (R+M)/2
Energy,
Btu/gal net
H/C ratio
Unwashed gums,
mg/100 ml
Carbon Weight
Fraction
BASE
Blending base
6.93
138.7
223.5
324.0
31.4
4.2
0.59
6
0.747
93.2
84.7
89.0
18436
1.82735
1.4
0.867
BASER
Add butane
9.08
127.2
221.0
324.5
28.5
3.9
0.58
6
0.742
93
85
89.0
18487
1.86184
2.4
0.865
BASERB
Add benzene
9.01
126.7
219.6
324.1
28.1
4.0
1.10
6
0.742
92.5
85.3
88.9
18488
1.86267
2.2
0.865
BASERBS
Add sulfur
9.05
127.8
220.6
324.0
28.1
4.0
1.09
32
0.743
92.6
85.3
89.0
18486
1.86127
2.2
0.865
CARFG
California RFG
6.95
136.8
210.0
305.3
21.2
6.7
0.41
5
0.733
91.0
83.7
87.4
18609
1.94208
not
measured
0.860
Figures 6.11-1 through 6.11-4 show conceptual overviews of the test procedures. All test
cycles consisted of the cold start Federal Test Procedure (FTP). Figure 6.11-1 shows the order in
which we tested seven of the nine vehicles on the program fuels. The remaining two vehicles
were tested in a different order. In cases where the sulfur cleanout prep was indicated, two
replicates of the EPEFE high-speed, high-load cycle were conducted immediately before the
6-91
-------
Final Regulatory Impact Analysis
final drain and fill.p The purpose of this type of prep procedure was to equilibrate the catalyst
with the low sulfur fuel. Where a sulfur loading prep was indicated, a 3-hour 35 mph cruise was
conducted immediately before the final drain and fill. The purpose of this prep procedure was to
equilibrate the catalyst with higher sulfur fuel, simulating conservatively the conditions that
might occur in typical suburban driving. The term LA4 indicates a drive cycle commonly used
for preps, which is an abbreviated portion of the FTP consisting of the first two bag periods.
Figure 6.11-1. Conceptual Overview of Testing Procedures.
Oil change procedure
Sulfur cleanout prep
Sulfur loading prep
•
Sulfur cleanout prep
•
Standard prep
•
Standard prep
•
Standard prep
BASE fuel x 3 reps
BASERBS fuel x 3 reps
BASERBfuelxS reps
BASER fuel x 3 reps
CARFG fuel x 3 reps
BASE fuel x 3 reps
p EPEFE is the European Programme on Emissions, Fuels and Engine Technologies, which developed a
protocol for purging contaminates from aftertreatment systems consisting of repeated cycles of high speed cruising
and extended accelerations.
6-92
-------
Final Regulatory Impact Analysis
Figure 6.11-2. Conceptual Overview of Oil Change Procedure
f Proceed to test \
I program J
Figure 6.11-3. Conceptual Overview of Vehicle Prep Procedure,
6-93
-------
Final Regulatory Impact Analysis
Figure 6.11-4. Conceptual Overview of Data Collection Procedure
C
In all, 162 tests were executed to cover all the fuels and vehicles. Each test resulted in
regulated and unregulated emissions data, for a total of 2,592 individual three-bag composite
emissions observations across all pollutants.
6.11.2 Key Findings and Next Steps
Data collected during the test program were analyzed both by EPA and an independent
statistician under contract from the Alliance of Automobile Manufacturers. Table 6.11-2
summarizes the findings of the contract statistician.38
Table 6.11-2. Summary of Significant Effects from Contract Statistician.
Pollutant
THC
NMHC
CO
NOX
C02
Fuel
CARFG
CARFG
BaseRBS
BaseRBS
CARFG
Significant Effect
less than
less than
greater than
greater than
less than
Relative To
All Other Fuels
All Other Fuels
Base
All Other Fuels
Base, BaseRB
An independent analysis of the data conducted internally by EPA generated more
detailed results, and generally found similar trends where the two analyses overlapped. This
work used the SAS software system to run a mixed model on log-transformed 3-bag composite
measurements. Depending on the context of the experiments, the model can accommodate
parameters as either random or fixed. In this case, parameters indicating which lab and which
vehicle were being tested were assumed to be random effects, while the fuel effect was taken as
fixed. This allowed for greater use of all the data collected. For example, since the addition of
benzene does not have a significant effect on VOC or NOX emissions, the effect of RVP for these
pollutants can be determined by comparing the base fuel to both the BASER and BASERB fuels.
6-94
-------
Final Regulatory Impact Analysis
Due to the limited size of the test program, we also used an alpha value of 0.90 instead of 0.95
as the criterion for determining statistical significance.
Table 6.11-3 summarizes the EPA findings. The effect of changes in fuel sulfur content
was relatively large and statistically significant on NOX and CO, and smaller though still
statistically significant for total hydrocarbons (THC). Another finding of importance is that the
change in fuel benzene content produced a statistically significant change in exhaust benzene
consistent with the estimated benefits of the fuel controls as stated in the proposal of this
rulemaking.39 Thus, the effect of fuel benzene on benzene exhaust emissions appears to be little
affected by changes in vehicle technology. Also worth noting is that unlike past programs on
older technology vehicles, these data suggest that reducing gasoline volatility from 9 to 7 psi
RVP under normal testing conditions (75°F) may actually increase as opposed to decrease
exhaust emissions of toxic VOC compounds. It also appears that there is a large statistically
significant effect of fuel benzene on acetaldehyde emissions, though the mechanism for this is
uncertain. If borne out in future testing, reducing fuel benzene will provide additional air toxics
benefits as well. Further details of the results are given in the table.
Table 6.11-3. Summary of Findings from EPA analysis.
Pollutant
Total Hydrocarbons
CH4
NMHC
CO
NOX
1,3 -Butadiene
Acetaldehyde
Benzene
Ethylbenzene
Formaldehyde
n-Hexane
Styrene
Toluene
M,P-Xylene
O-Xylene
RVP (7 to 9 psi)
NS
NS
NS
NS
NS
NS
NS
NS
-11.72
NS
NS
NS
-12.24
-10.95
-12.08
Effects (% Difference)*
Benzene (0.6 to 1.1 vol%)
NS
NS
NS
NS
NS
NS
36.82
18.53
NS
NS
NS
NS
NS
NS
NS
Sulfur (6 to 32 ppm)
12.07
47.62
NS
20.23
48.44
NS
NS
NS
NS
19.81
NS
NS
NS
NS
NS
*Statistical significance was determined using a = 0.90; NS indicates no significant effect at this level. Percent difference is
positive if there is an increase in emissions when the content of the listed fuel property is increased. Regulated pollutants are
shown in italics.
Clearly the data from this scoping study indicate that there may be benefits to future fuel
controls, though in many cases the size of the test program was not sufficient to determine
effects with statistical confidence. At this time, EPA is hoping to conduct a more comprehensive
fuel effects test program, as directed by the Energy Policy Act of 2005, in cooperation with
stakeholders and other interested parties, to generate new data over the next several years. We
expect that work will produce updated emissions models, as well as sufficient data to make
decisions about future fuels programs.
6-95
-------
Final Regulatory Impact Analysis
6.12 Analysis of Future Need for RFG Surveys of Toxics and NOX
Performance under MSAT2
The RFG surveys were created by EPA as part of the RFG program to ensure compliance
with a provision of the Clean Air Act which states that all RFG areas must have gasoline
meeting certain performance requirements.40 Gasoline is often produced far away from where it
is consumed and shipped via a distribution system that treats it as a fungible commodity. The
RFG retail surveys were put in place as a way to measure and document fuel properties at the
point of consumption.
Once the MSAT2 program is fully implemented, our analyses indicate that all gasoline
will meet or exceed statutory requirements under the RFG program as well as existing NOX
performance standards. Therefore, we will no longer require demonstration of compliance with
these programs, and believe it follows that retail surveys for these standards are no longer
necessary.q To verify that this is a reasonable course of action, we have conducted an analysis of
projected emissions performance for future RFG.
6.12.1 Total Toxics Reduction
Within a given RFG area, total toxics emissions as defined by the Complex Model must
be reduced over Clean Air Act baseline gasoline by 20.0% on a per-gallon basis, or 21.5% on an
annual average basis. 41 Once the MSAT2 and gasoline sulfur programs have been fully
implemented, our analyses show that emissions of total toxics from RFG will be reduced beyond
what is required by the applicable statutory and regulatory requirements.
To verify that this will be the case in all RFG areas, we performed a refinery-by-refinery
analysis for each refinery that produced RFG in 2004. We used 2004 batch report data as a
baseline, and then modified each refinery's sulfur level to meet a 30 ppm average standard and
benzene level to meet what our cost modeling projects as the applicable PADD-average RFG
benzene content/ We also removed all ethers and replaced them with 3.5 weight percent oxygen
as ethanol. This change in oxygenate blending is outlined in the documents generated for the
NPRM of the RFS rulemaking.42 Resulting PADD-average RFG fuel parameter values are given
in Table 6.12-1. Note that the analysis was done for each refinery, but due to control of
confidential business information and the need to use PADD-averages for some input
assumptions, PADD aggregates are shown here.
q More discussion of this topic can be found in Section VLB. 3 of the preamble of this rulemaking.
r See section 6.5.4 of this chapter.
6-96
-------
Final Regulatory Impact Analysis
Table 6.12-1. Projected PADD-Average RFG Fuel Parameters Under MSAT2.
MTBE Ethanol Sulfur RVP E200 E300 Aromatics Olefins Benzene
PADD wt%O wt%O ppm psi vol% vol% vol% vol% vol%
1
2
3
0
0
0
3.5
3.5
3.5
30
30
30
9.51
9.65
9.15
50.86
53.50
52.34
82.91
85.26
83.09
20.34
17.97
18.31
14.03
5.04
11.35
0.52
0.61
0.55
Using the individual refinery fuel parameters, we calculated projected total toxics
emissions reductions. The results indicate that no refinery's RFG is expected to fall below 25%
total toxics reduction on an annual average basis. In fact, there is considerable overcompliance
of all RFG beyond what is required by applicable statutes and/or regulations, and we do not
believe there will be any risk of noncompliance in any particular area. These results indicate that
continuation of RFG surveys for toxics performance under MSAT2 is not needed. More detailed
results are given in Table 6.12-2.
Table 6.12-2. Projected RFG Toxics Reductions Under MSAT2.
Annual Average Lowest Refinery Annual Average
Total Toxics Reduction Total Toxics Reduction
PADD Over CAA Baseline Gasoline Over CAA Baseline Gasoline
1 28.1%
2 30.3%
3 29.8%
25.5%
27.4%
25.5%
6.12.2 NOX Reduction
Within a given RFG area, NOX emissions as defined by the Complex Model must be
reduced over Clean Air Act baseline gasoline by 5.0% on a per-gallon basis during the VOC
season (summer), or 6.8% on an annual average basis.43
To verify this will occur in all RFG areas under the MSAT2 program, we performed a
refinery-by-refinery analysis in parallel to the one described above for toxics using the same
model and the same adjusted fuel parameters. The results of this analysis indicate that no
refinery's RFG is expected to fall below 9% reduction in NOX emissions over the baseline
gasoline in the summer season, or approximately 8% reduction on an annual average basis.
More detailed results are given in Table 6.12-3.
Table 6.12-3. Projected RFG NOX Reductions Under MSAT2.
Annual Average Lowest Refinery Annual Lowest Refinery Summer Average
NOX Reduction Over CAA Average NOX Reduction NOX Reduction Over CAA
PADD Baseline Gasoline Over CAA Baseline Gasoline Baseline Gasoline
1
2
o
J
11.4%
15.6%
13.7%
8.3%
13.0%
11.3%
9.4%
10.6%
10.9%
Given these results, we arrive at the same conclusion as for toxics: that there will be no
6-97
-------
Final Regulatory Impact Analysis
significant risk of noncompliance with NOX requirements in any particular RFG area. Therefore,
continuation of RFG surveys for NOX performance under MSAT2 is not needed.
6-98
-------
Final Regulatory Impact Analysis
Appendix 6A: Additional Background on Refining and Gasoline
We believe our discussion of how the benzene content of gasoline can be reduced would
be enhanced with a deeper discussion of how refineries work. In addition to discussing the
various units involved in producing gasoline, we also discuss aspects of crude oil - the primary
feedstock for refineries - gasoline and other products produced by refineries. Because of the
affect of benzene control on octane, we discuss the octane specifications in detail as well. The
information in this Appendix supplements some important information about refineries presented
above. Section 6.1 provides an overview of refining. Section 6.3 provides a detailed discussion
of how reformers work as well as a discussion about the technologies which reduce the benzene
levels in gasoline.
6A.1 Petroleum Refining
Petroleum refineries have been part of our general landscape for at least 150 years. The
earliest examples were little more than a barrel or bucket sitting on rocks or blocks over an open
fire. During those early years, the heavy fractions of crude oil were more valuable when used as
grease for wheels and fuel for heating and lights. The light fractions were either boiled off or
poured-out into a nearby ditch or pond.
Today, petroleum refining is an altogether different industry. The most identifiable
characteristic of most refineries in the U.S., apart from their names, of course, are their crude
throughputs, in barrels per day (bpd). The largest domestic refineries run up to 490,000 bpd of
crude shipped to them by ocean-going barges, pipelines, and trucks from all over the world. The
smaller refineries, of which there are few, run about 10,000 bpd, on average. Even these smaller
facilities occasionally run some foreign crude supplied to them by pipeline; some from Canada is
shipped by pipeline while most of the rest is hauled by marine tankers to terminals along our
coasts. From there the crude is shipped to various parts of the country via pipeline, rail, and
truck.
Most petroleum refineries are much alike, regardless of crude throughput; they consist of
processing units with nearly identical names, the most important of which are: crude units,
vacuum units, reformers, isomerization units, fluid catalytic crackers, hydrocrackers, cokers, and
sulfur recovery units. All refineries have at least one crude unit; many of the larger refineries
have more than one. Most, if not all have at least one or more vacuum units. If a refiner sells
gasoline, he certainly has a reformer. As a refiner adds units to improve his ability to convert
crude barrels into lighter, more valuable products (especially gasoline in the U.S.), he increases
the complexity of his facility. The main differences among the refineries are the sizes or
capacities of the units. Admittedly, all refineries don't have all the units; but to the extent a
refinery has them, it is similar to the others. We believe we should also make the point that even
though two or more refiners may have nearly identical units of some kind, none will likely
produce identical products. Similarities notwithstanding, crude variations and operating
philosophies tend to make significant variations in finished products.
We feel it is neither possible, nor for that matter necessary, to describe every possible
refinery configuration in order to explicate the effects we believe this rule have on refinery
6-99
-------
Final Regulatory Impact Analysis
operations and finished gasoline following the promulgation of this rule.
The "refinery" to which we refer in the following discussion should not be construed to
be any specific refinery or refineries in the U.S., or the world for that matter. None of the units
will have a specific flow rate, unless it is germane to our discussion. Our discussion is
qualitative; we most certainly do not imply nor will we provide any sort of weight or volume
material balance around any unit or the total refinery. Many refineries may have a few of,
several of, or all of the units we discuss. Our discussion of the crudes, intermediates, and
finished products will also be generic by nature, but will hopefully depict them well enough to
be clear about what is meant. We will focus mainly on how benzene is currently produced, and
how and why it is usually found in gasoline; we will then discuss ways refiners may be able to
reduce its final concentration in their gasoline.
We will briefly describe how the primary units operate within an average refinery, with
slightly more detailed discussions of the units that affect the final concentration of benzene in
gasoline. However, the first topic we will discuss is crude oil, since it is both the primary
feedstock to most U.S. refineries and since most crude contains at least some naturally occurring
benzene.
6A.2 Crude Oil
While crude oil is the main feedstock for most refineries, occasionally other stocks may
be purchased which are either processed further or blended directly into finished products.
Crude oil is generally described as a complex mixture of hundreds of different compounds made
up of carbon and hydrogen, the molecular weights of which vary from 16 for methane, the
simplest, to perhaps several hundred, for the most complex. The principal hydrocarbon species
are paraffins (alkanes), naphthenes (cycloparaffms), and aromatics; benzene, the subject of this
rule, is an aromatic. There are also many combinations of these species, such as alkyl
naphthenes, alkyl aromatics, and polycyclic compounds (two or more aromatic compounds
joined into a single molecule). Crude also contains inorganic substances including atoms of
sulfur, nitrogen, and oxygen, as well as metals such as iron, vanadium, nickel, arsenic, and
chromium, in varying concentrations depending on the source of the crude. Collectively,
because these atoms are neither carbon nor hydrogen, they are sometimes called "heteroatoms."
More commonly, they are referred to simply as contaminants. Certain heavy crude oils from
younger geologic formations (e.g., Venezuelan crudes) contain less than 50 percent
hydrocarbons and a high proportion of organic and inorganic compounds containing
heteroatoms. Over the years, many refinery processes have been developed to remove or reduce
their concentrations to low-levels because they damage catalysts. Likewise, our recent rules
were promulgated in order to reduce the negative effects some of these heteroatoms have had on
the environment.
In the world each day, a huge volume of crude oil is produced, shipped, and refined. It is
sold according to its quality and availability. The market price of a particular crude is usually
calculated according to formulae that relate its API Gravity and sulfur content, and perhaps other
criteria, to an agreed upon index. These indexes vary according to other indexes, depending on
where the crude located. Nevertheless, at any given time, it is a reasonable expectation that
6-100
-------
Final Regulatory Impact Analysis
nearly any refinery in the U.S. could be processing crude from almost any country in the world.
As a brief introduction to what follows, we note that the gasoline produced by most
modern refineries consists of several blendstocks, most of which are usually produced in that
refinery. We used the term "usually" in the previous sentence, since from time-to-time, refiners
purchase feedstocks and blendstocks from other sources. During the early days, refiners used
simple distillation (fractionation) technology, to recover as much naturally occurring straight-run
gasoline as possible. During the past 60 to 70 years, there has seen a steady drive to develop
processes and catalysts that convert as much as possible of any given crude barrel into high-
quality, light products such as gasoline and diesel. Today, in the U.S., there is very little finished
fuel that hasn't in some fashion been upgraded after it leaves the crude unit. This has been
especially the case for gasoline. However, even now or at least in the near future, relatively
more kerosene and diesel will be processed as a result of recent low-sulfur rules.
As far as reducing the benzene content of gasoline is concerned, a refiner may be
fortunate enough to purchase crude with less naturally occurring benzene and fewer benzene-
precursors. Regardless, since much crude contains at least some benzene and benzene-
precursors, the crude unit is usually the first opportunity a refiner has to begin controlling the
final benzene concentration in his gasoline. However, that "first opportunity" doesn't come at
the beginning of the process. Consequently, we feel our discussion will be made more
intelligible by describing the entire process, beginning with the crude unit and including several
other benzene producing processes. We will then high light the points where process changes
can be made to control both the naturally occurring benzene and the reformer feed benzene
precursor content which will ultimately reduce the overall content in the gasoline going to
market.
6A.2.1 Crude Desalting
Usually, water, or brine, from a variety of sources is recovered with crude at the time it's
produced. Crude and water are often produced as an emulsion as a result of the recovery pump's
shearing action. One of the main reasons the water is called brine is that it usually contains a
variety of water-soluble salts and suspended solids, which are potentially corrosive and
otherwise damaging, but also tend to stabilize the emulsions. Depending on the oil's
composition, its pH, and to some extent, the quantity of suspended solids, some emulsions
gradually "break" on their own in a field tank. Occasionally, however, tight emulsions form that
can only be broken using heat and sometimes an emulsion breaker. One of the first and most
important lab tests run on raw crude is called the test for "Basic Sediment & Water" (BS&W).
Oil field operators are usually able to reduce the BS&W of most crude to around one percent or
less, by volume, before the crude is shipped to a refiner
While some contaminants may settle-out in the feed tank with the water, refiners have
learned that desalting ahead of the crude unit is usually economically very beneficial. Even at
1% or less, BS&W can still cause problems. Inorganic, water-soluble salts, e.g., sodium,
calcium, and magnesium chlorides can hydrolyze in a crude furnace and eventually combine
with water (condensed stripping steam) usually found in most crude tower-overhead systems to
form acidic solutions that are very corrosive to the overhead internals. Consequently, most
6-101
-------
Final Regulatory Impact Analysis
refiners choose to desalt their crude ahead of the crude unit. Desalting is a continuous operation,
during which warm crude is vigorously mixed at the proper mix-ratio with clean water and
occasionally some proprietary chemical or other, after which the oil/water mixture is allowed to
separate with the aid of an electrostatic precipitator. The water and sediment are continuously
withdrawn and sent to water disposal facilities. The washed crude is fed to the crude preheat
train.
6A.2.2 Atmospheric Crude Unit
We will use the term "straight-run" from time-to-time in the following discussion. It
refers specifically to any product produced from crude by an atmospheric unit, especially the
crude unit. We believe this is a fairly common usage. As such, the rest of the streams in the
refinery are processed further in some manner and are no longer "straight-run" products.
6A.2.3 Preflash
Most crude contains some light gas, most of which is butane; crude occasionally contains
some propane and isobutane, but their percentages are usually quite low. Often, refiners use a
preflash unit to remove the butanes and occasionally propane. Occasionally, a preflash unit may
be used to make a single distillation cut between the Cs's and Ce's or the Ce's and Cy's. In
effect, this sets the final boiling point (FBP) of the light cut, which is fed to an isomerization
unit. A refiner also has the option of making the preflash cut between the Ce's and Cy's, and
sending the Ce- cut over the top. This cut is then fed to the main crude column above the heavy
straight run tray. This is usually done in order to unload the feed zone and reduce the vapor
traffic in the lower rectification sections of the main column.
Preflash units, often referred to in the early days simply as knock-out drums or tanks,
were and still are, usually located somewhere in the feed line after the feed pump. Early on, they
were often no more than a simple tank with a diameter-to-height (or length/diameter or head-
space) ratio sufficient to reduce the flowrate enough for the gas to separate from the liquid phase
and be removed under pressure control. Initially, many of these drums were horizontal, bullet-
type, tanks similar to those used to store liquefied petroleum gas (LPG) and/or other light-
hydrocarbons. Over time, a variety of internals, such as baffles and packing, were added to
improve the separation efficiency. Again, depending on the volume, the off gas is usually sent to
the suction-side of the wet gas compressor in the FCC gas concentration (gas-con) unit for
recovery; if the volume is small it is ordinarily sent to the fuel gas system.
As discussed above, the actual vessel may not have been more than a simple flash drum
that would provide at most only one or two theoretical separation stages and essentially no
stripping. Ordinarily, a refiner doesn't expect to accomplish much more than to make a
reasonably clean, if somewhat inconsistent gas/liquid separation; clean liquid/liquid cuts were
seldom really possible, of course depending on the equipment and controls. Nevertheless, it was
usually sufficient for degassing purposes; preflash units have become increasingly more complex
and efficient as refiners have geared-up to increase efficiency, refine an increasing variety of
crudes, and to meet the more stringent quality and compositional requirements necessary for
low-sulfur and reduced toxics compliance. Currently, many, if not most units include a
6-102
-------
Final Regulatory Impact Analysis
distillation-type tower (similar to a crude tower, but usually much shorter), with trays or packing
and a reboiler (thermosiphon or heater/furnace type) to provide stripping. Generally, this kind of
preflash unit will not only efficiently remove the light gas referred to above, but can also make a
fairly decent or clean, single, overhead/bottoms cut to remove the Cs/Ce light ends from the rest
of the crude; we note here that preflash towers usually don't have side-draws. In recent years,
electronic process controls, e.g., distributed control systems (DCS), have begun to play a
significant roll in helping operators make cleaner cuts than were previously possible using the
older pneumatic controllers to control what were fairly inefficient preflash towers/vessels.
The preflash operating conditions, such as flowrate, feed temperature, tower pressure,
and reflux and reboiler rate, would be set according to the feed composition and the desired cut.
The overhead, consisting of pentanes and lighter and some hexanes is condensed, cooled, and
collected in an overhead accumulator and degassed, e.g., the non-condensable gases are removed
from the accumulator under pressure control. Part of this condensed hydrocarbon is pumped as
reflux to the tower's top tray or, if the tower is packed rather than trayed, to the top of the
packing; ordinarily, there are no side-draws. The off-gas from the preflash is usually sent to the
wet-gas compressor in the fluid catalytic cracker (FCC) gas-concentration (gas-con) unit, if there
is enough gas and the refinery has a gascon, as most modern refineries do. The excess overhead
liquid, under level control, is sent to a naphtha splitter.
6A.2.4 Crude Unit
Regardless, the desalted crude preheated in feed/effluent heat exchangers against hot
crude tower product rundowns to recover process heat. It is subsequently fed either to the
preflash or to the crude charge furnace for trim heating to about 650° to 700° F and fed to the
flash zone of the crude tower at a pressure slightly higher than atmospheric. An ordinary crude
tower consists of a steel cylindrical column, which is usually around 100 ft. to 120 ft. tall to
accommodate the number of trays and their spacing, and whose diameter is set according to the
design feedrate. We won't discuss the minutiae of the heat and mass transfer dynamics of crude
fractionation at this point, but we will mention that the tower diameter is set according to the
feedrate, such that the vapor/liquid velocities in the tower and the tray liquid volume and
residence times will allow the transfer of heat and material to reach a condition of stable
equilibrium at each tray. A common assumption that may cast some light on the vapor/liquid
traffic in a crude tower is that, at equilibrium, the moles of liquid traveling down the tower will
equal the moles of vapor traveling up the tower.
The distillation or fractionation "tray" of which we speak, is a type of plate or tray
(usually a type of steel or steel alloy about a quarter-inch thick) installed at equal distances apart,
one above the other, beginning just above the feed zone and continuing up the entire height of
the column. These are ordinarily called distillation, fractionation, or simply tower trays and are
usually designed and spaced according to specific criteria involving far too many factors for us
to discuss here. Regardless, on average, while there could be as many as or seven or eight trays
between each draw tray, there may be as few as four or five. The number usually has to do with
desired product purity, but is also related to tray design limitations such as pressure drop per tray
and with column height.
6-103
-------
Final Regulatory Impact Analysis
The trays are designed to maintain a specified liquid level on their surface, deep enough
for good vapor/liquid contact, but as more condensed liquid falls onto a tray and reaches the
predetermined maximum level, there must be a mechanism by which excess liquid can fall down
to the next tray. A couple of ways are to drill specified diameter holes in the tray (these trays are
usually called "sieve trays") or to install "down-comers" from one bubble-cap tray to next tray
below.
Please note that we have mentioned only two types of trays, sieve and bubble cap, which
are quite common and have been in use for many years. There are in fact several others, many of
which are of proprietary design. There are many designs, but the purpose of all of them is to
provide a way for the vapor traveling up and liquid traveling down to come in contact in order to
provide for heat and mass transfer at as low-pressure drop as possible. At each tray the liquid is
enriched with heavier components and the vapor is enriched with lighter components. At
specific levels in the column, design engineers predict that the condensed liquid will look like
one of the products the refiner would like to produce. They install draw trays at these levels,
from which the straight-run products are each withdrawn.
As we mentioned in the first paragraph of this section, the hot crude is fed to the feed or
flash zone of the atmospheric crude column or tower. Within the flash or feed zone, the
components whose characteristics, e.g., boiling points, are such that they vaporize, separate from
those components that remain in the liquid phase at tower conditions. The vapors begin to rise
into the rectifying section of the tower while the heavier liquid falls into the tower stripping
section. We will briefly discuss the tower bottom operation first, followed by a discussion of the
vapor phase as it leaves the flash zone. The last crude tower stream we'll discuss will be the
heavy straight run, which is fed to the reformer to become one of the more important gasoline
blendstocks. Our discussion of gasoline and how it's produced will proceed from there.
6A.2.5 Atmospheric Tower Gasoil and Residuum; Vacuum Unit
The heavy ends of the crude, which didn't vaporize in the feed zone, fall down over three
or four stripping trays installed in the crude tower bottom. High-pressure steam is injected under
the bottom tray to strip out any remaining light-ends. The stripped crude tower bottoms (ATB)
are removed, cooled against feed and sent to storage. There are times when the ATB's may be
fed directly to a vacuum tower; regardless, there is usually provision for sending at least a
slipstream to storage.
Vacuum Unit: We have included a discussion of the vacuum unit as part of this section.
It plays an important role in producing road asphalt, and lube oil feedstocks as well as feed for
the FCC, an important gasoline and diesel producing process and occasionally the coker. In
some cases, the AGO, which we will presently discuss is fed to the FCC while the ATB is fed to
a vacuum unit rather than directly to the FCC.
A vacuum unit is necessary in order to process the heavy or high boiling ATB stream to
recover the components which, separately, are more valuable in other markets. Most crude
begins to thermally crack at around 700° F and atmospheric pressure; some crude will begin to
crack at as low as 650° F, while others may not begin until upwards of 750° F. It is therefore
6-104
-------
Final Regulatory Impact Analysis
necessary to use a vacuum unit to lower the boiling points of the ATB components. The vacuum
may be generated using stream driven eductors or, more recently by using vacuum pumps. As a
rule, the greater the vacuum is, the better. The entire design of the unit is of course critical in
order to make the desired separations and recoveries. One very important issue is the design of
the tower feed line and the tower flash zone. If the feed has not sufficiently vaporized in the
tower feed line, it may explosively vaporize in the flash zone, to not only make the vapor/liquid
separation as clean as possible, but rapidly expanding vapors can also dislodge tower internals.
If the tower is being used to produce asphalt, the flash zone operation is critical. If the feed
vaporizes explosively in the flash zone, the high velocity vapor components may carry
asphaltenes upward with them, and eventually contaminate the heavy vacuum gasoil.
A vacuum tower ordinarily produces a low-volume overhead that boils in the heavy
naphtha to kerosene range. These are generally light components that didn't strip out of the
ATB with stripping stream at the conditions in the crude tower bottom, but which readily
separate out under vacuum tower conditions. The unit usually produces a small volume of light-
vacuum gasoil, which is routinely fed to the distillate hydrotreater and eventually to distillate
blending. The lower side cut is called heavy vacuum gas oil, HVGO. We use the term "cut" for
convenience, knowing that the draws from the vacuum tower aren't "true" distillation cuts in the
technical sense of the term, used when discussing fractional distillation. The number of
theoretical stages in a vacuum tower is usually quite low compared to a crude tower; perhaps no
more than nine or 10 theoretical stages for the entire tower. Depending on the crude source,
HVGO may qualify as lube stock; otherwise, it would be fed to an FCC. If the original crude
was asphaltic, the vacuum resid or vacuum tower bottoms (VTB) may qualify as asphalt for use
in the paving and roofing industries or could also be fed to a hydrocracker or a coker. Another
important difference between vacuum towers and crude towers is that vacuum towers are true
distillation towers. The draw trays are referred to as total draw trays; that is, there is liquid
released from the tray down to the section below it, so there is no true internal reflux. The
"internal reflux" is provided by "pump-arounds." That is, light and heavy vacuum gasoil is
pumped into a distribution nozzle some distance above each of the two draws. There may also
be "pump-back" streams, which are pumped back to the tower under a draw tray. Another
important stream is the one pumped back under the HVGO draw tray, which washes
contaminants such as asphaltenes from the vapors leaving the flash zone. Most vacuum units
can produce several grades of asphalt, a few of which may be back-blended to produce others, as
needed. Some refiners use solvent deasphalting to produce finished asphalt. High-flash point
asphalt is usually air-blown in a plant designed specifically to produce roofing asphalt. We also
note that not all asphalts are alike. Some are especially good for producing road oil and asphalt,
but not for producing roofing asphalt; the reverse is also true. Polymer modified asphalt has
become very popular with highway engineers. Some types of asphalt work well when blended
with polymers to improve their highway performance, while others do not. With few exceptions,
asphalt qualities and the uses for which asphalt may be produced are closely related to the crude
from which the asphalt was originally derived. Vacuum tower bottoms may also be fed to a
coker, from which liquids may be recovered along with the coke.
For several reasons, the products derived from a barrel of average crude coming directly
from a crude unit have become increasingly less useful for market. There appear to be at least
two reasons; there are probably others. One is that the average crude barrel available to U.S.
6-105
-------
Final Regulatory Impact Analysis
refiners has gradually become heavier (e.g., has a lower percentage of light straight-run products
such as naphtha and diesel and more heavy cuts such as the AGO and ATB that we've just
discussed). Moreover, heavier crude usually contains increasingly higher percentages of
contaminants, which must be removed by some type of downstream processing. Secondly, not
only has the demand for light products (especially gasoline and diesel) grown quite rapidly, but
likewise the finished product quality specs, apart from those imposed by government regulations,
have become very high.
We will now discuss the crude tower operation above the flash zone. The fraction of the
crude that vaporizes in the feed or flash zone at the above referenced temperatures and pressures,
separates from the heavy liquid fraction and (the vapor) begins to rise upward through the tower.
As it rises it becomes progressively cooler and the heaver fractions begin to condense. In effect,
once the tower reaches a state of dynamic equilibrium, the vapor traveling up and condensed
liquid falling down the column are continually contacting each other to exchange heat and mass.
The first draw tray above the flash or feed zone will begins to fill with liquid which eventually
becomes atmospheric gasoil (AGO) when it is finished.
In this section, we will discuss the specifics of how the AGO draw is handled. We note
that the other side-draws above the AGO are handled in much same manner; other than listing
them, they won't be discussed. The withdrawn liquid is fed to a steam stripper to adjust its flash
point. This is necessary because the liquid taken from the column will always contain at least
some of the lighter, lower boiling components, which condense higher in the column, but that are
continually part of the traffic in that section. This withdrawn liquid contains components,
besides the AGO cut, such components as diesel, kerosene, heavy and light naphtha, and steam
used to strip the tower bottoms. These are all removed from the AGO by steam stripping. A
steam stripper is a small cylindrical vessel, into which about four to six perforated (sieve trays)
are installed. The draw liquid is fed into the side of the column at the top through a distribution
nozzle or pipe and falls down over the trays, while high pressure (>150 psi) steam is injected
into the column under the bottom tray. The stripping steams does not actually physically strip
the light ends from the liquid. Rather, its presence changes the partial pressure of the light ends
and helps them disengage from the hot liquid, following which they are carried up and out of the
stripper top along with the steam. These gaseous components are fed back into the crude tower
just above the draw tray and once again become part of the tower traffic. The stripper bottoms
are usually cooled against crude feed in a feed/effluent exchanger, water cooled, and sent to
storage.
The vapor above the AGO draw continues up the tower, progressively cooling and
condensing as it travels. Draw trays are installed at levels where diesel, kerosene, and heavy
naphtha (heavy straight-run, HSR), are each withdrawn from the tower in that respective order
proceeding upward. Each is stripped, cooled, and sent to storage much the same as we described
for the AGO.
The crude tower overhead, which usually consists of Cs's thru Cn's, is ordinarily fed to a
naphtha splitter (see below). The usual configuration has a feed flow controller, which
maintains a steady feedrate to the splitter. It is installed in a pipe or line position from which it
can control the crude tower overhead flow such that it can feed the splitter directly from the
6-106
-------
Final Regulatory Impact Analysis
crude tower overhead drum. However, if the crude tower overhead rate becomes too high for the
splitter, the splitter feed controller can open a valve in another line that will send the excess to
storage. On the other hand, if the crude tower overhead flow is too low, the splitter feed
controller can close the valve to storage and open still another valve to draw makeup feed
volume through a different line from storage. In other words, this arrangement not only
maintains a constant feedrate to the splitter, but the crude overhead storage tank provides surge
capacity for the crude unit as well as feed to the splitter should either come down unexpectedly.
Additionally, some refiners use a reformer feed tank to which splitter bottoms run down and
from which the reformer is fed to provide some surge capacity for the reformer in case of
splitter-unit problems.
6A.2.6 Naphtha Splitter
The naphtha splitter cuts the C5's and some C6's into the overhead while most of the C6's
and Cy+ cut is removed from the tower bottom. Pentanes do not make good reformer feed. They
are not converted into aromatics and although they have a relatively decent octane, it is
somewhat lower than usual reformate and actually dilutes the reformate octane. Another
drawback of having pentanes in the reformer feed is that they usually crack to gas and thus
actually reduce finished liquid yield.
We believe it is noteworthy that until recently, most of the Ce's were typically fed to the
reformer. Cyclohexane, for example, with a clear RON of around 83.0, is usually converted to
benzene which has an octane blending value >100. Also, naturally occurring benzene boils in
approximately the same boiling range and has been an important gasoline blending component
for many years. Nevertheless, despite best efforts, some Ce's ended up in the isom feed. We
believe it is also worth noting that prior to the lead phase down this stream was routinely called
light-straight run and was very susceptible to tetraethyl lead (TEL). As a rule, TEL raised the
clear LSR by around 15 numbers; this varied somewhat depending on the crude source.
Fortunately, most refiners were able to install isom units to replace the octane lost with the
removal of lead.
The splitter overhead typically contains at least some of the following light
hydrocarbons: isopentane, normal pentane, cyclopentane, 2, 2 dimethylbutane, 2, 3
dimethylbutane, 2 methylpentane, 3 methylpentane, normal hexane, methylcyclopentane,
cyclohexane, and benzene. The isomerization (isom) unit bottoms are routinely fed to a naphtha
reformer. Until recently, e.g., promulgation of the MSAT rules, the splitter distillation cut was
made approximately between the C5's and C6's, providing a C5 minus cut to the isom and the C6
- FBP cut to the reformer. We will discuss these cuts as they apply to benzene reduction in more
detail later.
6A.2.7 Hydrotreating
We will discuss hydrotreating technology because it plays an important role in the feed
preparation for many of the units we will be discussing. Hydrotreaters use catalysts at high
temperatures and pressures with fairly pure (>75% and often >95% pure hydrogen to remove
contaminates, such as sulfur, nitrogen, and heavy metals from a variety of feedstocks to other
6-107
-------
Final Regulatory Impact Analysis
units. The "hydro-" prefix indicates hydrogen is used in the main reactions. Hydrotreaters may
be referred to by a variety of names such as hydrodesulfurization units (specifically remove
sulfur), distillate hydrotreater, or hydrodenitrification units (specifically remove nitrogen). Also,
the acronym HDT is often used when referring to a distillate hydrotreater; HDN refers to a
naphtha treater, an important pretreater for a reformer. There are also FCC feed hydrotreaters,
usually called "cat feed hydrotreaters." There are of course, pumps, compressors, heat
exchangers, high- and low-pressure separators, as well as flashpoint stabilization units associated
with these units. Hydrotreaters use hydrogen from either a steam/methane reformer or a catalytic
naphtha reformer.
The catalyst usually consists of a combination of cobalt, molybdenum and nickel, applied
to the surface of an alumina extrudate. Over time the catalyst deactivates as a result of coking
and/or metal poisoning and must be either decoked or else replaced. When the catalyst
deactivates, the coke can be burned off (either in the reactor or off-site by a contractor) and
reused. Typically catalyst can be used a few times before it needs to be replaced. It is ordinarily
not possible to regenerate a poisoned catalyst.
Sulfur compounds are converted into hydrogen sulfide, which is routinely removed from
the process recycle and/or off gas in an amine extraction unit, following which the hydrogen
sulfide is removed from the amine and converted into elemental sulfur. Nitrogen is removed
using a sour water stripper, as ammonia, which is removed in an ammonia recovery plant.
The reactor is the dominant feature. Hot feed, the temperature of which depends on the
catalyst type, the stream being treated and the contaminants being removed, is usually mixed at
high pressure with hot hydrogen gas, usually from a catalytic reformer and fed down-flow
through a distribution tray, onto the catalyst bed. If the reactor is tall and has several beds, the
mixed hydrocarbon/hydrogen stream being treated may be withdrawn from open spaces or gaps
between some of the beds and fed back to the next bed through a re-distribution tray. This helps
prevent channeling, especially if the stream is liquid. Catalyst is not consumed in the process,
but lowers the activation energy of the chemical reactions needed to remove the contaminants.
As a rule, the heavier the feed and the more difficult the contaminants are to remove, then the
higher will likely be the temperature and pressure of the process. Catalyst type obviously plays a
pivotal role in setting the operating conditions. For example, if a catalyst is a "hot catalyst" the
operating condition may be less severe than for a less-active catalyst. We mention here that the
reformer and the FCC are units whose feeds are usually hydrotreated. If the FCC doesn't have a
feed hydrotreater, the heavy crackate, a potential gasoline blendstock, may need to be treated in
order to meet sulfur specs. The light cycle oil will also need to be treated before it is used in
distillate blending; if the light cycle oil can be stored separately, it could potentially be sold in
the fuel oil market; otherwise, it would need to be hydrotreated before it could be sold into the
ULSD market
6A.2.8 Fluid Catalytic Cracker
Generally FCC feedstocks are made up of heavy or lower API Gravity fractions, such as
AGO, ATB, and HVGO. For many years, before the demand for light products reached the level
it is today, these fractions were marketed as fuel oil, mostly in heavy industry. However, the
6-108
-------
Final Regulatory Impact Analysis
demand for light products, especially for gasoline, was a great motivator for the development of
processes that would convert these low-value heavy oils into higher-value light products.
Cracking, a generic reference to the process began to be used commercially the early part of the
20th century. The first units were called thermal crackers which used high temperatures to
thermally crack heavy stocks. Eventually, fixed-bed catalytic crackers were used, one of which
was the Houdry fixed bed process the success of which was recognized in the late 1930's.
Around that time, work was going on to develop a process using finely powdered catalyst, which
subsequently led to the development of the fluidized bed catalyst cracker or fluid catalytic
cracker (FCC). Originally, grinding fixed-bed catalyst material produced the finely powdered
catalyst. More recently it has been produced by spray-drying a slurry of silica gel and aluminum
hydroxide in a stream of hot flue gas. If done properly, a catalyst can be produced consisting of
small spheres in the range of 1-50 microns particle-size.
FCC feed hydrotreaters have become more common as a result of recent government
regulations limiting sulfur in diesel and gasoline. Many refiners have determined that feed
hydrotreaters improve the liquid volume recovery sufficiently, in some cases, to earn a
reasonable return on their investment.
Regardless of whether the feed has been hydrotreated, the fresh feed and possibly FCC
fractionator bottoms or heavy cycle oil are fed into a riser with hot catalyst; the catalyst is
typically regenerated, a topic of which we will speak in a moment. The charge can be heated by
an available source, e.g., furnace or heat exchange. As the feed vaporizes, the cracking reactions
begin and entire mix is carried upward through the riser. At the riser top, the mixture is fed into
a reactor from which the catalyst and hydrocarbons are separated. The reactor effluent
hydrocarbon stream is fed to the FCC fractionator, while the catalyst falls down a pipe into the
catalyst regenerator. During the cracking reactions, coke forms on the catalyst and deactivates it.
The coke is burned off in the regenerator and essentially reactivated and prepared for reuse; an
air blower supplies the required combustion air to the regenerator. The regenerated catalyst
passes down the regenerator standpipe to the bottom of the riser, where it joins the fresh feed and
the cycle repeats. Over time, part of the catalyst becomes unusable, e.g., is crushed into fines,
and is replaced on a continual basis from catalyst storage, such that a proper amount of catalyst
of sufficient activity is always available. In what is sometimes referred to as a power recovery
system, a stream of flue gas drives a turbine, which is connected to the air blower. In that
catalyst fines would quickly erode the turbine vanes, the flue gas stream passes through several
small cyclone separators before it reaches the turbine. The waste heat in the flue gas is finally
used to generate steam.
The fractionator separates the reactor effluent into three main streams. The crackate or
cat gasoline and mixed olefins are removed in the overhead; the light cycle oil, a side cut, is
steam stripped and sent to storage to eventually be used in distillate blends; the fractionator
bottoms are often referred to as slurry oil or heavy cycle oil. Occasionally the heavy cycle oil is
fed as a recycle stream back to the FCC riser, but is seldom recycled to extinction; it may also be
fed to a coker. The light olefins are sent to the gas concentration unit (gascon) for recovery and
further processing into polymer gasoline and alkylate.
While the FCC cat gasoline does contain some benzene, it is not a major contributor to
6-109
-------
Final Regulatory Impact Analysis
the total benzene concentration in finished gasoline. We don't expect much will be done to
reduce the benzene in cat gasoline.
6A.2.9 Alkylation
The alkylation process combines a mixture of propylene and butylene which are usually
produced by the FCC, with isobutane in the presence of an acid catalyst, usually either sulfuric
or hydrofluoric acid. The product, alkylate, is a mixture of high-octane, branched-chain
paraffinic hydrocarbons. Alkylate is considered to be a high-grade blendstock because it has
high octane and contains essentially no contaminants. Two of the more common processes use
either sulfuric or hydrofluoric acid as catalyst.
In the sulfuric acid catalyzed process, propylene, butylene, amylene, and isobutene are
used. Isobutane, often produced by a butane isomerization unit, and the acid catalyst are mixed
and fed through reaction zones in a reactor. The olefms are fed through distributors into each
zone as the sulfuric acid/isobutane mixture flows over baffles from zone to zone.
The reactor effluent is separated into hydrocarbon and acid phases in a settler, from
which the acid is recycled to the reactor for reuse. Some acid is routinely lost and must be made
up. The hydrocarbon phase is washed with caustic for pH control (to completely neutralize the
acid) before it is fed, in series, to a depropanizer, a deisobutanizer, and a debutanizer. The
deisobutanizer bottoms or alkylate can be sent directly to gasoline blending; the isobutane is
usually recycled back to feed and the propane may be recycled back to the gascon unit for
propane recovery.
6A.2.10 Thermal Processing
Thermal processing was one of the first ways early refiners processed crude. There are
essentially three current processes that qualify as thermal processors: delayed coking, fluid
coking, and visbreaking. All are used for the purpose of producing more valuable products such
as catalytic cracker feed and to reduce fuel oil make. Of themselves, they produce only minor
volumes of naphtha which must be severely hydrotreated and generally reformed before it can be
used as a gasoline blendstock.
6A.3 Gasoline
A previous rule provided several important health benefits by reducing the benzene
content in gasoline. We believe the health data gathered since then provides strong support for
removing even more benzene. We will review the refining processes that produce the usual
components from which gasoline is formulated; our discussion of specific units that produce
benzene will be more detailed. We believe this will provide coherence to our discussion of how
refiners can reduce gasoline benzene content. It is important to note that regardless of the
negative health effects, benzene also contributes to gasoline octane and, thereby, to our ability to
produce the engines that help power the world's economy. We will also discuss ways refiners
may be able to recover the octane lost as a result of removing benzene.
6-110
-------
Final Regulatory Impact Analysis
Refineries in the U.S. are complex industrial plants that process various crude oil
feedstocks into many important products. Among the most important of these, but certainly not
limited to them, are gasoline, jet fuel, kerosene, diesel fuel, fuel oil, and asphalt. Many refinery
intermediate streams, such as those produced by fluid catalytic cracking (FCC), become
feedstocks to processes in the chemical industry. The sophistication of these refineries varies,
from simple to very complex. The level of complexity is defined by the various types of
equipment (i.e., units) in use at the refinery. Refineries have been built (or added to) during
different engineering 'eras', e.g. they utilize different generations or technologies to achieve
similar refining goals, all the while attempting to maximize profitability. While, modern day
refineries process crude oil from nearly all countries of the world, the crude oil processed at
each, varies geographically, according to availability and pricing, and of course according to
where it markets its products. We will discuss how a refinery works in somewhat more detail in
a later section. Our focus for this section is automotive gasoline.
6A.3.1 Gasoline as a Complex Mixture
While gasoline is not actually formulated around its chemical composition, per se, it does
have a few specific characteristics, somewhat related to the chemicals of which it consists, that
are very important and should be high-lighted. With regard to those specific chemical or
compositional characteristics, we describe modern gasoline as a complex mixture of
hydrocarbons (compounds of carbon and hydrogen) which boil in the range of about 100° F to
around 410° F (C5 to C12, paraffins, isoparaffins, aromatics, naphthenes, and olefins). Gasoline
has a specific gravity of around 0.7; its API Gravity is about 65. We note that this is the boiling
range for the fraction of gasoline that is liquid at ambient temperature and the sea level air
pressure. Most gasoline, regardless of the season, contains some n-butane (boiling point at sea
level: around 31° F), used to adjust the RVP; gasoline RVP varies seasonally from around 7 psi
to!5 psi. Many regions, cities, etc., of the nation vary both below and above that range. If a
sample of gasoline is allowed to stand in an open container, the butane (and probably some
volume of the other light components) will likely weather-off, quite rapidly. The next species, in
the boiling order, would be isopentane, which boils at about 82° F, followed by n-pentane, which
boils at about 96° F; this accounts for the initial boiling temperature we reported above. A
chromatogram would likely detect all the low-boiling species, but a normal ASTM D-86
distillation would only pickup those species boiling above the ambient temperature. The low-
boiling components, which don't normally condense in the non-pressurized lab equipment,
would be reported as losses; even so this would, in fact, be a measure of their percentage in the
gasoline sample.
Gasoline is formulated to fire, modern spark-ignited, internal-combustion engines.
Diesel, a much heavier product, is used to fire pressure-ignited engines, an altogether different
technology. The initial boiling point (IBP) is controlled so as to provide easy cold and hot start,
prevent vapor lock, and maintain low evaporation and running-loss emissions. Midpoint
volatility is controlled to promote quick warm-up and reasonable short-trip fuel economy, power,
and acceleration. The final boiling point (FBP) is controlled to promote fuel economy and to
provide good energy density.
6-111
-------
Final Regulatory Impact Analysis
As we discussed earlier, IBP of standard gasoline is around 100° F. However, as we also
discussed, low-boiling components, such as n-butane, which usually don't show up in a boiling-
point table, are added to increase volatility; there must be components present that will vaporize
at lower than ambient temperature and pressure, otherwise, an engine won't start, especially
during cold times. Only gasoline vapor burns; the liquid does not. Normal-butane also changes
the partial pressure of the mix to allow other heavier components to more easily vaporize.
Isopentane also plays an important role in this process. Consequently, during cold months, the
amount of n-butane in gasoline is normally increased. On the other hand, older engines with
carburetors, had problems if there was too much light product in the fuel; the carburetor could
vapor-lock and the engine wouldn't start. Fuel-injected engines have reduced that problem.
Even so, the issue of lower vapor-pressure today has more to do with reducing the volume of
unburned hydrocarbons being released into the environment. We mentioned above, that at
ambient conditions, n-butane will quite rapidly evaporate from gasoline. If it isn't maintained at
lower concentrations and otherwise carefully controlled, during warm and hot months, it will
likely evaporate.
The FBP of gasoline is usually controlled around two factors. Reformers produce
reformate, one of the important octane producers for the gasoline pool. Reformers convert Cg-
Cn cycloparaffms and alkyl-paraffins into alkylbenzenes (propyl-, isopropyl-benzene), which
have high blending octanes, but which also boil at about 400° F to 420° F. Other important
reactions take place in the reformer, which we will discuss in more detail in the reformer section.
The combustion pattern in current spark-ignited engines will efficiently burn only hydrocarbons
that boil at or below the referenced temperature. Gasoline is formulated around a fairly delicate
balance of light and heavy components. Depending on the several factors, a refiner may choose
or be asked to either raise or to lower the FBP of his gasoline. If the FBP is raised, it may be
possible to use more butane to makeup the RVP; if it is lowered, less butane can be added. It
should be clear that there are practical limits to either raising or lowering the FBP. If lowered
too far, little butane can be added, and regardless, the entire blend becomes relatively more
volatile and more difficult to control in an automobile fuel tank.
Even though we intend to discuss fluid catalytic cracking (FCC) later, we will mention
here that as a result of "cracking" (mostly FCC) most gasoline currently sold in the U.S. contains
at least some olefms (hydrocarbon compounds which have at least one double-bond between two
carbons). These compounds are quite unstable and over even short time periods tend to
polymerize into long-chained, highly branched compounds commonly referred to as "gums."
Olefms are a particular problem around the injector nozzles of fuel-injected engines. If
detergents aren't added, deposits tend to build up and disrupt injector operation. Additives are
used that interrupt the oxidation of these compounds, including during combustion, and thus help
reduce gum deposits. Other additives are also used to enhance performance and provide
protection against oxidation and rust formation.
With regard to gasoline as a blended, marketable liquid fuel, we describe it as a mix of
intermediate streams from a variety of refinery units. The manner in which an individual refinery
is configured and operated, including purchasing additional blendstocks from other refineries,
affects the final batch quality. Two refineries, even with similar configurations and similar crude
feeds, but operated differently produce gasolines with quite different chemical compositions.
6-112
-------
Final Regulatory Impact Analysis
Gasoline is exposed to a wide variety of mechanical, physical, and chemical environments. Thus
the properties must be balanced to give satisfactory engine performance over a very wide range
of operating conditions. In nearly every case, the composition of a gasoline batch sold in a
specific area of the country is the result of a variety of compromises among both automobile and
fuel manufacturers.
Each batch or blend is comprised of a unique distribution of compounds, mostly
hydrocarbons, which when mixed properly achieve the performance-based requirements for
commercial gasoline. It would not be unusual to find that as many as 14, or more, different
blendstocks may be available at a single complex refinery; a few of these are: light straight run
(LSR), isomerate, reformate, cracked light and heavy gasoline, hydrocracked gasoline, polymer
gasoline (cat poly gasoline), alkylate, n-butane, and perhaps other additives in minor amounts.
The percentages of these stocks usually fluctuate, up and down, in each blend; from time-to-
time, for a variety of reasons, a component may not be used at all. Gasoline and the stocks from
which it is composed are sometimes referred as "the gasoline pool." We also note that multiple
units produce blendstocks of a similar type. For example, three different reformers usually
produce reformate with slightly different properties. Several of the large, complex refineries
have several units in multiples. The overall variety of blend stocks provides refiners with a
multitude of options for producing gasoline that meets ASTM and performance-based
requirements.
Gasoline with ethanol is not shipped by pipeline but is splash-blended at the terminal as
the gasoline is loaded onto a truck for delivery to an end-user. This makes it necessary for
refiners to produce a low-vapor pressure gasoline component or blendstock which can be
shipped via pipeline, into which the ethanol can be blended. The vapor pressure of the final mix
must meet local RVP requirements.
All gasolines are not created equal, because, as we mentioned, gasoline is formulated
according to performance- and not compositional-based specs; few if any gasolines, including
batches from within the same refinery, end up having the same chemical composition. The
'recipe' for blending a specific gasoline grade at any given refinery depends upon several factors
including, (1) inventories of the various blendstocks, (2) the operating status of the various
refining units, (3) the specific regulatory requirements for the intended market, and, of course,
(4) maximizing profit. Most modern refineries have engineers, economists, and marketers that
continually run linear programs (LP) using input from several sources, including lab, operations,
and inventory data, gathered from over the entire refinery, in real-time. Blending can be
automated and almost automatically self-adjust, as in-line monitors and other data-gathering
devices provide continuous feedback on product properties and unit production rates. As crude
and product supplies and costs shift up and down, along with market effects and processing
costs, LP operators are able to make adjustments to blending recipes, as often as from batch to
batch.
While some blending (e.g., addition of some oxygenates) may occur at the final
distribution terminal, the majority of a gasoline's properties are achieved through the blending
that occurs within the refinery, although many gasoline service stations blend regular and
premium gasoline to produce mid-grade at the pump. Though it may be obvious, we,
6-113
-------
Final Regulatory Impact Analysis
nevertheless, point out that such an operation means refiners and shippers needn't ship a third
grade of gasoline.
6A.3.2 Octane
Historical Context
Much of where we are today with regard to how hydrocarbon fuels, including those
which contain benzene, and the internal combustion engine have come to affect the environment,
has to do with the somewhat parallel development and eventual convergence of several
discoveries, inventions, and wars that occurred over an approximately 150-year span of recent
history. We believe a brief outline of that history will provide a helpful context for the
discussion that follows.
As has often happened in history, the discovery or invention of one thing has lead to the
invention, discovery, or new use of something else. As is likewise often the case, the demand or
supply for one or another of these "things" causes an ebb and flow in the supply and demand of
the other. Such was very much the case with crude oil and its many derivatives, such as
gasoline, diesel, and jet fuel and the internal combustion engine and the turbine or jet engine.
Crude oil and a few of its derivatives have been used in many parts of the world for centuries.
On the other hand, the internal combustion engine, by historical standards, is a fairly recent
invention.
By the early 1880's researchers and inventors eventually determined that internal
combustion engines "knocked" or "pinged" less when fired with gasoline produced from certain
varieties of crude oil than with that derived from others, but no one knew exactly why.
Eventually, they learned that, for a specific engine compression-ratio, gasoline produced
from certain varieties of crude oil knocked less than gasoline derived from others. According to
our current knowledge regarding the naturally occurring gasoline components that boost octane,
we suspect that one reason for the differences may have been that the "anti-knock" gasoline had a
higher concentration of branched-chain hydrocarbons in the C5 - C9 range. It is also possible
that the fuel contained some concentration of natural occurring aromatics. Since "poorly"
processed natural gasoline made up most of the available supply (although some volume was
recovered from natural gas wells), engine and auto manufacturers were forced to limit the
effective compression ratio and therefore the horsepower of their engines.
It was evident, early on, that compression-ratio and horsepower were related. For
example, an early (1901) 3-cylinder engine had a compression ratio of 2 to 1. It had only six to
eight horsepower and a top speed of about 20 miles per hour. Within eight or nine years, Henry
Ford's model T engine had a compression ratio of about 4.5 to 1 and at 20 horsepower was
capable of speeds above 30 miles per hour. These engines began to "knock" or "ping" at about
this compression-ratio using the fuel available at the time. As demand grew, the supply of
usable gasoline gradually became limited and its quality decreased. As fuel supplies worsened,
engine manufacturers tried to adjust, until for example, in 1916, the Model T engine's
compression-ratio had been reduced to 3.8 to one. Some chemicals, including benzene and
alcohol, which allowed higher compression ratios without engine knock, were widely used in
6-114
-------
Final Regulatory Impact Analysis
high performance racing engines of the era. It was through race-track testing (much the same as
happens today with race cars and developments in the auto/fuels industry) that benzene and other
aromatics came into common use, if not as single component fuels, certainly, as additives.
Octane Number
Until "octane number" was established, the only practical way to determine whether a
fuel would ping in an engine was to fire it in the engine. If the compression ratio of the engine
was already set, the only way to eliminate the ping was to continue trying various fuels or adding
chemicals such as benzene, toluene, alcohol, or whatever was available until the pinging
stopped. It was possible to set the compression ratio of an engine to match the available fuel, but
eventually that fuel would run out. During this early period, when little was really known about
gasoline, many attempts were made to determine which component or components were
responsible for reducing or eliminating pre-ignition ping. Neither then, nor since then, has
anyone been able to clearly explain "why" one chemical species helps reduce or eliminate ping
while a different species not only does not help, it may even exacerbate the problem. Nor has
anyone been able to produce a single component, full-purpose gasoline. We discussed earlier
that gasoline has been formulated according to performance criteria: made from components
light enough to readily ignite, even in cold conditions; with others heavy enough to not require
pressurized containment and to provide some energy density.
Eventually, a mechanism was deduced which helped explain how, in a particular engine
at a specified compression ratio, one gasoline knocked or pinged while another did not. Ideally,
a carefully timed spark ignites an air/fuel mixture, injected above the piston of a spark-ignited
engine, just as the piston compression stroke begins to increase the pressure, temperature, and
density of the mixture. A flame front, likewise ideally, should spread out somewhat smoothly
and uniformly across the piston-face from the point of the spark, to consume what remains of the
unburned mixture. Further, and again ideally, the gaseous products of combustion expand and
produce a gradually increasing "push" against the piston until all the fuel is consumed as the
piston reaches the top of the compression stroke and then begins its power stroke. To return to
the instant the spark fires and as the compression stroke continues, radiant heat from the burning
fuel rapidly raises the temperature of the unburned fuel. Additionally, as the flame front spreads
across the piston, the hot combustion gases expand at an increasing rate and tend to compress the
unburned part of the air-fuel mixture, further increasing its density and raising its temperature. If
the unburned air-fuel mixture is heated beyond its ignition temperature before the piston reaches
its proper position it "autoignites," instantaneously and explosively. When this happens it causes
a pressure wave to interfere with the ideal or at least more desirable pressure wave in the
cylinder. This wave-interaction generates a wildly fluctuating, third pressure wave. The
combination of these wildly interacting, fluctuating waves is responsible for the knocking or
pinging sound. This violent mistimed release of energy and the subsequent abnormal pressure
waves can be quite destructive and may shorten the life of the engine. (We note again, that while
it's helpful to understand how or why an engine knocks, we still don't know why some
chemicals reduce knock and others don't.)
It gradually became clear, as mentioned previously, that some types of chemicals reduced
pre-ignition ping. That is, that Cs to Ci2 branched paraffins contribute high octane blending
6-115
-------
Final Regulatory Impact Analysis
values; straight-chain paraffins have very low numbers. We also know that aromatics, such as
benzene, toluene, mixed xylenes, and other alkylbenzenes have high octane blending values.
An interesting phenomenon presents itself when gasoline octane is compared to diesel
cetane. We are not making a full-on technical comparison, but would like to merely point out
the following, as a matter of some interest. Aromatics, as a general rule improve the octane of
gasoline; straight-chained paraffins are poor octane producers. On the other hand, aromatics
reduce diesel cetane, while paraffins improve cetane number. The interesting part of the
comparison is that diesel engines are compression-ignited engines and compression
(compression ratio) is very much involved in pre-ignition ping or knock, especially if aromatic
content is low and paraffin content is high. A rather simplistic explanation seems to be that
paraffins promote compression ignition. This is not a conclusion; merely a comment. (See our
discussion, above, of the combustion process in a spark-ignited engine.)
To select a way of rating the propensity of a particular gasoline batch to knock, the
Cooperative Fuel Research Committee (CFRC) was set up in 1927 made up of representatives
from the American Petroleum Institute, the American Manufacturers Assn., the National Bureau
of Standards, and the Society of Automotive Engineers. A single-cylinder, variable compression-
ratio engine was built and fuel samples were prepared of various pure hydrocarbons, including
normal heptane distilled from the sap of the Jeffrey Pine. This engine or perhaps more precisely
the variable compression-ratio technology incorporated into it, allowed researchers to fire
mixtures of pure hydrocarbons and at the same time vary the engine compression-ratio to
determine the compression-ratio at which a particular fuel or fuel mixture would knock.
Likewise, the engine could be used to determine which fuel, from among a variety of
formulations, would not knock or ping at a specified compression-ratio.
In 1929, as part of the effort to standardize fuel quality, a proposal came before the
CFRC to actually use a variable compression-ratio engine to rate the ignition characteristics of
various gasolines. Although a few committee members were concerned that such an engine
would be far too complicated for routine use, by 1931 a prototype was built and displayed at a
meeting of the American Petroleum Institute. Eventually the skeptics were persuaded and
thousands of the engines were subsequently built, many of which continue to be in use.
"Octane number" eventually became the numerical measure by which the ignition
characteristics of a fuel would be defined. It is a unit-less figure that represents the resistance of
gasoline to autoignite when exposed to the heat and pressure of a combustion chamber in an
internal-combustion engine. Such premature detonation is indicated by the knocking or pinging
noises as discussed above. Eventually, the industry agreed to recognize the octane number
determined by comparing the performance of a test gasoline with the performance of a mixture
of iso-octane (2, 2, 4 trimethyl pentane) and normal heptane as a valid measure of a gasoline's
resistance to autoignition. The octane number is, simply, the percentage of iso-octane in a
mixture whose performance is the same as that of the gasoline being tested. For example, the
gasoline is given an 80 octane rating, if the test gasoline performs the same as a mixture of 80%
2, 2, 4, trimethyl pentane and 20% normal heptane. Straight-line extrapolation is used to
determine octane numbers higher than 100.
6-116
-------
Final Regulatory Impact Analysis
The CFRC subsequently determined that several tests would be required in order to
provide an octane rating that was useful over the entire range of potential operating conditions.
Around 1926, a test using an engine, similar to the one described above, was developed and
designated: Motor Octane Number (MON). A similar, but improved method, Research Octane
Number (RON) was developed in the late 1930's. Subsequently, two methods were developed
and recognized by the American Society of Testing Materials (ASTM): the Motor Method or
MON (ASTM D357) and the Research Method or RON (ASTM D908). The results of the two
test methods vary from gasoline to gasoline.
Currently, the RON is determined by a method that measures fuel antiknock level in a
single-cylinder engine under mild operating conditions; namely, at a moderate inlet mixture
temperature and a low engine speed. RON tends to indicate fuel antiknock performance in
engines at wide-open throttle and low-to-medium engine speeds. Generally, a gasoline's
performance under high loads and at high speeds is reflected in the MON, while its performance
under lighter loads and at lower speeds is reflected in the RON results.
MON is determined by a method that measures fuel antiknock level in a single-cylinder
engine under more severe operating conditions than those employed in the RON method;
namely, at higher inlet mixture temperature and higher engine speed. It indicates fuel antiknock
performance in engines operating at wide-open throttle and high engine speeds. Also, Motor
octane number tends to indicate fuel antiknock performance under part-throttle, road-load
conditions.
Three octane numbers are currently in use in the United States. The MON and RON
numbers are determined, as described above. Usually the RON is higher than the MON. The
third octane number is an average of the MON and RON numbers, (R+M)/2. By definition, this
is the octane rating of a gasoline that can be legally sold to the public and by federal mandate
must be clearly posted on all pumps that dispense gasoline to the public. Accordingly, regular,
unleaded gasoline has an octane number of about 87 (R+M)/2, while premium unleaded gasoline
is rated at about 93 (R+M)/2. In other parts of the country, usually in higher elevations, regular
unleaded may be 85 (R+M)/2 and premium 91 or 92 (R+M)/2.
Octane requirements can change with altitude, air temperature, and humidity, depending
on a vehicle's control system. Newer vehicles have sensors to measure and computers, to adjust
for such changes in ambient conditions. Regardless of changes in ambient conditions, these
vehicles are designed to use the same octane rated gasoline at all ambient operating conditions.
This new technology began to be used extensively in 1984. This technology, while constantly
evolving and improving, is used on almost all new vehicles. The octane requirements of an older
vehicles decrease as altitude increases. One of the problems of increasing altitude is that the
decreased air pressure doesn't provide adequate oxygen in the air/fuel mixture.
We mention here that fuel with antiknock ratings higher than required for knock-free
operation, do not improve engine performance. On the other hand, as we mentioned previously,
pre-ignition knock can damage an engine.
6A.4 Kerosene and Diesel
6-117
-------
Final Regulatory Impact Analysis
This information is provided mainly to complete our discussion of the crude fractionation
column. The first or upper side draw on the crude column usually produces kerosene. If the
refinery doesn't have a preflash, the overhead will essentially be LSR for isorn feed while the
first side draw will then be heavy straight-run, HSR. Whereas in the past the Air Force used
naphtha based JP-4 turbine fuel, the kerosene based fuel JP-8 is now being used. As such, some
refiners may be fortunate enough to produce some volume of straight-run JP-8 from this draw.
Regardless, the stream is steam stripped to set the vapor pressure, cooled, and sent to storage to
be used in blends to produce a variety of distillate range fuels, including possibly JP-8.
The diesel is drawn from the tower several trays below the kerosene draw. Diesel is used
in a wide variety of ways including to power highway vehicles, construction and mining
equipment, and locomotive and marine engines; it is also use to generate electricity and to heat
homes in several areas of the U.S. Nowadays, most kerosene and diesel is hydrotreated. High
sulfur diesel can be used to heat homes and aviation turbine fuel may have sulfur up a
concentration of about 0.5 wt. %. It is common practice in colder regions of the country for
truckers to mix some volume of kerosene into their diesel to improve his diesel's cold flow
properties during winter months. Prior to ultra-low sulfur diesel (ULSD), common straight-run
kerosene was used for this purpose, since the kerosene sulfur content was usually not so high as
to cause sulfur compliance problems for the diesel. However, as a result of the recent ULSD
rules, refiners may need to hydrotreat or desulfurize more, if not most, of their kerosene for this
market. Consequently, many refiners will likely hydrotreat the combined kerosene/diesel stream
and re-separate them where the market justifies it. We recognize that there may be other ways of
handling this problem.
6-118
-------
Final Regulatory Impact Analysis
References for Chapter 6
1 Nelson, W.L., Petroleum Refinery Engineering, 4th Ed. pp. 469-527 for further
discussion.
2 UOP, www.uop.com/refining/1061_l.html, definition included in Gasoline
Desulfurization discussion.
3 Gary, James H., Handwerk, Glenn E., Petroleum Refining Technology and Economics,
4th Ed., p. 204 for further discussion.
4 Maples, Robert. E., Petroleum Refinery Process Economics, 2nd Ed., oo, 263-281;
Gary, James H., Handwerk, Glenn E., Petroleum Refining Technology andEconcomics, 4th Ed.,
pp 189-203. See both for further discussion.
5Li quid-Li quid Extraction,, Chemical and Process Technology Encyclopedia, Considine,
Douglas M. Editor-in-Chief, 1974, p. 1024.
6 Fluid Catalytic Cracking,, Chemical and Process Technology Encyclopedia; Considine,
Douglas M. Editor-in-Chief, 1974, pp. 505-509.
7Sterba, Melvin J., Alkylation. Chemical and Process Technology Encyclopedia;
Considine, Douglas M. Editor-in-Chief, 1974, pp. 70-75.
8 Defined in 40 CFR 80.41
9 North American Catalysis Society (NACS), 50 Years of Catalysis, see year 1949;
webpage: www.nacatsoc.org/history.asp?HistoryID= 1
10 Gary, James H., Handwerk, Glenn E., Petroleum Refining Technology and Economics,
4th Ed., p. 189-204; for further discussion.
11 Maples, Robert E., Petroleum Refinery Process Economics, 2nd Ed., pp. 263-267 for
further discussion.
12 Gary, James H., Handwerk, Glenn E., Petroleum Refining Technology and Economics,
4th Ed., p. 189-204; for further discussion.
13Sterba, Melvin J., Isomerization. Chemical and Process Technology Encyclopedia,
Considine, Douglas M. Editor-in-Chief, 1974, pp. 662-665.
14
15
www.uop.com/objects/Bensat.pdf
Sullivan, Dana K., UOP Bensat Process, Handbook of Petroleum Refining Processes,
rd
Robert B. Myers, Ed., 3ra Ed., p. 9.3 - 9.6.
16 www.cdtech.com/techProfilesPDF/Selective_Hydrogenation_Benz_Cyclohexane-
CDHydro.pdf
17 http://www.cdtech.com/techProfilesPDF/Selective_Hydrogenation_Benzene_
Reformate- CDHydro.pdf
18 Gentry, Joseph, Kumar, Sam, Wright-Wytcherley, Randi, GTC Technology, Houston
Texas, Extractive Distillation Applied, Paper No. 70e, AIChE Spring Meeting, New Orleans,
LA, April 2003.
19 Perry, Robert H., Chilton, Cecil H., Kirkpatrick, Sidney D., Extractive and Azeotropic
Distillation, Chemical Engineers Handbook. 4th Ed., 1963, pp. 13.46-13.47.
20 In Depth Look at Extractive Distillation, Online Chemical Engineering Information,
Cheresources.com; webpage: www.cheresources.com/extrdist.shtml.
6-119
-------
Final Regulatory Impact Analysis
21 Kolb, Jeff, Abt Associates, Task 3 - Benzene Recovery from Light Reformate via
Cracking and Distillation, Memorandum to EPA under contract WA 0-01, EP-C-06-094, January
11,2007.
22 Personal conversation with David Netzer, October 2006.
23 Kolb, Jeff; Benzene Content of Light Gasoline Blendstocks: Information Provided by
Refiners, Memorandum provided to EPA under contract WA 0-01, EP-C-06-094, November 6,
2006.
24 Kolb, Jeff; Benzene Control for Light Refinery Streams, Memorandum provided to
EPA under contract WA 0-01, EP-C-06-094, November 21, 2006.
25 Since this is not a technical paper on distillation, we have chosen not to discuss the finer points of
IBP/FBP overlap or separation.
26 Kolb, Jeff, Abt Associates, Task 4 - Control of Benzene in FCC Naphtha,
Memorandum to EPA under contract WA 0-01, EP-C-06-094, January 12, 2007.
T7
Key worth, Donald A., FCC Applications Manager et al, Akzo Catalysts Inc,
Controlling Benzene Yield from the FCCU, technical paper AM-93-49 presented at the 1993
National Petroleum Refiners Association Annual Meeting.
28 Moncrief, P. and Ragsdale, R., "Can the U.S. E&C Industry Meet the EPA's Low
Sulfur Timetable?" NPRA 2000 Annual Meeting, March 26-28, 2000, Paper No. AM-00-57.
29 40 CFR 80 Subpart D
30 The Complex Model is defined in 40 CFR 80.45
31 40 CFR 80 Subpart E
32 40 CFR 80 Subpart J
33 See Chapter 2 of the Draft RIA for the Renewable Fuels Standard rulemaking, EPA-
420-D-06-008
34 Energy Policy Act of 2005, Section 1504
35 Energy Policy Act of 2005, Section 1501
36 See Chapter 2 of the Draft RIA for the Renewable Fuels Standard rulemaking, EPA-
420-D-06-008
See www.eia.doe.gov
38 Final report to Alliance of Automobile Manufacturers prepared by Richard Gunst is
available in the docket.
39 See section 2.2.2 of this RIA.
40 Clean Air Act §2 ll(k)
41 Clean Air Act §21 l(k); Code of Federal Regulations, Title 40, Parts 80.4 l(e) and
80.41(f).
42 See section 2.1.4.6 of the Draft RIA for the Renewable Fuels Standard rulemaking,
EPA-420-D-06-008
43 Code of Federal Regulations Parts 80.4 l(e) and 80.4 l(f).
6-120
-------
Final Regulatory Impact Analysis
Chapter 7: Table of Contents
Chapter 7: Portable Fuel Container Feasibility and Test Procedures 2
7.1 Permeation Emissions 2
7.2 Permeation Emissions Controls 3
7.2.1 Sulfbnation 3
7.2.2 Fluorination 7
7.2.3 Barrier Platelets 11
7.2.4 Multi-Layer Construction 13
7.3 Diurnal Emissions 14
7.4 Testing Procedures 15
7.4.1 Temperature Profile, Length of Test, Fill Level 16
7.4.2 Test Fuel 16
7.4.3 Preconditioning and Durability Testing 17
7.4.3.1 Preconditioning 17
7.4.3.2 Durability Testing 17
7.4.4 Reference Container 19
7-1
-------
Final Regulatory Impact Analysis
Chapter 7: Portable Fuel Container Feasibility and Test Procedures
Section 183 (e) of the Clean Air Act provides statutory criteria that EPA must evaluate in
determining standards for consumer products. The standards must reflect "best available
controls" as defined by section 183 (e)(3)(A). Determination of the "best available controls"
requires EPA to determine the degree of reduction achievable through use of the most effective
control measures (which extend to chemical reformulation, and product substitution) after
considering technological and economic feasibility, as well as health, energy, and environmental
impacts. Chapters 1 through 3 discuss the environmental and health impacts of portable fuel
container (PFC) emissions. Chapter 10 discusses the economic feasibility of PFC controls and
the fuel savings associated with controlling PFC emissions. This chapter presents the
technological feasibility of controlling emissions from PFCs. All of these analyses and
information form the basis of EPA's belief that the evaporative emission standards reflect the
"best available controls" accounting for all the above factors.
This chapter presents available data on baseline emissions and on emission reductions
achieved through the application of emission control technology. In addition, this chapter
provides a description of the test procedures for determining evaporative emissions.
Evaporative emissions from PFCs containing gasoline can be very high.A This is largely
because PFCs are often left open and vent to the atmosphere and because materials used in the
construction of the plastic PFCs generally have high permeation rates. Evaporative emissions
can be grouped into three main categories:
DIURNAL: Gasoline evaporation increases as the temperature rises during the day,
heating the PFC and venting gasoline vapors.
PERMEATION: Gasoline molecules can saturate plastic PFCs, resulting in a relatively
constant rate of emissions as the fuel continues to permeate through the walls of the PFC.
REFUELING: Gasoline vapors are always present in typical containers. These vapors are
forced out when the container is filled with liquid fuel.
The use of PFCs also results in losses through spillage, both during transportation and
usage of the cans to refill vehicles and equipment.
7.1 Permeation Emissions
The California Air Resources Board (ARB) investigated permeation rates from PFCs
with no emissions controls.1>2 The ARB data is compiled in several data reports on their web
site and is included in our docket. Table 7.1-1 presents a summary of this data which was
A Diesel and kerosene fuels have very low volatility levels and therefore much lower evaporative emissions
compared to gasoline.
7-2
-------
Final Regulatory Impact Analysis
collected using the ARB Test Method 513.3 Although the temperature in the ARB testing is
cycled from 65 - 105° F with 7 pound per square inch (psi) Reid Vapor Pressure (RVP) fuel, the
results would be similar if the data were collected at the temperature range and fuel used by EPA
of 72-96° F with 9 psi RVP fuel. This is because the lower temperature and higher RVP
effectively offset one another. The average permeation emissions from uncontrolled containers
were 1.57 g/gallon/day.
Table 7.1-1. Permeation Rates for HOPE PFCs Tested by ARB
PFC Capacity
[gallons]
1.0
1.0
1.0
1.0
1.0
1.0
1.3
1.3
1.3
1.3
1.3
1.3
1.3
1.3
2.1
2.1
2.1
2.1
2.5
2.5
5.0
5.0
5.0
5.0
5.0
5.0
5.0
6.6
Permeation Loss
[g/gal/day]
1.63
1.63
1.51
0.80
0.75
0.75
0.50
0.49
0.51
0.52
0.51
0.51
1.51
1.52
1.88
1.95
1.91
1.78
1.46
1.09
0.89
0.62
0.99
1.39
1.46
1.41
1.47
1.09
7.2 Permeation Emissions Controls
7.2.1 Sulfonation
The California Air Resources Board (ARB) collected test data on permeation rates from
sulfonated PFCs using California certification fuel.4 The results show that sulfonation can be
used to achieve significant reductions in permeation from plastic fuel containers. This data was
7-3
-------
Final Regulatory Impact Analysis
collected using a diurnal cycle from 65 - 105° F. The average emission rate for the 32
sulfonated PFCs was 0.35 g/gal/day; however, there was a wide range in effectiveness of the
sulfonation process for these PFCs. Some of the data outliers were actually higher than baseline
emissions. This was likely due to leaks in the PFCs which would result in large emission
increases due to pressure built up with temperature variation over the diurnal cycle. Removing
these five outliers, the average permeation rate is 0.17 g/gal/day with a minimum of 0.01
g/gal/day and a maximum of 0.64 g/gal/day. This data suggests that more than a 90% reduction
in permeation is possible through sulfonation. This data is presented in Table 7.2-1.
Table 7.2-1. Permeation Rates for Sulfonated
Plastic PFCs Tested by ARE
PFC Capacity
[gallons]
1
1
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
5
5
5
5
5
5
5
Permeation Loss
[g/gal/day]
0.05
0.05
0.05
0.06
0.06
0.06
0.08
0.12
0.14
1.23
1.47
1.87
0.02
0.02
0.48
0.54
1.21
0.03
0.08
0.32
0.38
0.42
0.52
0.64
0.80
0.01
0.04
0.05
0.06
0.11
0.13
0.15
7-4
-------
Final Regulatory Impact Analysis
Variation can occur in the effectiveness of this surface treatment if the sulfonation
process is not properly matched to the plastic and additives used in the container material. For
instance, if the sulfonater does not know what ultraviolet (UV) inhibitors or plasticizers are used,
they cannot maximize the effectiveness of their process. Earlier data collected by ARB showed
consistently high emissions from sulfonated fuel containers; however, ARB and the treatment
manufacturers agree that this was due to inexperience with treating fuel containers and that these
issues have since been largely resolved.5
ARB also investigated the effect of fuel slosh on the durability of sulfonated surfaces.
Three half-gallon fuel tanks used on small SI equipment fuel tanks were sulfonated and tested for
permeation before and after being rocked with fuel in them 1.2 million times.6'7 These fuel
tanks were blow-molded high density polyethylene (HDPE) tanks used in a number of small SI
applications including pressure washers, generators, snowblowers, and tillers. The results of the
testing show that an 85% reduction in permeation was achieved on average even after the slosh
testing was performed. Table 7.2-2 presents these results which were recorded in units of
g/m2/day. The baseline level for Set #1 is an approximation based on testing of similar fuel tanks,
while the baseline level for Set #2 is based on testing of those tanks.
The sulfonater was not aware of the materials used in the fuel tanks sulfonated for the
slosh testing. After the tests were performed, the sulfonater was able to get some information on
the chemical make up of the fuel tanks and how it might affect the sulfonation process. For
example, the UV inhibitor used in some of the fuel tanks is known as HALS. HALS also
reduces the effectiveness of the sulfonation process. Two other UV inhibitors, known as carbon
black and adsorber UV, are also used in similar fuel tank applications. These UV inhibitors cost
about the same as HALS, but have the benefit of not interfering with the sulfonation process.
The sulfonater claimed that if HALS were not used in the fuel tanks, a 97% reduction in
permeation would have been seen.8 To confirm this, one manufacturer tested a sulfonated tank
similar to those in Set #2 except that carbon black, rather than HALS, was used as the UV
O Q
inhibitor. This fuel tank showed a permeation rate of 0.88 g/m /day at 40°C which was less
than half of what the CARB testing showed on their constant temperature test at 40°C.10 A list
of resins and additives that are compatible with the sulfonation process is included in the
docket.11'12
7-5
-------
Final Regulatory Impact Analysis
Table 7.2-2. Permeation Rates for Sulfonated Fuel Tanks
with Slosh Testing by ARE Over a 18-41°C Diurnal
Technology Configuration
Set #1 Approximate Baseline
Set #1 Sulfonated
Set #1 Sulfonated & Sloshed
Set #2 Average Baseline
Set #2 Sulfonated
Set #2 Sulfonated & Sloshed
Units
g/m2/day
g/m2/day
% reduction
g/m2/day
% reduction
g/m2/day
g/m2/day
% reduction
g/m2/day
% reduction
Tankl
10.4
0.73
93%
1.04
90%
12.1
1.57
87%
2.09
83%
Tank 2
10.4
0.82
92%
1.17
89%
12.1
1.67
86%
2.16
82%
Tank3
10.4
1.78
83%
2.49
76%
12.1
1.29
89%
1.70
86%
Average
10.4
1.11
89%
1.57
85%
12.1
1.51
88%
1.98
84%
About a year and a half after the California ARB tested the Set #2 fuel tanks, we
performed permeation tests on these fuel tanks. During the intervening period, the fuel tanks
remained sealed with California certification fuel in them. We drained the fuel tanks and filled
them with fresh California certification fuel. We then measured the permeation rate at 29°C.
Because this is roughly the average temperature of the California variable temperature test,
similar permeation rates would be expected. The untreated fuel tanks showed slightly lower
permeation over the constant temperature test as compared to the ARB test. This difference was
likely due to the difference in the temperature used for the testing. However, the Sulfonated fuel
tanks showed an increase in permeation as compared to the ARB test. This increase in
permeation appears to be the result of the 1.5 year additional fuel soak. After this long soak, the
average permeation reduction changed from 84% to 78%. Table 7.2-3 presents this comparison.
Table 7.2-3. Permeation Rates [g/m2/day] for Sulfonated Fuel Tanks Tested by
ARB and EPA on CA Certification Gasoline with a \1A Year Fuel Soak Differential
Technology Configuration
Baseline, CARB testing
Baseline, EPA testing after
1.5 year additional fuel soak
Sulfonated, CARB testing
Sulfonated, EPA testing after
1.5 year additional fuel soak
Temperature
18-41°C
29°C
% change
18-41°C
29°C
% reduction
from EPA
baseline
Tankl
12.1
11.5
-5%
2.09
2.48
78%
Tank 2
12.1
11.4
-6%
2.16
2.73
76%
Tank3
12.1
11.2
-7%
1.70
2.24
80%
Average
12.1
11.4
-6%
1.98
2.5
78%
7-6
-------
Final Regulatory Impact Analysis
After the above testing, we drained the fuel tanks and filled them with certification
gasoline splash-blended with 10% ethanol (E10). We then soaked the fuel tanks for 20 weeks to
precondition them on this fuel. Following the preconditioning, we tested these fuel tanks for
permeation at 29°C (85°F). Table 7.2-4 presents these emission results compared to the
emission results for three baseline tanks (untreated) that were subject to the same
preconditioning. Percent reductions are presented based on the difference between the
sulfonated fuel tanks and the average results of the three untreated fuel tanks.
Table 7.2-4. Permeation Rates for Sulfonated Fuel Tanks on E10 Fuel at 29°C
Technology Configuration
Baseline (untreated)
Sulfonated
Units
g/m2/day
g/m2/day
% reduction
Tankl
13.9
3.91
72%
Tank 2
13.7
4.22
70%
Tank3
14.4
2.92
79%
Average
14.0
3.69
74%
One study looked at the effect of alcohol in the fuel on permeation rates from sulfonated
fuel tanks.13 In this study, the fuel tanks were tested with both gasoline and various methanol
blends. No significant increase in permeation due to methanol in the fuel was observed.
7.2.2 Fluorination
Another barrier treatment process is known as fluorination. The fluorination process
causes a chemical reaction where exposed hydrogen atoms are replaced by larger fluorine atoms
which form a barrier on surface of the container. In this process, PFCs are generally processed
post production by stacking them in a steel container. The container is then voided of air and
flooded with fluorine gas. By pulling a vacuum in the container, the fluorine gas is forced into
every crevice in the fuel containers. As a result of this process, both the inside and outside
surfaces of the PFCs are treated. As an alternative, containers can be fluorinated on-line by
exposing the inside surface of the PFC to fluorine during the blow molding process. However,
this method may not prove as effective as off-line fluorination which treats the inside and outside
surfaces.
We tested one fluorinated HDPE fuel tank which we bought off the shelf and sent to a
fluorinater for barrier treatment. The fuel tank type used was a 6-gallon portable marine fuel
tank. The fuel tank was soaked for 20 weeks with certification gasoline prior to testing. We
measured a permeation rate of 0.05 g/gal/day (0.56 g/m2/day), which represents more than a 95
percent reduction from baseline. We then began soaking this fuel tank on E10, subjected it to the
required pressure and slosh testing, and retested the fuel tank. The post-durability testing
showed a permeation rate of 0.6 g/gal/day (6.8 g/m2/day). As discussed below, we believe that
the impact of the durability testing on the effectiveness of fluorination can be minimized if the
fluorination process and material properties are matched properly. In addition, this fuel tank was
treated to a significantly lower level of fluorination than is now available. However, this data
supports the need for the durability testing requirements included in the program.
7-7
-------
Final Regulatory Impact Analysis
The California Air Resources Board (ARB) collected test data on permeation rates from
fluorinated fuel containers using California certification fuel.14'15 The results show that
fluorination can be used to achieve significant reductions in permeation from plastic fuel
containers. This data was collected using a diurnal cycle from 65 - 105°F. For the highest level
of fluorination, the average permeation rate was 0.04 g/gal/day, which represents a 95 percent
reduction from baseline. Earlier data collected by ARB showed consistently high emissions
from fluorinated PFCs; however, ARB and the treatment manufacturers agree that this was due
to inexperience with treating fuel containers and that these issues have since been largely
resolved.16 The ARB data is presented in Table 7.2-5.
Table 7.2-5. Permeation Rates for Fluorinated
Plastic PFCs Tested by ARB
Barrier Treatment*
Level 4
(average =0.09 g/gal/day)
Level 5
(average =0.07 g/gal/day)
SPAL
(average =0.04 g/gal/day)
PFC Capacity
[gallons]
1
1
1
5
5
5
1
1
1
1
1
1
1
1
1
2.5
2.5
2.5
2.5
2.5
5
5
5
5
5
5
Permeation Loss
[g/gal/day]
0.05
0.05
0.06
0.11
0.11
0.15
0.03
0.04
0.05
0.05
0.07
0.08
0.11
0.11
0.12
0.04
0.04
0.05
0.07
0.07
0.05
0.10
0.11
0.04
0.04
0.04
* designations used in ARB report; shown in order of increasing treatment
7-8
-------
Final Regulatory Impact Analysis
All of the data on fluorinated PFCs presented above were based on PFCs fluorinated by
the same company. Available data from another company that fluorinates fuel containers shows
a 98 percent reduction in gasoline permeation through a HDPE fuel tank due to fluorination.17
ARB investigated the effect of fuel slosh on the durability of fluorinated surfaces. Two
sets of three fluorinated fuel tanks were tested for permeation before and after being sloshed with
fuel in them 1.2 million times.18'19 These fuel tanks were 0.5 gallon, blow-molded HDPE tanks
used in a number of small SI applications including pressure washers, generators, snowblowers,
and tillers. The results of this testing show that an 80% reduction in permeation was achieved on
average even after the slosh testing was performed for Set #1. However, this data also showed a
99 percent reduction for Set #2. This shows the value of matching the barrier treatment process
to the fuel tank material. Table 7.2-6 presents these results, which were recorded in units of
g/m2/day. The baseline level for Set #1 is an approximation based on testing of similar fuel tanks,
while the baseline for Set #2 is based on testing of those tanks.
Table 7.2-6. Permeation Rates for Fluorinated Fuel Tanks
with Slosh Testing by ARB Over a 65-105° F Diurnal
Technology Configuration
Set #1 Approximate Baseline
Set #1 Fluorinated
Set #1 Fluorinated & Sloshed
Set #2 Approximate Baseline
Set #2 Fluorinated
Set #2 Fluorinated & Sloshed
Units
g/m2/day
g/m2/day
% reduction
g/m2/day
% reduction
g/m2/day
g/m2/day
% reduction
g/m2/day
% reduction
Tankl
10.4
1.17
89%
2.38
77%
12.1
0.03
>99%
0.07
99%
Tank 2
10.4
1.58
85%
2.86
73%
12.1
0.00
>99%
0.11
99%
Tank3
10.4
0.47
96%
1.13
89%
12.1
0.00
>99%
0.05
>99%
Average
10.4
1.07
90%
2.12
80%
12.1
0.01
>99%
0.08
99%
About a year and a half after the California ARB tests on the Set #2 fuel tanks, we
performed permeation tests on these fuel tanks. During the intervening period, the fuel tanks
remained sealed with California certification fuel in them. We drained the fuel tanks and filled
them with fresh California certification fuel. We then measured the permeation rate at 29°C.
Because this is roughly the average temperature of the California variable temperature test,
similar permeation rates would be expected. The untreated fuel tanks showed slightly lower
permeation over the constant temperature test. This difference was likely due to the difference in
the temperature used for the testing. However, the fluorinated fuel tanks showed an increase in
permeation. This increase in permeation appears to be the result of the 1.5 year additional fuel
soak. Even after this long fuel soak, the fluorination achieves more than a 95% reduction in
permeation. Table 7.2-7 presents this comparison.
7-9
-------
Final Regulatory Impact Analysis
Table 7.2-7. Permeation Rates [g/m /day] for Fluorinated Fuel Tanks Tested by
ARE and EPA on CA Certification Gasoline with a \1A Year Fuel Soak Differential
Technology Configuration
Baseline, CARD testing
Baseline, EPA testing after 1.5
year additional fuel soak
Fluorinated, CARB testing
Fluorinated, EPA testing after
1.5 year additional fuel soak
Temperature
18-41°C
29°C
% change
18-41°C
29°C
% reduction
from EPA
baseline
Tankl
12.1
11.5
-5%
0.07
0.56
95%
Tank 2
12.1
11.4
-6%
0.11
0.62
95%
TankS
12.1
11.2
-7%
0.05
0.22
98%
Average
12.1
11.4
-6%
0.08
0.47
96%
After the above testing, we drained the fuel tanks and filled them with certification
gasoline splash-blended with 10% ethanol (E10). We then soaked the fuel tanks for 20 weeks to
precondition them on this fuel. Following the preconditioning, we tested these fuel tanks for
permeation at 29°C (85°F). Table 7.2-8 presents these emission results compared to the
emission results for three baseline tanks (untreated) that were subject to the same
preconditioning. Percent reductions are presented based on the difference between the
fluorinated fuel tanks and the average results of the three untreated fuel tanks. The slight
increase in permeation on the E10 fuel was similar for the baseline and fluorinated fuel tanks and
still resulted in reductions above 95 percent.
Table 7.2-8. Permeation Rates for Fluorinated Fuel Tanks on E10 Fuel at 29°C
Technology Configuration
Baseline (untreated)
Fluorinated
Units
g/m2/day
g/m2/day
% reduction
Tankl
13.9
0.43
97%
Tank 2
13.7
0.62
96%
Tank3
14.4
0.62
96%
Average
14.0
0.56
96%
Another study also looked at the effect of alcohol in the fuel on permeation rates from
fluorinated fuel tanks.20 In this study, the fuel tanks were tested with both gasoline and various
methanol blends. No significant increase in permeation due to methanol in the fuel was observed.
One automobile manufacturer used fluorination to reduce permeation on HDPE fuel
tanks to meet the LEV I vehicle standards. This manufacturer used similar or more stringent
requirements for fuel soak, durability, and testing than finalized today. At 40°C, this
manufacturer stated that they measured 0.15-0.2 g/day for fluorinated tanks compared to over 10
g/day for untreated HDPE fuel tanks.21
7-10
-------
Final Regulatory Impact Analysis
7.2.3 Barrier Platelets
Another approach for reducing permeation emissions is to blend a low permeable resin in
with the HDPE and extrude it with a single screw. The low permeability resin, typically ethylene
vinyl alcohol (EVOH) or nylon, creates non-continuous platelets in the HDPE fuel tank which
reduce permeation by creating long, tortuous pathways that the hydrocarbon molecules must
navigate to pass through the container walls. The trade name typically used for this permeation
control strategy is Selar® for nylon and Selar KB® for EVOH. Although the barrier is not
continuous, this strategy can still achieve greater than a 90 percent reduction in permeation of
gasoline. EVOH has much higher permeation resistance to alcohol than nylon; therefore, it
would be the preferred material to use for meeting our new standard, which is based on testing
with a 10 percent ethanol fuel.
We tested several portable PFCs and marine fuel tanks molded with low permeation non-
continuous barrier platelets. Six of the containers tested were constructed using nylon as the
barrier material. The remainder of the containers were constructed using EVOH as the barrier
material. The sixth container was tested on E10 (10% ethanol) to evaluate the effectiveness of
this material with alcohol blended fuel. The containers with the EVOH barrier were all tested on
E10.
Testing was performed after the containers had been filled with fuel and stored at room
temperature. We soaked the containers with gasoline for 22 weeks and the tanks with E10 for 37
weeks. The purpose of the soak period was to ensure that the fuel permeation rate had stabilized.
The containers were drained and then filled with fresh fuel prior to the permeation tests. We did
not run slosh and pressure tests on these containers. However, because the barrier platelets are
integrated in the can wall material, it is not likely that pressure or slosh testing would
significantly affect the performance of this technology.
Table 7.2-9 presents the results of the permeation testing on the containers with barrier
platelets. These test results show more than an 80 percent reduction for the nylon barrier tested
on gasoline. However, the nylon barrier does not perform as well when a fuel with a 10%
ethanol blend is used. Testing on a pair of 2 gallon containers with nylon barrier showed 80%
percent higher emissions when tested on E10 than on gasoline. We also tested PFCs that used
EVOH barrier platelets. EVOH has significantly better resistance to permeation on E10 fuel
than nylon. For the containers blended with 6% EVOH, we observed a permeation rate of about
0.08-0.09 g/gal/day on E10 fuel.
7-11
-------
Final Regulatory Impact Analysis
Table 7.2-9. Permeation Rates for Plastic Fuel Containers
with Barrier Platelets Tested by EPA at 29°C
Percent
Selar®*
Capacity
[gallons]
Test Fuel
Nylon barrier platelets
unknown* *
unknown* *
4%
4%
4%
4%
2
2
5
5.3
6.6
6.6
gasoline
E10
gasoline
gasoline
gasoline
gasoline
EVOH barrier platelets
2%
4%
4%
6%
6%
6.6
6.6
6.6
6.6
6.6
E10
E10
E10
E10
E10
Fuel Soak
[weeks]
40
40
22
22
22
22
37
37
37
37
37
g/gal/day
g/m2/day
0.54
0.99
0.35
0.11
0.15
0.14
-
4.1
1.2
1.6
1.5
0.23
0.14
0.15
0.08
0.09
3.0
1.9
2.0
1.4
1.4
*trade name for barrier platelet technology used in test program
** designed to meet California permeation requirement
Manufacturers raised a concern about whether or not a container using barrier platelets
would have a stabilized permeation rate after 20 weeks. In other words, manufacturers were
concerned that this technology may pass the test, but have a much higher permeation rate in-use.
We tested one of the 4% and 6% EVOH containers on E10 again after soaking for a total of 104
weeks (2 years). The measured permeation rates were 2.0 and 1.4 g/m2/day for the 4% and 6%
EVOH containers, respectively, which represents no significant changes in permeation from the
37 week tests. In contrast, we measured the 4% nylon tanks again after 61 weeks and measured
permeation rates of 2.8 and 2.7 g/m2/day, which represented about an 80-90% increase in
permeation compared to the 22 week tests.
The California ARB collected test data on permeation rates from PFCs molded with
Selar® low permeation non-continuous barrier platelets using California certification fuel. This
data was collected using a diurnal cycle from 65-105°F. The results show that this technology
can be used to achieve significant reductions in permeation from plastic fuel containers. This
test data showed that more than a 90 percent reduction in permeation is achievable through the
use of barrier platelets. However, all of this testing was performed on California certification
fuel, which does not include ethanol.
7-12
-------
Final Regulatory Impact Analysis
Table 7.2-10. Permeation Rates for PFCs
with Barrier Platelets Tested by ARE on California Fuel
Percent Selar®*
4%
(average =0.12 g/gal/day)
6%
(average =0.09 g/gal/day)
807
/O
(average =0.07 g/gal/day)
Container Capacity
[gallons]
5
5
5
5
5
6
6
5
5
5
5
5
5
6
6
5
5
6
6
Permeation Loss
[g/gal/day]
0.08
0.09
0.13
0.16
0.17
0.08
0.10
0.07
0.07
0.07
0.08
0.12
0.17
0.06
0.07
0.08
0.10
0.05
0.06
*trade name for barrier platelet technology used in test program
Table 7.2-11 presents permeation rates for HDPE and three Selar KB® blends when
tested at 60°C on xylene.22 Xylene is a component of gasoline and gives a rough indication of
the permeation rates on gasoline. This report also shows a reduction of 99% on naptha and 98%
on toluene for 8% Selar KB®.
Table 7.2-11. Xylene
Composition
100% HDPE
10%RB215/HDPE
10%RB300/HDPE
15%RB421/HDPE
Permeation, g mm/m2/day
285
0.4
3.5
0.8
% Reduction
99.9%
98.8%
99.7%
7.2.4 Multi-Layer Construction
7-13
-------
Final Regulatory Impact Analysis
PFCs may also be constructed out of multiple layers of materials, and some PFC
manufacturers have started using this technology. In this way, the low cost and structural
advantages of traditional materials can be utilized in conjunction with higher grade materials
which can provide effective permeation resistance.
Coextruded barrier technology has been long established for blow-molded automotive
fuel tanks. Data from one automobile manufacturer showed permeation rates of 0.01-0.03 g/day
for coextruded fuel tanks at 40°C on EPA certification fuel. They are using this technology to
meet LEV II vehicle standards. For comparison, they reported permeation rates of more than 10
g/day for standard HDPE fuel tanks.
23
Another study looks at the permeation rates, using ARB test procedures, through multi-
layer vehicle fuel tanks.24 The fuel tanks in this study were 6 layer coextruded plastic tanks with
EVOH as the barrier layer (3% of wall thickness). The outer layers were HDPE and two
adhesive layers were needed to bond the EVOH to the polyethylene. The sixth layer was made
of recycled polyethylene. The two test fuels were a 10 percent ethanol blend (CE10) and a 15
percent methanol blend (CM15). See Table 7.2-12.
Table 7.2-12. Permeation Results for a Coextruded Fuel Tank Over a 65-105°F Diurnal
Composition
100% HDPE (approximate)
3% EVOH, 10% ethanol (CE10)
3% EVOH, 15% methanol (CM15)
Permeation, g/day
6-8
0.2
0.3
% Reduction
97%
96%
7.3 Diurnal Emissions
The above sections discuss permeation emissions and permeation emissions control.
These emissions are part of the overall evaporative emissions, or diurnal emissions, from PFCs.
PFCs as a system also emit evaporative emissions from seals and spouts. PFCs have high
evaporative emissions when they are left open. In order to meet emissions standards,
manufacturers would use cans with spouts that automatically close and seal well around the
opening to the can where the spout attaches. Automatic closing spouts have been designed for
the California program. These spouts are typically manufactured with springs that close the cans
automatically when the cans are not being used to refill equipment. In addition, these cans vent
through the spouts, and the vents typically found on the back of the cans are removed. This is
important because open vents can be a significant source of evaporative emissions.
CARB conducted a feasibility study for their PFC standards and concluded that a 0.3
g/gal/day standard was feasible in the 2009 time-frame.25 CARB conducted testing of three
different PFCs designed to meet emissions standards. They were tested in two ways: with the
spout attached and with the spouts removed and the PFCs sealed. The results for the sealed cans
7-14
-------
Final Regulatory Impact Analysis
represent the amount of permeation emissions observed. This data was collected using a diurnal
cycle from 65-105°F with 7 RVP fuel. As noted above, the results would be similar if the data
were collected at the temperature range and fuel used by EPA of 72-96°F with 9 psi RVP fuel,
because the lower temperature and higher RVP offset one another. The PFCs with spout were
soaked for 160 days and the sealed cans were soaked for 174 days prior to testing. The results of
the testing are provided below in Table 7.3-1. The results show the average of three identical
cans per manufacturer. CARB did not identify the manufacturers or the permeation barriers used.
Table 7.3-1. Results of CARB Diurnal Testing (g/gal/day)
Manufacturer A
Manufacturer B
Manufacturer C
Sealed PFC
0.1
0.0
0.2
PFC w/ Spout
0.2
0.7
0.2
CARB indicated that the results from Manufacturer B increased because of one faulty
spout which significantly increased the average emissions. The results indicate that the 0.3
g/gal./day standard is feasible. The results also indicate that a faulty spout or seal around the
opening of the PFC would likely lead to emissions significantly above the standard.
Manufacturers would need to focus on controlling variability in their manufacturing process to
ensure spouts are durable and well matched to the PFCs and do not allow evaporative emissions
to escape.
7.4 Testing Procedures
The test procedure for diurnal emissions is to place the PFC with the spout attached in a
SHEDB, vary the temperature over a prescribed profile, and measure the hydrocarbons escaping
from the fuel container. The final result would be reported in grams per gallon where the grams
are the mass of hydrocarbons escaping from the fuel tank over 24 hours and the gallons are the
nominal PFC capacity. The test procedure is based on the automotive evaporative emission test
described in 40 CFR Part 86, Subpart B, with modifications specific to PFC applications. The
hydrocarbon loss must be measured either by weighing the cans before and after the diurnal
cycle or by measuring emissions directly from the SHED. Three identical containers must be
tested for three diurnal cycles. The daily emissions for each container are to be averaged
together for comparison with the standard, rounded to the nearest one-tenth of a gram. Each
container must meet the standard to demonstrate compliance with the standard.
Manufacturers must test cans in their most likely storage configuration. The key to
reducing evaporative losses from PFCs is to ensure that there are no openings on the cans that
could be left open by the consumer. Traditional cans have vent caps and spout caps that are
easily lost or left off cans, which leads to very high evaporative emissions. We expect
manufacturers to meet the evaporative standards by using automatic closing spouts and by
Sealed Housing for Evaporative Determination
7-15
-------
Final Regulatory Impact Analysis
removing other openings that consumers could leave open. However, if manufacturers choose to
design cans with an opening that does not close automatically, we are requiring that containers
be tested in their open condition. If the PFCs have any openings that consumers could leave
open (for example, vents with caps), these openings thus must be left open during testing. This
applies to any opening other than where the spout attaches to the can. We believe it is important
to take this approach because these openings could be a significant source of in-use emissions.
Spouts must be in place during testing because this would be the most likely storage
configuration for the emissions compliant cans. Spouts will likely still be removable so that
consumers will be able to refill the cans, but we would expect the containers to be resealed by
consumers after being refilled in order to prevent spillage during transport. We do not believe
that consumers will routinely leave spouts off cans, because spouts are integral to the cans' use
and it is obvious that they need to be sealed. Testing with spouts in place will also ensure that
the cans seal properly at the point where the nozzle attaches to the can. If cans do not seal
properly, emissions will be well above the standards.
7.4.1 Temperature Profile, Length of Test, Fill Level
PFCs will be tested over the same 72-96°F (22.2-35.6°C) temperature profile used for
automotive applications. This temperature profile represents a hot summer day when ground
level ozone emissions (formed from hydrocarbons and oxides of nitrogen) would be highest.
This temperature profile would be for the air temperature in the SHED.
The automotive diurnal test procedure includes a three-day temperature cycle. The
purpose of this test length is to ensure that the carbon canister can hold at least three days of
diurnal emissions without vapor breaking through the canister. For PFCs, we do not believe that
a three-day test is necessary. Prior to the first day of testing, the fuel will be stabilized at the
initial test temperature. Following this stabilization, a single 24-hour diurnal temperature cycle
will be run. Because this technology does not depend on purging or storage capacity of a
canister, multiple diurnal cycles per test should not be necessary.
Diurnal emissions are not only a function of temperature and fuel volatility, but of the
size of the vapor space in the PFC as well. The fill level at the start of the test will be 50% of the
nominal capacity of the PFC. Nominal capacity, defined as the volume of fuel to which the PFC
can be filled when sitting in its intended position, is to be specified by the manufacturer. The
vapor space that normally occurs in a PFC, even when "full," is not considered to be part of the
nominal capacity of the PFC.
7.4.2 Test Fuel
Consistent with the automotive test procedures, we are requiring that the test take place
using 9 RVP certification gasoline. About 20-30% of fuel sold in the U.S. contains ethanol and
this percentage is expected to increase due to the Energy Policy Act. We are requiring the use of
E10, which is a blend of 90% certification gasoline blended with 10% ethanol for diurnal testing
7-16
-------
Final Regulatory Impact Analysis
of PFCs. As noted in Section 7.2, ethanol in the fuel can increase permeation emissions for some
permeation barriers such as nylons if not properly accounted for in the design of the PFCs. Other
available permeation barriers do not allow significantly higher emissions when ethanol is present
in the fuel. Testing with E10 helps ensure that manufacturers would select materials with
emissions performance that does not degrade significantly when ethanol is present in the fuel.
7.4.3 Preconditioning and Durability Testing
We are applying essentially the same preconditioning and durability testing requirements
for PFCs that we have established for permeation control requirements for recreational vehicles.
We are also requiring a durability demonstration for spouts. As with the diurnal testing, the
preconditioning and durability testing are to be performed on the complete PFC with the spout
attached (except for pressure cycling as noted below).
7.4.3.1 Preconditioning
It takes time for fuel to permeate through the walls of containers. Permeation emissions
will increase over time as fuel slowly permeates through the container wall, until the permeation
finally stabilizes when the saturation point is reached. We want to evaluate emissions
performance once permeation emissions have stabilized, to ensure that the emissions standard is
met in-use. Therefore, we are requiring that prior to testing the PFCs, the cans need to be
preconditioned by allowing the can to sit with fuel in them until the hydrocarbon permeation rate
has stabilized. Under this step, the PFC must be filled with E10, sealed, and soaked for 20 weeks
at a temperature of 28 ± 5°C. As an alternative, we are allowing that the fuel soak could be
performed for 10 weeks at 43 ± 5°C to shorten the test time. During this fuel soak, the PFCs
must be sealed with the spout attached. We have established these soak temperatures and
durations based on protocols EPA has established to measure permeation from fuel tanks made
of FtDPE.26 These soak times should be sufficient to achieve stabilized permeation emission
rates. However, if a longer time period is necessary to achieve a stabilized rate for a given PFC,
we are requiring that the manufacturer to use a longer soak period (and/or higher temperature)
consistent with good engineering judgment.
7.4.3.2 Durability Testing
To account for permeation emission deterioration, we are specifying three durability
aging cycles: slosh, pressure-vacuum cycling, and ultraviolet (UV) exposure. They represent
conditions that are likely to occur in-use for PFCs, especially for those cans used for commercial
purposes and carried on truck beds or trailers. The purpose of these deterioration cycles is to
help ensure that the technology chosen by manufacturers is durable in-use, represents best
available control, and the measured emissions are representative of in-use permeation rates. Fuel
slosh, pressure cycling, and UV exposure each impact the durability of certain permeation
7-17
-------
Final Regulatory Impact Analysis
barriers, and we believe these cycles are needed to ensure long-term emissions control. Without
these durability cycles, manufacturers could choose to use materials that meet the certification
standard but have degraded performance in-use, leading to higher emissions. We do not expect
these procedures to adversely impact the feasibility of the standards, because there are
permeation barriers available at a reasonable cost that do not deteriorate significantly under these
conditions. As described above, we believe including these cycles as part of the certification test
is preferable to a design-based requirement.
For slosh and pressure cycling, we are requiring the use of durability tests that are based
on draft recommended Society of Automotive Engineers (SAE) practice for evaluating
permeation barriers.27 For slosh testing, the PFC must be filled to 40 percent capacity with E10
fuel and rocked for 1 million cycles. The pressure-vacuum testing contains 10,000 cycles from -
0.5 to 2.0 psi. The pressure cycling may be performed by applying pressure/vacuum through the
opening where the spout attaches, rather than by drilling a hole in the container. The third
durability test is intended to assess potential impacts of UV sunlight (0.2 |im - 0.4 jim) on the
durability of a surface treatment. In this test, the PFCs must be exposed to a UV light of at least
0.40 Watt-hour/meter2 /minute on the PFC surface for 15 hours per day for 30 days.
Alternatively, PFCs may be exposed to direct natural sunlight for an equivalent period of time.
We have also established these same durability requirements as part of our program to control
permeation emissions from recreational vehicle fuel tanks.28 While there are obvious differences
in the use of PFCs compared to the use of recreational vehicle fuel tanks, we believe the test
procedures offer assurance that permeation controls used by manufacturers will be robust and
will continue to perform as intended when in use.
We are also allowing manufacturers to do an engineering evaluation, based on data from
testing on their permeation barrier, to demonstrate that one or more of these factors (slosh, UV
exposure, and pressure cycle) do not impact the permeation rates of their PFCs and therefore that
the durability cycles are not needed. Manufacturers would use data collected previously on
PFCs or other similar containers made with the same materials and processes to demonstrate that
the emissions performance of the materials does not degrade when exposed to slosh, UV, and/or
pressure cycling. The test data must be collected under equivalent or more severe conditions as
those noted above.
In its recently revised program for PFCs, California included a durability demonstration
for spouts. We are requiring a durability demonstration consistent with California's procedures.
Automatically closing spouts are a key part of the emissions controls expected to be used to meet
the new standards. If these spouts stick or deteriorate, in-use emissions could remain very high
(essentially uncontrolled). We are interested in ways to ensure during the certification
procedures that the spouts also remain effective in use. California requires manufacturers to
actuate the spouts 200 times prior to the soak period and 200 times near the conclusion of the
soak period to simulate spout use. The spouts' internal components are required to be exposed to
fuel by tipping the can between each cycle. Spouts that stick open or leak during these cycles are
considered failures. The total of 400 spout actuations represents about 1.5 actuations per week
on average over the average container life of 5 years. In the absence of data, we believe this
7-18
-------
Final Regulatory Impact Analysis
number of actuations appears to reasonably replicate the number that can occur in-use and will
help ensure quality spout designs that do not fail in-use. We also believe that adopting
requirements consistent with California will help manufacturers to avoid duplicate testing.
The order of the durability tests would be optional. However, as discussed above, we
require that the PFC be soaked to ensure that the permeation rate is stabilized just prior to the
final permeation test. If the slosh test is run last, the length of the slosh test may be considered as
part of this soak period. Where possible, the deterioration tests may be run concurrently. For
example, the PFC could be exposed to UV light during the slosh test. In addition, if a durability
test can clearly be shown to not be necessary for a given product, manufacturers may petition to
have the test waived. For example, manufacturers may have data showing that their permeation
barrier does not deteriorate when exposed to the conditions represented by the test procedure.
After the durability testing, once the permeation rate has stabilized, the PFC is drained
and refilled with fresh fuel, the spout is placed back on the container, and the PFC is tested for
diurnal emissions.
7.4.4 Reference Container
We are requiring the use of a reference container during testing. In cases where the
permeation of a PFC is low, and the PFC is properly sealed, the effect of air buoyancy can have a
significant effect the measured weight loss. Air buoyancy refers to the effect of air density on
the perceived weight of an object. As air density increases, it will provide an upward thrust on
the PFC and create the appearance of a lighter container. Air density can be determined by
measuring relative humidity, air temperature, and air pressure.29
One testing laboratory presented data to EPA on their experience with variability in
weight loss measurements when performing permeation testing on PFCs.30 They found that the
variation was due to air buoyancy effects. By applying correction factors for air buoyancy, they
were able to greatly remove the variation in the test data. A technical brief on the calculations
they used is available in the docket.31
A more direct approach to accounting for the effects of air buoyancy is to use a reference
container. In this approach, an identical PFC to that being tested would be tested without fuel in
it and used as a reference PFC. Dry sand would be added to this PFC to make up the difference
in mass associated with the test cans being half full of fuel. The reference PFC would then be
sealed so that the buoyancy effect on the reference PFC would be the same as the test PFCs. The
measured weight loss of the test PFC could then be corrected by any measured changes in weight
in the reference can. The California Air Resources Board has required this approach for
measuring PFC emissions, and they refer to the reference PFC as a "trip blank."32
7-19
-------
Final Regulatory Impact Analysis
References for Chapter 7
1 www.arb.ca.gov/msprog/spillcon/reg.htm, Updated March 26, 2001, Copy of linked data
reports available in Docket.
2 Email from Jim Watson, California Air Resources Board, to Phil Carlson, U.S. EPA, "Early
Container Data," August 29, 2002.
3 "Test Method 513; Determination of Permeation for Spill-Proof Systems," California Air
Resources Board, Adopted July 6, 2000.
4 www.arb.ca.gov/msprog/spillcon/reg.htm, Updated March 26, 2001, Copy of linked data
reports available in Docket.
5 Email from Jim Watson, California Air Resources Board, to Phil Carlson, U.S. EPA, "Early
Container Data," August 29, 2002.
6 "Durability Testing of Barrier Treated High-Density Polyethylene Small Off-Road Engine Fuel
Tanks," California Air Resources Board, June 21, 2002.
7 "Durability Testing of Barrier Treated High-Density Polyethylene Small Off-Road Engine Fuel
Tanks," California Air Resources Board, March 7, 2003.
8 Conversation between Mike Samulski, U.S. EPA and Tom Schmoyer, Sulfo Technologies, June
17, 2002.
9 "Sulfo Data", E-mail from Tom Schmoyer, Sulfotechnologies to Mike Samulski, U.S. EPA,
March 17, 2003.
10 "ADDENDUM TO: Durability Testing of Barrier Treated High-Density Polyethylene Small
Off-Road Engine Fuel Tanks," California Air Resources Board, March 27, 2003.
11 "Resin and Additives - SO3 Compatible," Email from Tom Schmoyer, Sulfo Technologies to
Mike Samulski and Glenn Passavant, U.S. EPA, June 19, 2002.
12 Email from Jim Watson, California Air Resources Board, to Mike Samulski, U.S. EPA,
"Attachment to Resin List," August 30, 2002.
13Kathios, D., Ziff, R., "Permeation of Gasoline and Gasoline-Alcohol Fuel Blends Through
High-Density Polyethylene Fuel Tanks with Different Barrier Technologies," SAE Paper 920164,
1992.
14 www.arb.ca.gov/msprog/spillcon/reg.htm, Updated March 26, 2001, Copy of linked data
reports available in Docket.
15 "Permeation Rates of Blitz Fluorinated High Density Polyethylene Portable Fuel Containers,"
California Air Resources Board, April 5, 2002.
16 Email from Jim Watson, California Air Resources Board, to Phil Carlson, U.S. EPA, "Early
Container Data," August 29, 2002.
17 www.pensteel.co.uk. A copy of this site, as downloaded on August 13, 2004, is available in
Docket.
18 "Durability Testing of Barrier Treated High-Density Polyethylene Small Off-Road Engine Fuel
Tanks," California Air Resources Board, June 21, 2002.
19 "Durability Testing of Barrier Treated High-Density Polyethylene Small Off-Road Engine Fuel
Tanks," California Air Resources Board, March 7, 2003.
7-20
-------
Final Regulatory Impact Analysis
20Kathios, D., Ziff, R., "Permeation of Gasoline and Gasoline-Alcohol Fuel Blends Through
High-Density Polyethylene Fuel Tanks with Different Barrier Technologies," SAE Paper 920164,
1992.
21 "Fluorination Information," e-mail from Doug McGregor, BMW, to Mike Samulski, US EPA,
Augusts, 2003.
22 "Selar RB Technical Information," Faxed from David Zang, Dupont, to Mike Samulski, U.S.
EPA on May 14,2002.
23 "Fluorination Information," e-mail from Doug McGregor, BMW, to Mike Samulski, US EPA,
Augusts, 2003.
24Fead, E., Vengadam, R., Rossi, G., Olejnik, A., Thorn, J., "Speciation of Evaporative
Emissions from Plastic Fuel Tanks," SAE Paper 981376, 1998.
25 "Quantification of Permeation and Evaporative Emissions From Portable Fuel Container",
California Air Resources Board, June 2004.
26 Final Rule, "Control of Emissions from Nonroad Large Spark-ignition engines, and
Recreational Engines (Marine and Land-based)", 67 FR 68287, November 8, 2002.
27 Draft SAE Information Report J1769, "Test Protocol for Evaluation of Long Term Permeation
Barrier Durability on Non-Metallic Fuel Tanks," (Docket A-2000-01, document IV-A-24).
28 Final Rule, "Control of Emissions from Nonroad Large Spark-ignition engines, and
Recreational Engines (Marine and Land-based)", 67 FR 68287, November 8, 2002.
29 Dickson, A., Goyet, C., "Handbook of Methods for the Analysis of the Various Parameters of
the Carbon Dioxide System in Sea Water; Version 2," Prepared for the U.S. Department of
Energy, SOP21 "Applying air buoyancy corrections," September 29, 1997.
30 Testing Services Group, "CARB TM-513: Portable Fuel Container Permeation Testing,"
Presented to U.S. EPA on August 18, 2005.
31 Testing Services Group, "Technical Brief: Buoyancy Correction for Mass Measurement of
Fuel Containers," TB 50819-1, August 19, 2005.
32 California Air Resources Board, "TP-502: Test Procedure for Determining Diurnal Emissions
from Portable Fuel Containers," proposed July 22, 2005.
7-21
-------
Final Regulatory Impact Analysis
Chapter 8: Table of Contents
Chapter 8: Impact of New Requirements on Vehicle Costs 2
8.1 Costs Associated with a New Cold Temperature Standard 2
8.1.1 Hardware Costs 2
8.1.2 Development and Capital Costs 2
8.1.3 Total Per Vehicle Costs 4
8.1.4 Annual Total Nationwide Costs 5
8.2 Costs Associated with Evaporative Standards 8
8-1
-------
Final Regulatory Impact Analysis
Chapter 8: Impact of New Requirements on Vehicle Costs
Chapter 5 on vehicle feasibility describes the changes to Tier 2 vehicles we believe will
be needed to meet new cold temperature NMHC standards and new evaporative emissions
standards. This section presents our analysis of the average vehicle-related costs associated with
those changes.A For our analysis, we considered incremental hardware costs and up-front costs
for research and development (R&D), tooling, certification, and facilities. This section includes
both per vehicle and nationwide aggregate cost estimates. All costs are in 2003 dollars.
8.1 Costs Associated with a New Cold Temperature Standard
8.1.1 Hardware Costs
As described in Chapter 5, we are not expecting hardware changes to Tier 2 vehicles in
response to new cold temperature standards. Tier 2 vehicles are already being equipped with
very sophisticated emissions control systems. We expect manufacturers to use these systems to
minimize emissions at cold temperatures. We were able to demonstrate significant emissions
reductions from a Tier 2 vehicle through recalibration alone. In addition, a standard based on
averaging allows some vehicles to be above the numeric standard as long as those excess
emissions are offset by vehicles below the standard. Averaging would help manufacturers in
cases where they are not able to achieve the numeric standard for a particular vehicle group, thus
helping manufacturers avoid costly hardware changes. The phase-in of standards and emissions
credits provisions also help manufacturers avoid situations where expensive vehicle
modifications would be needed to meet a new cold temperature NMHC standard. Therefore, we
are not projecting hardware costs or additional assembly costs associated with meeting new cold
temperature NMHC emissions standards.
8.1.2 Development and Capital Costs
Manufacturers would incur research and development costs associated with a new cold
temperature standard and some may also need to upgrade testing facilities to handle increased
number of cold tests during vehicle development.
R&D
Manufacturers currently have detailed vehicle development processes designed to ensure
Tier 2 vehicles meet all applicable emissions standards throughout the useful life. These
processes include cold temperature development and testing for the cold CO standard. New
NMHC standards would add engineering effort and emissions testing to the Tier 2 vehicle
development cycle for each vehicle durability group. Manufacturers would need to calibrate
emissions controls to optimize emissions performance and potentially refine those calibrations to
ensure acceptable vehicle performance. Based on discussions with manufacturers and our
A This chapter discusses costs for Tier 2 vehicles. We believe the costs would be the same or lower for California
certified LEV-II vehicles. Tier 2 and LEV-II must meet very similar emissions standards. LEV-II vehicles,
however, must currently meet a 50°F standard which may reduce the costs associated with meeting a 20°F.
8-2
-------
Final Regulatory Impact Analysis
feasibility testing described in Chapter 5, we are projecting an average increase of 160 hours of
engineering staff time and 10 additional cold temperature development tests for each durability
group.8 The level of effort is likely to vary somewhat by durability group and also by
manufacturer, depending on their engines and emissions control systems. However, we believe
our estimate is conservatively high based on our test program. We were able with less than 80
hours of engineering effort to significantly reduce emissions from a heavier test weight vehicle
with relatively high emissions to levels well below the 0.5 g/mile fleet average standard level.
We understand that additional engineering time may be needed as the vehicles proceed through
their development cycle so we have doubled the hours needed to 160 hours. We also believe that
the average R&D costs are likely conservatively high because the projection ignores the
carryover of knowledge from the first vehicle groups designed to meet the new standard to others
phased-in later.
We estimate that the R&D costs would be incurred on average three years prior to
production. We increased the R&D costs by seven percent each year prior to introduction to
account for time value of money. This resulted in an average R&D cost per durability group of
about $42,400. To determine a per vehicle cost, we divided total annual vehicle sales by the
number of durability groups certified by manufacturers (16,948,000 vehicles sold divided by 295
durability groups) to determine an estimate of average number of vehicles sold per durability
group (about 57,500 vehicles/durability group). 1>2 Finally, for the cost analysis, the fixed R&D
costs were recovered over five years of production at a rate of seven percent.
Test Facility Upgrades
Manufacturers currently have testing facilities capable of cold temperature testing due to
the cold CO standard and also for vehicle development. We are anticipating additional vehicle
development testing due to the new cold temperature NMHC standard. During discussions with
manufacturers, manufacturers expressed a wide range of concern regarding their testing
capabilities. Some manufacturers will likely be able to absorb this additional testing with their
current facilities. Other manufacturers expressed the need to upgrade facilities to handle the
additional volume of testing. We believe that the proposed phase-in of the standards helps to
minimize the number of additional tests that will be needed in any given year and that major new
facilities will not be needed. However, we recognize that facility upgrades may be needed in
some cases to handle additional test volumes. For our cost analysis, we are including an average
facilities cost of $10 million for each of the six largest manufacturers which make up about 88
percent of the vehicles sold. This is based on discussions with manufacturers and our general
experiences with testing facilities costs. We believe the remaining manufacturers have limited
product lines with relatively few durability groups and will either be able to cover the additional
testing with their current facilities or by contracting out a small number of tests as needed.
We estimate that the facility costs will be incurred on average three years prior to the start
of the program because the facilities will be needed during vehicle development. As with R&D
costs, we increased the facilities costs by seven percent each year prior to introduction to account
for time value of money. This resulted in an overall facility cost industry-wide of about
$73,500,000. We projected that the facilities costs will be recovered over 10 years of production
' We estimated costs using $60 per engineering hour and $2,500 per test.
8-3
-------
Final Regulatory Impact Analysis
at a seven percent rate of return. To determine an average per vehicle cost, we divided the
annualized cost by annual sales.
Certification Costs
We are not projecting an increase in certification costs. Manufacturers are currently
required to measure HC when running the cold CO test procedure during certification.3 We do
not believe the standard adds significantly to manufacturers' current certification process.
Development testing is included in the estimated R&D costs described above.
8.1.3 Total Per Vehicle Costs
Our estimated per vehicle cost increase due to the new standards is relatively small
because we are projecting no hardware costs, tooling costs, or certification costs, and fixed costs
for R&D and facilities are recovered over large unit sales volumes. We estimate the average per
vehicle cost will be about $0.62 due to both the R&D and facilities costs during the first five
years of the program. The costs would be reduced to $0.44 after the five year recovery period
for R&D costs.
As discussed above, we believe the cold temperature standards are feasible for Tier 2
vehicles. We are also including other program provisions such as lead time, phase-in, averaging,
and early emissions credits that would help ease the transition to the new standards and avoid
costly vehicle redesign and new hardware. Costs associated with the new standard are fixed
costs for facilities upgrades and vehicle development. We are projecting average vehicle
development costs for vehicle recalibration and software design for cold temperature emissions
control. The costs associated with facilities are well understood based on past experience with
testing facilities and will vary depending on the current facilities of each manufacturer. The
development costs will also vary due to the wide variety of vehicles and the averaging program.
Costs could be higher if vehicles not yet phased in to the Tier 2 fleet are more difficult to control
than anticipated relative to those already phased in to the Tier 2 program. Costs may be lower
because the above analysis does not consider manufacturers being able to transfer knowledge
and experience from one vehicle family to the next. However, we do not expect the average per
vehicle cost to be considerably higher or lower than the costs projected. These fixed costs are
recovered over a large number of vehicles. Although we don't believe we have significantly
over or underestimated costs, even if the costs are twice those projected here, the per vehicle
costs would remain under $1.30 per vehicle.
We received comments from one limited product line manufacturer that it believes it will
be unable to meet the new standard without additional hardware "such as a secondary air
injection system or hydrocarbon trap or significantly alter our United States fleet mix to 100%
expensive SULEV certified vehicles." The commenter did not provide cost information in their
comments. Other manufacturers' comments supported our leadtime, phase-in, and other
transitional provisions as providing the flexibility needed to meet the standards with Tier 2
vehicle hardware. We continue to believe that manufacturers will be able to meet the standards
through vehicle development without additional hardware. However, we conducted a sensitivity
analysis in response to this comment, assuming the commenter would use new hardware to meet
8-4
-------
Final Regulatory Impact Analysis
the cold temperature standard. The commenter's sales represent about 1% of US light-duty
vehicle sales. If one percent of new vehicles required additional hardware costing $100 - $200
per vehicle, the average cost would increase from $0.62 to the range of $1.60 - $2.60 per vehicle.
We used this relatively large range of cost because it is not clear what new hardware or
combination of hardware the commenter might use on its vehicles. Also, we believe there will
be significant incentive for manufacturers to find alternative to using additional hardware in
order to remain competitive, considering that other manufacturers are unlikely to be making
hardware changes. Additional discussion of the comments received on the vehicle cold
temperature standard is provided in Chapter 3 of the Summary and Analysis of Comments for
this rule.
8.1.4 Annual Total Nationwide Costs
To estimate annual costs, we distributed the R&D costs over the phase-in schedule shown
below in Table 8.1-1 and amortized the costs over a five-year time period after vehicle
introduction using a seven percent discount rate. Based on certification data, we estimated that
about 14% (42 out of 295) of durability groups are HLDT/MDPV durability groups. The phase-
in schedule is needed to reasonably account for the timing of the R&D investment.
Table 8.1-1. Phase-in Schedule Used in Cost Analysis
Vehicle GVWR
(Category)
< 6000 Ibs
(LDV/LLDT)
> eoooibs
(HLDT/MDPV)
2010
25%
2011
50%
2012
75%
25%
2013
100%
50%
2014
75%
2015
100%
For the facilities cost, we projected that all facility modifications would occur prior to the
start of the program and would be amortized over a ten-year time period. We do not expect the
phase-in schedule to impact the timing of facilities upgrades. Manufacturers will likely upgrade
facilities prior to the first year of the phase-in. Table 8.1-2 provides annual nationwide cost
estimates. Table 8.1-3 provides non-annualized aggregate costs.
8-5
-------
Final Regulatory Impact Analysis
Table 8.1-2. Annual Nationwide Vehicle Costs
Calendar
Year
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
LDV/LLDT
Cost
0
653,858
1,307,715
1 ,961 ,573
2,615,430
2,615,430
1 ,961 ,573
1,307,715
653,858
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
HLDT/MDFV
Cost
0
0
0
108,546
217,091
325,637
434,182
434,182
325,637
217,091
108,546
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Facilities
Cost
0
10,465,114
10,465,114
10,465,114
10,465,114
10,465,114
10,465,114
10,465,114
10,465,114
10,465,114
10,465,114
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Total cost
0
11,118,971
11,772,829
12,535,232
13,297,635
13,406,181
12,860,869
12,207,011
1 1 ,444,608
10,682,205
10,573,659
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
8-6
-------
Final Regulatory Impact Analysis
Table 8.1-3. Non-Annualized Nationwide Vehicle Costs
Calendar
Year
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
LDV/LLDT
Cost
0
2,188,450
2,188,450
2,188,450
2,188,450
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
HLDT/MDFV
Cost
0
0
0
363,300
363,300
363,300
363,300
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Facilities
Cost
0
60,000,000
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Total cost
0
62,188,450
2,188,450
2,551,750
2,551,750
363,300
363,300
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
8-7
-------
Final Regulatory Impact Analysis
8.2 Costs Associated with Evaporative Standards
The standards for evaporative emissions, which are equivalent to the California LEV II
standards, are technologically feasible now. As discussed earlier in Chapter 5, the California
LEV II program contains numerically more stringent evaporative emissions standards compared
to existing EPA Tier 2 standards, but because of differences in testing requirements, we believe
the programs are essentially equivalent. This view is supported by manufacturers and current
industry practices. (See section V.C.5 of today's rule for further discussion of such test
differences — e.g., test temperatures and fuel volatilities.) A review of recent model year
certification results indicates that essentially all manufacturers certify 50-state evaporative
emission systems.4 Based on this understanding, we do not expect additional costs since we
expect that manufacturers will continue to produce 50-state evaporative systems that meet LEV
II standards.
As discussed in the section V.C.3 of final rule, some manufacturers are still developing
flex fuel vehicles (FFVs) and the evaporative control systems in some cases have not been fully
field tested and certified on the non-gasoline fuel (for example E-85, which consists of 85
percent ethanol and 15 percent gasoline). Only a few FFV systems have been certified thus far
to California LEV-II standards on the non-gasoline fuel. It is likely, however, that other vehicles
will be certified to LEV-II standards in the future so that the vehicles can be offered for sale in
California as FFVs. We are providing more lead time to manufacturers to certify to the new
evaporative standards on the non-gasoline fuel. At this time, however, we do not expect
significant hardware changes to these evaporative control systems or a significant increase in the
average costs for vehicles due to the new standards. The few systems already on the market
available in California are not significantly different from the systems used on current Tier 2-
certified FFVs.
-------
Final Regulatory Impact Analysis
References for Chapter 8
1 Ward's Automotive Yearbook 2006, Calendar Year 2005 Light-duty Vehicle Sales.
2 Certification data for the 2005 model year.
3 40 CFR Subpart C.
4 Update for FRM: U.S. EPA, Evaporative Emission Certification Results for Model Years 2004
to 2007, Memorandum to Docket EPA-HQ-OAR-2005-0036 from Bryan Manning, January 4,
2007.
8-9
-------
Final Regulatory Impact Analysis
Chapter 9: Table of Contents
Chapter 9: Costs of the Gasoline Benzene Program and Other Control Options Considered 2
9.1 Methodology 2
9.1.1 Overview 2
9.1.2 Changes to the Cost Analysis since the Proposal 5
9.1.3 LP Refinery Modeling Methodology 13
9.1.4 Summary of Refinery-by-Refinery Model Methodology 15
9.1.4.1 Estimating Individual Refinery Gasoline Blendstock Volumes 16
9.1.4.2 Refinery Blendstock Benzene Levels 23
9.1.4.3 Calibration of the Refmery-by-Refmery CostModel 24
9.2 Cost Inputs for the Benzene Control Technologies 26
9.2.1 Benzene Precursor Rerouting 26
9.2.2 Isomerizing Rerouted Benzene Precursors 27
9.2.3 Benzene Saturation 28
9.2.4 Benzene Extraction 29
9.3 Benzene Market and Prices 31
9.4 Refinery Modeling of Benzene Control Scenarios 34
9.5 Evaluation of the Refmery-by-Refmery CostModel 35
9.6 Refining Costs 36
9.6.1 Cost of the Benzene Program 37
9.6.2 Cost of Alternative Benzene Programs 43
9.6.3 Costs Used to Estimate Price Impacts of the Benzene Program 50
9.6.4 Projected Fuel Supply and Energy Impacts of the Benzene Program 54
9.7 Refinery Industry Cost Study 57
9-1
-------
Final Regulatory Impact Analysis
Chapter 9: Costs of the Gasoline Benzene Program and Other
Control Options Considered
This chapter provides a summary of the methodology used and the results obtained from
our cost analyses of the benzene control program as well as various other benzene control
options considered. We start by summarizing the refinery models used for our analysis. We
then describe our detailed methodology for estimating the benzene control costs for our benzene
program followed by the results. We present the results from our energy and supply analyses for
our benzene program. Finally, we discuss and compare the results of an oil industry cost
analysis for various benzene programs, including one which is similar to the benzene program
that was submitted as comments to the proposed rulemaking.
9.1 Methodology
9.1.1 Overview
Prior to the proposed rule, we retained the services of Abt Associates, Inc., (Mathpro)
under subcontract to ICF, Inc., to assess the cost of potential air toxics emissions control
programs. Abt Associates initially ran their linear program (LP) refinery cost model to
investigate various air toxic emissions control programs for gasoline. LP refinery models are
proven tools for estimating the costs for fuels programs which control fuel quality.1 A series of
gasoline quality control programs were evaluated using the LP refinery model including
benzene, total toxics and sulfur and RVP control.
While the LP refinery models are necessary and appropriate for many analyses, they also
have several important limitations of relevance here. When used to model the cost of nationwide
fuel control programs on the entire refining industry, LP refinery models are usually used to
model groups of refineries in geographic regions called PADDs which are defined above in
Chapter 6. The LP refinery model averages the costs over the refineries represented in the
PADDs; however, the technology chosen by the refinery model would normally be the lowest
cost technology found by the refinery model. This may represent an unreasonable choice of
technologies for individual refineries because of how refineries are configured and the
technologies to which they have access. While the choice of technologies can be limited based
on an approximate analysis of what mix of technologies would best suit the group of refineries
modeled in each PADD, this would only provide an approximate estimate of the cost incurred.
Also the LP refinery model would not be a sensible tool for estimating the credit averaging
between PADDs. The PADD trading issue could be partially overcome by iterating between
PADD refinery model runs, thus estimating the number of credits traded between PADDs and
estimating the level of benzene control in each PADD. However, the need to make multiple runs
for each PADD for each case, coupled with the need to run multiple control cases for different
benzene standards, would be very time consuming, costly and still would only result in
approximate estimates of the benzene levels achieved and the cost incurred.
For this reason, EPA contracted Abt Associates to develop a refinery-by-refinery cost
model which models the capability for each refinery to install the available benzene control
9-2
-------
Final Regulatory Impact Analysis
technologies available to them to reduce their gasoline benzene levels.2 The advantages that this
form of cost model has over the LP refinery model are that:
1. The cost for applying the benzene control technologies available to each refinery
can be modeled for each refinery;
2. The benzene level achievable by applying each benzene control technology can
be estimated for each refinery, which allows estimating the benzene level achievable in
each PADD and across the entire refining industry;
3. The benzene control cost-effectiveness (cost per amount of benzene reduction
achieved) for each benzene control technology modeled in each refinery can be compared
to that of the others;
4. The most cost-effective benzene control strategy for each refinery can be chosen
after considering the cost-effectiveness of benzene control technologies available at all
the refineries and considering the level of the benzene standard.
This strategy results in the optimum selection of benzene control technologies consistent
with how the ABT program would be expected to affect benzene control investments by the
refining industry attempting to minimize its costs. For this reason, the refinery-by-refinery cost
model was used to estimate the cost for various benzene standards both with and without ABT
programs, and the LP refinery model was used for the other air toxics control programs
considered. Because certain refinery-specific information necessary for estimating the cost of
benzene control with the refinery-by-refinery cost model was not publicly available, it was
necessary to find a way to estimate this information. The inputs and outputs from the LP
refinery cost model provide this needed information and it was utilized in the refinery-by-
refinery cost model. The information from the LP refinery model used in the refinery-by-
refinery cost model is described in Section 9.1.3.
Newly creating the refinery-by-refinery modeling tool raises questions about its viability.
For example, the LP refinery model has been used by Abt Associates for dozens, if not hundreds,
of refinery modeling studies for a variety of clients, including the oil industry, the automobile
industry, and government. These modeling studies have exposed this LP refinery modeling tool
to many opportunities for internal and external review and continued adjustment to better model
fuel quality changes imposed on the refining industry. Even though refinery modeling expertise
was relied upon during the creation of the refinery-by-refinery model, it still has not been
exposed to multiple opportunities for scrutiny. For this reason the refinery-by-refinery cost
model was evaluated three different ways. First, the model was reviewed by EPA's refining
modeling expert who has been conducting cost analyses on fuel programs for nearly 15 years.
Another check on the model was conducted by comparing its cost estimates for benzene control
with the same benzene control case evaluated with the LP refinery cost model. Finally, two peer
reviews were conducted on the refinery-by-refinery cost model by two refinery industry
consulting firms. These two refining industry consultant peer reviews were conducted late in the
proposal process, which did not allow for adjustments to the refinery model in time for the
proposal. However, their principal comments were addressed prior to undertaking the cost
analysis for the final rulemaking. The peer review comments and how we addressed them are
summarized at the beginning of Section 9.1.2.
9-3
-------
Final Regulatory Impact Analysis
A key assumption associated with the analysis is that the benzene reduction technologies
assumed to be used are those which reduce benzene levels from the feed or product streams (the
product stream is called "reformate") of the reformer, the unit in the refinery which produces
most of the benzene in gasoline.3 Basing the cost of this program on reformer benzene reduction
technologies is reasonable because the reformer contains the highest concentrations of benzene
and reformate comprises a large portion of the gasoline pool. More importantly, essentially all
the benzene reduction technologies which have been developed to date and used around the
world are designed to reduce reformer benzene levels. Thus, reducing benzene from reformate
would be expected to be the most cost-effective means for achieving benzene reductions. In
some unique situations additional benzene reduction might be available from other refinery units.
Despite considering the possibility for such reductions, we have not assumed this to be the case
here. Should it occur, it would only be at refineries where such control would be more
economical than reformate benzene control at other refineries - reducing the costs of the
program, but also increasing uncertainty that the benzene reductions that are estimated to occur
in each region of the country will actually occur. A detailed discussion on the technologies
available for benzene control is discussed in Chapter 6 of the Regulatory Impact Analysis.
A number of benzene programs were considered for the final rulemaking. These include
the proposed 0.62 vol% average benzene standard with an ABT program and several variants of
the proposed benzene standard. We evaluated some of these alternative benzene standards with
a second benzene standard called a maximum average standard. The maximum average standard
would place an additional constraint on refiners beyond the average standard. Under this option,
refiners would still be able to meet the average standard using credits; however, the maximum
average standard would require them to meet or exceed the maximum average standard in each
refinery before purchasing credits to show compliance with the average standard. The standard
effectively limits the degree to which credits can be used to demonstrate compliance. For
example, a refinery with a gasoline benzene level of 2 vol% and faced with a 1.3 vol% maximum
average standard and a 0.62 vol% average standard under a nationwide ABT program would
have to at least reduce its benzene level below 1.3 vol% to comply with this program. It could
remain above the 0.62 volume percent standard and comply with the standard through the
purchase of credits. However, its actual production would have to meet the 1.3 vol% maximum
average limit. The addition of a maximum average standard would force several high cost
refineries to take additional benzene control steps not required by the 0.62 vol% average
standard alone. The addition of a maximum average standard would thus tend to increase the
cost of a benzene program over a program without a maximum average standard.
We also evaluated a benzene standard without an ABT program. This type of benzene
program would require that the benzene levels of every refinery be reduced down to the benzene
standard. Because a number of refineries currently produce gasoline with very low benzene
levels, the average benzene level of a benzene program without an ABT program would likely
result in a national average benzene level that is lower than the standard (albeit far costlier, and
with far more negative impact on individual refineries). We also modeled several air toxics
control standards that would regulate total air toxics. Finally we modeled two different low RVP
programs and a lower sulfur standard.
9-4
-------
Final Regulatory Impact Analysis
After the proposed rule, we eliminated any further consideration of a more stringent
average gasoline sulfur standard, a low RVP standard, or any variant of a total air toxics
standard. Therefore we limited our cost analysis for the final rule to various benzene programs
above and below the proposed 0.62 vol% benzene standard, including variants with a maximum
average standard. For the final rule, we adopted a gasoline benzene content standard of 0.62
vol% benzene with a maximum average standard of 1.3 vol%. The benzene standards evaluated
for the final rule are summarized in Table 9.1-1 .
Table 9.1-1. Benzene Standards Modeled using Refinery-by-Refinery Model
Average Std.
0.50
0.60
0.60
0.62
0.62
0.62
0.62
0.62
0.62
0.65
0.65
0.70
0.70
0.71
Avg.-Max Std.
None
1.3
None
1.1
1.2
1.3
1.4
1.5
None
1.3
None
1.3
None
None
ABT Program
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
The final benzene levels and cost results for the benzene program and other benzene
standards considered are reported by PADD. This allows one to view the potential impact of the
benzene program on a regional basis. Moreover, since the PADD regions are the smallest
geographical unit of analysis for the LP refinery modeling case studies, reporting the cost results
for the benzene control cases also on a PADD-by-PADD basis allows a straightforward
comparison to the LP refinery modeling results which are reported on a PADD-basis.
Agreement of certain outputs between the refinery-by-refinery and LP models increases our
confidence in the results of both.
9.1.2 Changes to the Cost Analysis since the Proposal
In deriving the cost estimate for the final rule, we identified and made a number of
changes to the refinery modeling methodology used for the proposed rule. One of the primary
changes was to base the future year fuel prices on the Annual Energy Outlook (AEO) 2006
instead of AEO 2005. Perhaps the most important difference between the two AEO studies is
that the AEO 2006 projects a higher crude oil price of $47 per barrel for 2012, the year of the
final rule analysis, compared to the crude oil price projected by AEO 2005, which was $27 per
barrel. The primary difference caused by the higher crude oil price is that the cost of reduced
gasoline supply, such as when benzene is extracted from gasoline, is higher when the removed
benzene is replaced by other high octane petroleum compounds. AEO 2006 also projects higher
natural gas prices as well.
9-5
-------
Final Regulatory Impact Analysis
Another change was to update the refinery modeling base year to 2004 from 2003 - the
year used for the proposed rule analysis. The primary purpose for this change was to calibrate
each refinery's gasoline benzene levels and gasoline volumes to the most recent year that we
have information available. Each refinery's 2004 gasoline benzene level and volume is available
from the RFG data base.
The final rule analysis treated natural gasoline differently in the refinery-by-refinery cost
model compared to how it was treated in the proposed rule analysis. Natural gasoline contains
1.3 vol% benzene and we assumed for the proposed rule cost analysis that natural gasoline, and
other benzene-containing streams, are blended directly into gasoline without being treated to
reduce their benzene. For the final rule cost analysis, we assessed the feasibility for treating the
benzene in natural gasoline as well as the other benzene containing gasoline streams (these other
benzene-containing streams include, light straight run naphtha, light coker naphtha and light
hydrocrackate). Of these streams, the only one that we identified that refiners would treat to
reduce benzene with certainty is natural gasoline (see Chapter 6 of the RIA for a discussion of
the feasibility for treating the benzene of these other streams). The reason why we are confident
that refiners would treat the benzene in natural gasoline is because most refiners have rerouted
natural gasoline to the front of the refinery and are feeding it into the atmospheric crude tower to
facilitate the desulfurization of this stream to achieve compliance with the Tier 2 gasoline sulfur
standard. As the benzene of natural gasoline is routed through the refinery, it will be treated by
the isomerization unit, when the six carbon benzene compounds are rerouted around the
reformer, or by extraction and benzene saturation which post-treat the benzene in the reformate.
For some refineries which blend a lot of natural gasoline into their gasoline, this additional
benzene reduction can be significant.
The refinery modeling case studies conducted for the final rule were conducted on an
annual basis - which is different from the proposed rule, which was conducted on a summer
basis. As we acknowleded in the proposed rule, assessing the cost of benzene reductions solely
on a summer basis, which was done to allow the cost comparison with low RVP control, would
likely lead to a slightly conservative cost estimate for benzene reductions. For example,
recovering octane loss associated with benzene reduction is higher in the summer versus the
winter. Thus, assessing the cost of benzene reductions on an annual basis is expected to more
accurately estimate the cost of benzene reductions.
The cost analysis for the final benzene program excludes the participation of California
refineries - which differs from how the analysis was conducted for the proposed rule. After the
cost analysis was completed for the proposed rule, but before it was proposed, California state
officials decided not to be a part of the Federal benzene program and the state has maintained
this point of view. Not including California refineries in our cost analysis increases the cost of
benzene control slightly because non-California refiners cannot take advantage of the low-cost
benzene control credits that California refineries would provide them if they were included in the
program.
In addition to the above changes to our cost analysis that we identified, we also made
some adjustments that were based on public comments (from the American Petroleum Institute)
9-6
-------
Final Regulatory Impact Analysis
and peer review comments we received on the cost analysis that we conducted for the proposal.
Peer reviews on the refinery-by-refinery cost model were conducted by Jacobs Engineering and
A Second Opinion.4'5 They both are refining industry consulting firms which also have
consulted for EPA in the past. Both firms have conducted cost analyses on changes to fuel
quality - Jacobs uses a refinery cost LP refinery model while A Second Opinion has used
simpler cost estimation techniques. Based on the different experiences they each have in
conducting cost analysis, each firm brings a different perspective to the peer review process.
As expected, both peer reviewers agreed with aspects of the refinery modeling and took
issue with other aspects. Both reviewers found that the choices for benzene control technologies,
including benzene precursor rerouting with and without isomerizing this stream, benzene
saturation and benzene extraction, are sound choices for modeling the reduction in benzene
levels. Both reviewers thought, contrary to our modeling, that any benzene precursor rerouting
assumed to be occurring in the basecase would continue in the control case when benzene
saturation is applied. Applying this approach would slightly reduce the cost of the program, but
we believe a more conservative approach that results in deeper benzene reductions under the
credit trading program is more appropriate, thus relieving the need for some of the benzene
control by other refineries.
Both reviewers found that the calibration of each refinery's benzene level and gasoline
volume to their actual levels and volumes is important for establishing a sound refinery-specific
analysis, although one reviewer pointed to some anomalies in how a few specific refineries were
calibrated. Some anomalies can be expected when attempting to calibrate individual refineries
modeled using average gasoline blendstock production and quality information when their
operations deviate significantly from the average. Thus, this is not unexpected and we did not
make any changes to our methodology.
Jacobs commented that using the marginal cost of octane from the LP refinery model
(also termed shadow values) might underestimate the cost of making up lost octane since the cost
of the amount of additional octane needed might be greater than the marginal octane cost. Our
analysis of the octane made available from the Renewable Fuels standard mandated by EPAct
reveals that this octane entering into the gasoline pool would make up for the octane loss from
this benzene program several times over, and should ensure that many increments of octane
recovery could be made available at about the same price. Thus, we did not adjust our octane
cost methodology for the final rule analysis.
One of the peer review comments we received from Jacobs was in response to our
assumption that refiners assess what strategy they will take to reduce gasoline benzene levels
based on their desire to minimize their dollars expended per barrel of benzene reduced (dollars
expended includes capital amortized at 10% return on investment (ROI) after taxes). Jacobs
countered stating that refiners assess how to move forward on a particular refining strategy solely
on the desire to minimize their capital investments. We disagree with Jacob's statement. If
minimizing capital investment was a refiner's sole goal, then refiners would not have invested in
fluidized catalytic cracker (FCC) feed hydrotreating to reduce gasoline sulfur when lower capital
cost FCC naphtha hydrotreating is also available. Similarly, refiners would not opt for
hydrocrackers and would instead live with relative inflexibility of FCC units. However, we do
9-7
-------
Final Regulatory Impact Analysis
understand a refiner's desire to limit their capital investments. We contacted two refining
companies and asked them what payout they expect for their newly invested capital investments.
The two refiners said that they move forward with their capital investments when they are better
than 15% ROI. Thus for the final rule, we continued to assume that refiners assess benzene
control technology based on their dollars expended per barrel of benzene reduced, but we
amortized the capital investments involved based on the higher after-tax 15% ROI which values
the cost of capital more than the lower ROI.
Jacobs and API, in its comments on the proposed rule, provided capital cost estimates for
the benzene control technologies. We summarized our capital costs that we used for the proposed
rule analysis and those by Jacobs and API in Table 9.1-2.6
-------
Final Regulatory Impact Analysis
Table 9.1-2 EPA Capital Costs used for the Proposed Rule Compared to Jacobs and API Capital
Costs
LSR Rerouting
EPA (Abt)
Jacobs Consult
API (B&OB)
Benzene
Saturation
EPA (Abt)
Conv Saturation
Reform Spltr
Saturation Unit
Total
CD Hydro
Jacobs Consult
Reform Spltr
Saturation Unit
Total
API (B&OB)
Reform Spltr
Saturation Unit
Total
Benzene Extr
EPA (ABT)
Reform Spltr
Depentanizer
Sulfolane
Total
Jacobs Conslt
Reform Spltr
Sulfolane
Total
API (B&OB)
Aromatics Extr
Aromatics Extr
Reported/Estimated
Unit
Size
(K
b/sd)
15.0
20.0
30.0
30.0
6.3
30.0
20.0
9.1
30.0
10.0
30.0
6.3
5.4
1.8
20.0
10.4
1.0
20.4
14.3
ISBL
Cost
($MM)
7.69
9.10
7.52
6.00
2.76
8.76
7.20
9.10
10.12
19.22
7.52
9.09
16.61
6.00
1.07
19.00
26.07
9.10
17.05
26.16
134.63
134.63
$
Year
2004
2005
2Q06
2003
2003
2003
2005
2005
2Q06
2Q06
2003
2003
2003
2005
2005
2Q06
Adjustment to Standard Size and to
2004 Dollars
Scale
Factor
0.70
0.65
0.39
0.70
0.65
0.70
0.65
0.65
0.65
0.39
0.67
0.67
0.70
0.70
0.65
0.65
0.65
0.65
0.65
0.67
0.67
Infl.
Adj. (%)
100%
98%
95%
107%
107%
107%
107%
98%
98%
98%
95%
95%
107%
107%
107%
107%
96%
96%
96%
95%
95%
Std Size
(K b/sd)
15.0
15.0
15.0
30.0
6.3
30.0
30.0
30.0
13.7
30.0
30.0
10.0
30.0
30.0
6.3
5.4
1.8
16.2
8.4
0.8
8.4
5.9
ISBL
Cost
($MM)
7.69
7.41
5.45
6.43
2.96
9.39
7.72
11.62
12.91
24.53
7.14
8.62
15.76
6.43
1.15
20.37
27.95
7.77
14.56
22.33
70.41
70.41
2004 Dollars
Off -Site
Factor
12.0%
50.0%
70.2%
25.0%
25.0%
25.0%
25.0%
50.0%
75.0%
63.9
70.2%
70.2%
70.2%
25.0%
25.0%
40.0%
35.9%
50.0%
100.0%
82.6%
ISBL +
OSBL
Cost
$MM
8.61
11.11
9.27
8.04
3.70
11.74
9.65
17.43
22.60
40.03
12.15
14.68
26.82
8.04
1.43
28.52
38.00
11.66
29.11
40.77
Contingency
15%
15%
15%
Total
Capital
Cost
($MM)
8.61
11.11
10.66
8.04
3.70
11.74
9.65
17.43
22.60
40.03
13.97
17.68
31.65
8.04
1.43
28.52
38.00
11.66
29.11
40.77
113.18
113.18
Per
Barrel
Cost
($/bbl)
0.57
0.74
0.71
0.39
0.32
1.33
1.05
21.1
48.54
19.25
Comparing our capital costs used in our proposed rule analysis to those by Jacobs and
API we found that, for the most part, our capital costs were lower. We discovered that one
general reason why our capital costs were lower is that the base year for our capital costs is
several years ago, and capital costs have increased recently much faster than the rate of inflation.
For each benzene control technology, we also compared other aspects of our capital costs, such
as the offsite costs, to those used by Jacobs and API, and made additional changes to the capital
cost information we used for the proposed rule to update them for our final rule cost analysis.
Our proposed light straight run rerouting capital costs are about 80 percent those of
Jacobs and API. The inside battery limits (ISBL) portion of our LSR rerouting capital costs are
9-9
-------
Final Regulatory Impact Analysis
the highest of the three for a similar sized unit, but our 12% offsite factor is much lower. a Our
LSR rerouting offsite factor also seems low compared to the 25% offsite factor we assigned to
reformate splitters, which is another distillation column and arguably should have a similar
offsite factor. The offsite factor for Jacob's and API's LSR rerouting distillation column are
much higher at 50% and 70%, respectively. However, we believe that these are too high for a
distillation column. According to a presentation by Fluor engineers, the offsite factors for new
process units in refineries range from 10% to 80%, with the average being 40%.7 Distillation
columns are simple refinery units that we expect would have lower offsite costs. Thus we don't
believe that the higher offsite factor used by Jacobs is justified, and API's offsite factor seems
extremely high. In addition to API's very high offsite factor, API also applies a 15%
contingency factor. Contingency factors are usually reserved for estimates with significant
uncertainty, not for well proven technologies. It appears that API is being excessively
conservative in its cost estimate. After considering the different offsite factors, we decided to
increase our LSR rerouting offsite factor to 25% to make it consistent with the offsite factor for
reformate splitters.
Our proposed benzene saturation capital costs are about one third of those of Jacobs and
API. In conducting our capital cost comparison, we compared our capital costs individually for
each of the two units which comprise benzene saturation: the reformate splitter and the saturation
unit. Reviewing our reformate splitter costs we identified that its ISBL costs are lower than
API's and much lower than Jacobs'. After reviewing those costs, we found that our costs are
indeed low - perhaps solely because they are older. Updating them with cost information from
the year 2006, we increased our reformate splitter ISBL costs from $6.4 to $8.3 million for a
30,000 barrel per day unit expressed in 2004 dollars. As discussed above, our OSBL factor is
25% compared to Jacobs which is 50% and API which is 70%, along with a 15% contingency
factor. As discussed above, we have a high level of confidence with our 25% offsite factor for
distillation columns so we kept the same factor for reformate splitters.
Our proposed saturation unit capital costs are much lower than those by Jacobs and API.
We identified two reasons for our lower costs. First, our ISBL and offsite costs were much
lower than those by Jacobs and API. We reviewed our saturation unit ISBL cost and found that
it was indeed low. We obtained more recent capital cost information and based our saturation
unit ISBL capital costs on this new cost information, increasing them by about a factor of 2 1A.
Again our saturation offsite factor was much lower than that used by Jacobs and API. As
discussed above, the typical range for offsite costs is 10 to 70 percent. A benzene saturation unit
is more complicated than a simple distillation column, but less complicated than fluidized
catalytic cracker (FCC) or hydrocracker units, which would arguably have offsite costs at the
higher end of this range. For this reason, we believe that the offsite factor for a benzene
saturation unit should be about at the middle of the range for an offsite factor, so we assigned it a
a Onsite costs are for the primary unit including the distillation column, heat exchangers, pumps, heaters,
piping, valves and instrumentation. Offsite costs are for administration and control buildings, cooling tower,
electrical substation and switchgear, water and waste treatment facilities, feedstock and product storage and loading
and offloading, spare equipment kept onsite and catalysts. Normally refiners estimate offsite costs for each project
which can vary from zero to a factor several times greater than the onsite costs. For national fuel control programs,
cost estimation is averaged and a factor is used to indicate the fraction that offsite costs comprise of onsite costs.
This factor is applied for all the technologies requiring capital investment and is expressed as a single onsite and
offsite capital cost estimate.
9-10
-------
Final Regulatory Impact Analysis
40% offsite factor. We believe that our 40% factor is more appropriate than the higher offsite
factors used by Jacobs and API.
The second reason why our proposed benzene saturation costs were lower is that our ratio
of benzene saturation unit capacity to reformate splitter capacity was much lower compared to
the same ratios used by Jacobs and API. Benzene saturation units are always of lower capacity
than the reformate splitter because the reformate splitter concentrates the benzene into a single
stream separate from the rest of the reformate. If at a refinery, the six, seven and eight carbon
compounds are sent to the reformer, then the six carbon portion of reformate is likely to be on
the order of 33% of the reformate, provided that the mix of hydrocarbons are proportional for
each carbon number. However, most refiners also send nine carbon and even some higher
carbon number hydrocarbons to the reformer in addition to the six, seven and eight carbon
hydrocarbons. Thus, the six carbon hydrocarbons comprise 25% or less of the total mix of
hydrocarbons. For our proposed rule cost analysis, our benzene saturation unit capacity was
21% of the reformate capacity, while Jacobs and API assigned the benzene saturation unit
capacities which are 46% and 31% of the reformate splitter capacity, respectively. Since refiners
usually send (or want the capacity to send) the nine and heavier hydrocarbons to the reformer,
then it seems that the benzene saturation unit would only need to be sized to be about 25% of the
reformate splitter capacity, depending on whether or not a safety factor is also necessary. Based
on this reasoning, our assumption that the benzene saturation unit would be sized to be 21% of
the reformate splitter capacity would be low. We contacted a vendor of benzene saturation
technology to find out how they size their benzene saturation units relative to reformate splitters.
They typically size their benzene saturation units to be 28% of the capacity of the reformate
splitters. This relative benzene saturation unit capacity seemed reasonable based on the
discussion above, and is only slightly lower than API's but much lower than Jacobs' which
seems unnecessarily high. We changed the relative capacity of the benzene saturation unit for
our analysis to be 28% of the reformate splitter.
Our proposed benzene extraction capital costs were also lower than Jacobs', but about the
same as API's on a per-barrel basis. However, the API capital costs are for a BTX extraction
unit which is larger and therefore enjoys a better economy of scale. For a similar sized unit, the
per-barrel API capital costs would be $32 per barrel and therefore higher than ours at $21 per
barrel. We made several changes to our benzene extraction costs. First, as stated above, we
adjusted our reformate splitter ISBL capital costs higher for the benzene saturation unit and we
applied those same adjustments to our reformate splitter capital costs for benzene extraction. We
had included capital costs for a depentanizer, the purpose of which would be to ensure that no
five-carbon hydrocarbons would be sent to the extraction unit. However, after further
consideration we realized that that all reformers have a stripper that could be used to separate the
five carbon hydrocarbon compounds from the heavier hydrocarbons in reformate. Thus, adding
a depentanizer unit would be unnecessary, so we eliminated the depentanizer from our benzene
extraction costs. Finally, we assessed our capital costs for the benzene extraction unit, the
sulfolane unit. Our sulfolane unit ISBL capital costs are as high as or higher than those by
Jacobs and API. Therefore we did not adjust them. The offsite factor that we assigned to the
sulfolane unit was 40%, which is much lower than those used by Jacobs and API. Using the
reasoning that we used above for estimating the offsite factor, we believe that the offsite factor
should be higher than 40%. The offsite costs are usually very high for a benzene saturation unit
9-11
-------
Final Regulatory Impact Analysis
because of the need for adding special benzene and extraction chemical storage, offloading
facilities and the costly environmental controls necessary to control benzene fugitive emissions.
The offsite costs for benzene extraction are usually higher than FCC and hydrocracker units
which are other complex refinery units with high offsite factors. We therefore increased benzene
extraction unit's offsite factor to 100% of the ISBL capital costs. The last variable in the
extraction unit's costs is the relative capacity for the sulfolane unit compared to the reformate
splitter. For the saturation unit capital costs, we concluded that the saturation unit capacity
should be sized to be 28% of the reformate unit capacity. Since the reformate splitter will be
creating the same benzene-rich stream for extraction as it would for saturation, we assigned the
same relative ratio of extraction unit capacity to reformate splitter unit capacity, which is 28%.
Again, Jacobs used a very conservative ratio for the benzene sulfolate extraction unit capacity
compared to the capacity for the reformate splitter unit, which we believe is unjustified.
After making the above adjustments to our capital costs, we summarize our revised
capital costs in Table 9.1-3 below, comparing them to the Jacobs and API capital costs. The
values in Table 9.1-3 which are in bold are revised from the values presented in the proposed
rule.b
b After further reviewing the cost information for the benzene saturation technologies as we adopted the revised
capital cost estimates into our refinery cost model, we realized that the differences in capital costs from literature
between a Bensat unit and a CDHydro unit were greater than expected compared to how these technologies differ.
To remedy this, we conservatively assigned CDHydro's capital costs to be the same as those as Bensat, as described
in Section 9.2.3.
9-12
-------
Final Regulatory Impact Analysis
Table 9.1-3 Our Revised Capital Costs Compared to Capital Costs Provided by Jacobs and API
(values in bold indicated that they were updated since the proposed rule)
LSR Rerouting
EPA (Abt)
Jacobs Consult
API (B&OB)
Benzene
Saturation
EPA (Abt)
Conv Saturation
Reform Spltr
Saturation Unit
Total
CD Hydro
Jacobs Consult
Reform Spltr
Saturation Unit
Total
API (B&OB)
Reform Spltr
Saturation Unit
Total
Benzene Extr
EPA (ABT)
Reform Spltr
Sulfolane
Total
Jacobs Conslt
Reform Spltr
Sulfolane
Total
API (B&OB)
Aromatics Extr
Aromatics Extr
Reported/Estimated
Unit
Size
(K
b/sd)
15.0
20.0
30.0
30.0
8.4
30.0
20.0
9.1
30.0
10.0
30.0
8.4
1.8
20.0
10.4
1.0
20.4
14.3
ISBL
Cost
($MM)
7.69
9.10
7.52
8.79
8.67
17.46
5.86
9.10
10.12
19.22
7.52
9.09
8.79
25.20
33.99
9.10
17.05
26.16
134.63
134.63
$
Year
2004
2005
2Q06
2006
2006
2003
2005
2005
2Q06
2Q06
2006
2003
2005
2005
2Q06
Adjustment to Standard Size and to
2004 Dollars
Scale
Factor
0.70
0.65
0.39
0.70
0.65
0.70
0.65
0.65
0.65
0.39
0.67
0.70
0.65
0.65
0.65
0.65
0.65
0.67
0.67
Infl.
Adj. (%)
100%
96%
92%
95%
95%
95%
110%
98%
98%
98%
95%
95%
95%
110%
107%
96%
96%
96%
95%
95%
Std Size
(K b/sd)
15.0
15.0
15.0
30.0
8.4
30.0
30.0
30.0
13.7
30.0
30.0
10.0
30.0
30.0
8.4
1.8
16.2
8.4
0.8
8.4
5.9
ISBL
Cost
($MM)
7.69
7.41
5.45
8.34
8.22
16.56
6.44
11.62
12.91
24.53
7.14
8.62
15.76
8.34
27.73
36.07
7.77
14.56
22.33
70.41
70.41
2004 Dollars
Off -Site
Factor
25.0%
50.0%
70.2%
25.0%
40.0%
25.0%
40.0%
50.0%
75.0%
63.9
70.2%
70.2%
70.2%
25.0%
100.0%
82.7%
50.0%
100.0%
82.6%
ISBL +
OSBL
Cost
$MM
9.61
11.11
9.27
10.42
11.51
21.94
9.01
17.43
22.60
40.03
12.15
14.68
26.82
10.42
55.45
65.88
11.66
29.11
40.77
Contingency
15%
15%
15%
Total
Capital
Cost
($MM)
9.61
11.11
10.66
10.42
11.51
21.94
9.01
17.43
22.60
40.03
13.97
17.68
31.65
10.42
55.45
65.88
11.66
29.11
40.77
113.18
113.18
Per
Barrel
Cost
($/bbl)
0.64
0.74
0.71
0.73
0.30
1.33
1.05
36.60
48.54
19.25
9.1.3 LP Refinery Modeling Methodology
Although the benzene control costs estimated for the final rule were estimated using the
refinery-by-refinery cost model, certain inputs into that model were taken from the input tables
or from the results of the refinery modeling output from the LP refinery model - hence its
importance for the cost analysis. The information from the LP refinery model used in the
refinery-by-refinery model included the average benzene content of the various streams which
make up gasoline, the price of hydrogen, the cost for making up the octane-barrel loss of octane,
and the price of gasoline. Certain refinery operations information from the LP refinery model
was used for estimating the volume of gasoline produced in the refinery-by-refinery model,
9-13
-------
Final Regulatory Impact Analysis
including the utilization factors of individual refinery units, and the percentage that straight run
naphtha, FCC naphtha and hydrocrackate comprises of the feed volume of their respective units.
The means for using the specific inputs from the LP refinery model discussed here in the
refinery-by-refinery model are summarized below in the section discussing the refmery-by-
refmery model methodology.
LP refinery models are detailed mathematical representations of refineries. They are
used by individual refining companies to project how best to operate their refineries. They are
also used by government agencies, such as EPA and DOE, as well as by refining industry
associations and individual companies, to estimate the cost and supply impacts of fuel quality
changes. LP refinery models have been used for these purposes for decades and a certain
protocol has been established to conduct these studies. For estimating the cost and other impacts
of a future gasoline quality standard, the refinery modeling work is conducted in three steps.
The first step in conducting an LP refinery modeling analysis is the development of a
base case. The base case is a refinery modeling case that calibrates the refinery model based on
actual refinery unit capacity and input and output data. The base year for this study was the year
2000. Because much of the information available for establishing the base case is only available
for PADDs of refineries, the LP refinery modeling is conducted on a PADD-wide basis.
Refinery capacity information from the Oil and Gas Journal is aggregated by PADD and entered
into the LP refinery model.8 The year 2000 feedstock volumes including crude oil, oxygenates,
and gasoline blendstocks, were obtained from the Energy Information Administration and
entered into each PADD's model. Similarly, year 2000 product volumes such as gasoline, jet
fuel and diesel fuel, were obtained from EIA and entered into the cost model. The environmental
and ASTM fuel quality constraints in effect by 2000 are imposed on the products. This includes
the Reformulated Gasoline program and the 500 ppm highway diesel fuel sulfur standard. This
information was input into the LP refinery cost model for each PADD and each PADD model
was run to model the U.S. refinery industry for the year 2000, which is the base year. The
gasoline quality for each PADD refinery model was then compared to the actual gasoline quality
which is available from the RFG data base. Each model was calibrated to closely approximate
the gasoline quality of each PADD.
The next step in modeling is the development of a reference case. The purpose of the
reference case is to model the refining industry operations and cost in a future year, which is the
year that the air toxics cases are modeled to be in effect (serving as a point of reference to the
modeled air toxics cases for estimating costs). The benzene program was assumed to take effect
in 2012. The reference case is created by starting with the 2000 base cases for each PADD and
adjusting each base case to model the future year, accounting for the changes between the two
years.
Two different types of adjustments were made to the base case refinery models to enable
modeling the refining industry in 2012 for the reference case. First, the change in certain inputs
such as product volumes and energy prices need to be accounted for. U.S. refinery gasoline,
diesel fuel and jet fuel demand are projected by EIA to grow to meet increased demand.9 This
growth in demand is used to project refinery production for each PADD to meet that increased
demand. This projected growth in U.S. refinery production is entered into the reference case
9-14
-------
Final Regulatory Impact Analysis
version of the LP refinery model. Another adjustment is made to account for changes in energy
prices which are projected by EIA for future years.
The second adjustment made to model the reference cases is the application of fuel
quality changes. Environmental programs which have been implemented or which will largely
be implemented by the time that the prospective benzene program would take effect were
modeled in the reference case. These fuel quality changes include limits such as the 30 ppm
average gasoline sulfur standard, and 15 ppm caps on highway and nonroad diesel fuel, in
addition to the environmental programs which were already being modeled in the base case.
Additionally, we also modeled the implementation of EPAct, which requires a large increase in
the amount of ethanol to be blended into gasoline to comply with the renewable fuels standard
(RFS). In its AEO 2006, EIA has projected that the volume of ethanol blended into gasoline will
exceed the RFS required amounts, resulting in 9.6 billion gallons of ethanol blended into
gasoline by 2012. Other provisions of EPAct that we modeled included a nationwide ban on
MTBE and rescinding the RFG oxygenate standard.
The third step in conducting the LP refinery modeling was to run the various control
cases. The control cases are created by applying a specific fuel control standard to each PADD
reference case. The control cases are run with capital costs evaluated at a 15 percent rate of
return on investment (ROI) after taxes. The refinery model output for each PADD is then
compared to the reference case output and the changes in refining operations, fuel quality and
costs are reviewed and reported. In the reported results the capital costs are adjusted to a 7
percent rate of ROI before taxes.
9.1.4 Summary of Refinery-by-Refinery Model Methodology
The methodology used for estimating costs with the refinery-by-refinery cost model has
some similarities with the methodology used with the LP refinery cost model. Although the
refinery-by-refinery cost model is a separate cost estimation tool, the means for using the
mathematical representation of the benzene control technologies for estimating the cost and the
final gasoline benzene level by reducing benzene levels is very similar. The principal difference
is that the refinery-by-refinery cost model estimates the gasoline production and benzene level
for each refinery, while the LP refinery model estimates the benzene levels of the aggregate
gasoline produced by each PADD of refineries. As discussed above, the modeling of each
refinery is important to understanding the impact of the ABT program on compliance and cost.
However, attempting to model the refinery operations for each refinery has its own set of
challenges. This section presents various steps used in our methodology for estimating the
operations and benzene control costs for individual refineries.
The first step was to estimate year 2004 baseline operating conditions for each refinery.
This involves estimating the volumes and benzene levels of the gasoline blendstocks that
comprise each refinery's gasoline. As a final adjustment to our estimated gasoline volumes and
benzene levels, we calibrate them against actual refinery gasoline volume and benzene levels.
For seven refineries, we had gasoline blendstock volumes and benzene levels which the refining
companies shared with us in our previous discussions with them for MSAT1 concerning air
toxics control and during our discussions with refiners prior to the proposed rule. This specific
9-15
-------
Final Regulatory Impact Analysis
refinery information provided to us was entered into the refinery-by-refinery model avoiding the
need to estimate it.
The 2004 gasoline production volumes and refinery operating conditions were projected
to 2012, the year that we modeled the cost for gasoline benzene control. We chose the year 2012
for modeling the cost of benzene reductions because it represented a midyear in the range of
years that the benzene program is expected to phase in. The phase-in years range from 2007 to
2015 with the major benzene reductions expected to occur in 2015. Based on projections by the
Energy Information Administration, gasoline demand is expected to increase by 12.5 percent
between 2004 and 2012.10
The next step involves applying the various benzene control technologies as appropriate
in each refinery. This allows us to make a cost estimate for using each benzene control
technology in each refinery. The capital costs for installing the various benzene control
technologies in each refinery were evaluated based on a 15 percent rate of return on investment
(ROI) after taxes, but were adjusted post modeling to a 7 percent ROI before taxes for reporting
the results. We also report the cost estimates based on capital costs amortized at 6 and 10
percent ROI after taxes, to represent the typical return on investments experienced by refiners. A
key part of illustrating this step is a summary of the cost inputs for the various benzene control
technologies. We also describe how the four benzene control strategies were utilized to meet the
various benzene standards.
9.1.4.1 Estimating Individual Refinery Gasoline Blendstock Volumes
To calibrate each refinery to its current benzene levels and gasoline volumes, and to
provide the best opportunity for estimating the cost and ultimate level of benzene control, it is
necessary to understand the benzene levels and volumes of the various blendstocks which make
up each refinery's gasoline. Information on the volumes and benzene levels of each gasoline
blendstock contained in each refinery's gasoline is not publicly available, so it was necessary to
estimate them. This is accomplished by adjusting published refinery unit capacity information to
estimate the extent that each refinery unit is utilized, followed by a unit-specific analysis for
estimating how each refinery unit produces material for blending into gasoline. After the unit-
by-unit estimates are completed, we do an overall check by comparing our estimated gasoline
volumes with actual gasoline volume. We force the estimated gasoline volumes to match the
actual gasoline volume using a factor which adjusts the estimated gasoline volume of each
refinery unit.
The Oil and Gas Journal publishes, and the Energy Information Administration reports,
unit capacities for the principal refinery units for each refinery in the U.S.u 12 Information from
these two sources was reviewed for the year 2004, the base year for the cost model, and the
information judged best overall from the two sources was entered into the refinery-by-refinery
cost model. This information was used as a first step in the process to estimate the volumetric
contribution of each of the gasoline producing units to each refinery's gasoline pool. The units
analyzed include coking, fluidized catalytic cracking (FCC), hydrocracking, alkylation, dimersol,
polymerization, isomerization, reforming and aromatics extraction.
9-16
-------
Final Regulatory Impact Analysis
An initial assumption was made that each unit in each refinery is being operated at the
percent of capacity for the respective unit's percent of operating capacity for the PADD of
refineries being modeled by the LP refinery model. The initial percent of capacity utilization for
each unit as estimated by the LP refinery model for 2004 and 2012 is presented in Table 9.1-4.
Table 9.1-4. Initial 2004 and 2012 Percent of Refinery Unit Capacity used in Refinery-by-
Refinery Cost Model
Grade
Coking
FCC
Hydrocracker
Isomerization
Polymerization
Alkylation
Reforming
Aromatics
2004
2012
2004
2012
2004
2012
2004
2012
2004
2012
2004
2012
2004
2012
2004
2012
2004
2012
PADD1
101
103
97
88
94
96
100
100
98
98
90
101
100
103
88
93
100
100
PADD2
94
97
90
87
97
97
102
111
100
72
86
98
92
96
82
82
65
67
PADD3
97
100
96
104
95
96
77
100
100
100
64
87
71
75
85
96
88
94
PADD 4 & 5
exCA
89
98
100
100
100
111
100
110
103
100
10
71
89
95
85
72
-
-
The estimates of refinery unit capacity utilized in Table 9.1-4 are a product of how the LP
refinery model models the use of refinery units in each PADD of refineries. Normally, we would
expect year 2004 (baseyear) refinery unit utilization to be 80 to 95 percent of listed capacity. For
some units this is the case, but for many of the units this is not the case. There are two reasons
for this. First, listed refinery unit capacity can be wrong. For past refinery modeling efforts, we
have compared the listed unit capacity for specific refinery units between EIA and the Oil and
Gas Journal and have seen significant differences between the two sources. We do not know
which source is right, or if either of the sources is right. The second reason why there may be a
discrepancy is because LP refinery models attempt to model PADDs of refineries based on
average operating characteristics, which can vary substantially between refineries, and can vary
between PADDs based on regional differences in how the units are being operated. If such
average operating characteristics are not capturing the refining characteristics adequately, then
this could lead to over and underestimating refinery unit utilization. Despite the occasional
apparent anomaly in percent of operating capacity estimated by the LP refinery model, we chose
to use the LP refinery model's estimated refinery utilization factors.
Estimating refinery unit capacity and utilization of that capacity may or may not translate
directly into the gasoline blendstock volume produced by a specific refinery unit because some
of the refinery units produce more than one refinery product or they may affect the density of the
9-17
-------
Final Regulatory Impact Analysis
feedstock to that unit. How the refinery unit capacity and its utilization are used to estimate
gasoline blendstock volume is described in detail for each major refinery unit.
For the polymerization and alkylation units listed in Table 9.1-4, the actual capacity of
the unit coupled with its estimated utilization does establish the initial volume of gasoline
blendstock volume produced by those units. For example, a particular refinery unit in PADD 1
might have a 10,000 barrel per day alkylation unit. Table 9.1-4 shows that the alkylation units in
PADD 1 are estimated to be operating at 103 percent of its listed capacity in 2012, thus, alkylate
production is projected to be 10,300 barrels per day at that refinery.
Other gasoline blendstocks require additional steps to estimate their volumes, including
light straight run naphtha, FCC naphtha, coker naphtha and hydrocrackate. Each of these other
gasoline blendstocks are produced based on a portion of the unit capacities for the units used to
produce them. To illustrate the methodology used to estimate the volumes, we will use light
straight run naphtha as an example. Light straight run naphtha is principally comprised of five
carbon hydrocarbons which come directly from crude oil. Thus to model the volume of the light
straight run naphtha, it was necessary to estimate the volume of crude oil as well as the
percentage that light straight naphtha comprises of crude oil. The Oil and Gas Journal contains
reported capacities of the atmospheric crude oil towers for each refinery. The reported crude oil
tower capacity is adjusted using the percent of unit utilization estimates for the crude unit
contained in Table 9.1-4 applying the same adjustment to each refinery in each PADD. These
calculations provided us an estimate of the volume of crude oil processed by each refinery. The
fraction of light straight run naphtha in each refinery's crude oil was estimated from the
percentage that light straight run comprises of crude oil for each PADD in the LP refinery model.
This percentage is based on the types and quality of crude oil processed by all the refineries in
each PADD - information obtained from the Energy Information Administration.13 The
percentage that light straight run naphtha comprises of crude oil is applied to each refinery in the
refinery-by-refinery cost model. As summarized below in Table 9.1-5, the volume of light
straight run naphtha is estimated to be 4 to 5 percent of the crude oil volume processed
depending on the PADD.
Light straight run has three possible different fates depending on the refinery. Except for
PADD 2, a portion is designated to be sold into the petrochemicals market. For PADDs 1, 3, 4
and 5, although primarily in PADD 3, a portion of straight run naphtha is processed and sold to
petrochemical companies which use the material to make other hydrocarbon compounds. EIA
publishes the volume of naphtha which is sold into the petrochemicals market in each PADD.14
Since no source of information is publicly available that specifies the volume of naphtha sold by
each refinery to the petrochemicals market, the volume of light straight run naphtha sold into the
petrochemicals market by each refinery was assumed to be proportional to the percentage that its
crude oil processing capacity comprises of the total crude oil processing capacity in the PADD.
After accounting for the volume of light straight run naphtha sold to the petrochemicals market,
the balance of straight run naphtha is blended directly into gasoline for those refineries without
an isomerization unit. For refineries with an isomerization unit, the volume of light straight
naphtha not sent to the petrochemicals market is sent to the isomerization unit up to the capacity
of that unit, and the balance is blended directly into gasoline.
9-18
-------
Final Regulatory Impact Analysis
The hydrocracker and coker units produce some light naphtha material which plays a role
in blending up gasoline. The light naphtha material produced by the hydrocracker and coker are
termed light hydrocrackate and light coker naphtha, respectively. The portion of the material
processed by each of these units converted to light coker naphtha and light hydrocrackate is 5
percent for coker units across all the PADDs, and ranges from 23 to 32 percent for hydrocracker
units depending on the PADD. Table 9.1-5 below summarizes the percentage of total material
processed by these units into light naphtha.
The volume of isomerate, the product produced by the isomerization unit, is based on the
feed to the isomerization unit up to its capacity. As described above, the volume of light straight
run is estimated and that volume which is not assumed to be sold into the petrochemical markets
is assumed to be sent to the isomerization unit. An additional source of feed to the isomerization
unit, as described below, is a portion of the six carbon hydrocarbons which is estimated to be
sent to the isomerization unit to calibrate a refinery's benzene levels. This is one of the strategies
used by refiners to reduce their benzene levels today, although in a limited way since the
refinery-by-refinery model estimates that 26 refineries in the U.S. in 2012 are sending their six
carbon hydrocarbons to the isomerization unit. The six carbon hydrocarbons have priority to the
light straight run which is sent to the isomerization unit. In all cases, the volume of isomerate
produced by isomerization units is estimated to be 1.6 volume percent less than its feed.
The volume of reformate was estimated based on the feed to the unit as limited by each
unit's capacity. The feed to the reformer comes from various sources depending on the refinery
configuration. For virtually all refineries, part of the naphtha from the atmospheric crude tower
is sent to the reformer. Those refineries with a hydrocracker or a coker will send part of the
naphtha from these units to the reformer as well. The naphtha sent to the reformer from these
various units is that portion that is heavier than the light naphtha which is either sent to the
isomerization unit or blended directly to gasoline. This reformate feed naphtha contains the six,
seven, eight and usually the nine carbon compounds from these various sources. In some cases,
the six carbon compounds are separated from the rest of the reformate feedstock to reduce the
benzene in the final reformate. As discussed above, this rerouted six carbon stream is either
blended directly into gasoline or is sent to the isomerization unit for further benzene control.
The volume of the feed to the reformer is estimated on a PADD basis and is based on fractions of
the material processed in the atmospheric crude tower, hydrocracker and coker.
The fraction of crude oil that is fed to the reformer ranges from about 13 to 16 percent
depending on the PADD. About 18 percent of the material processed in the coker unit is
estimated to end up as feedstock to the reformer. Of the feed processed in the hydrocracker, a
range of 30 to 50 percent is estimated to end up as feed to the reformer unit, depending on the
PADD. The variance in the fraction of hydrocracker material sent to the reformer is due to the
significant flexibility that the hydrocracker has for producing either gasoline or diesel fuel. In
certain PADDs, such as PADD 4 and 5, there is a higher relative demand for diesel fuel
compared to gasoline so there is a lower conversion to naphtha than in other PADDs. The
product from the reformer experiences a volume decrease of about 18 percent relative to the
volume of feed due to the conversion of straight chain and cyclical hydrocarbons to energy dense
aromatics and other light products. This volume shrinkage and conversion to lighter products
increases with the severity and thus the conversion of the reformer unit. All the refineries in
9-19
-------
Final Regulatory Impact Analysis
each PADD are assumed to be operating their reformers at the same severity as estimated by the
LP refinery model. For PADDs 1 through 5, the reformer severity in 2012 falls within a range of
92 to 96 research octane number RON.c This range of reformer severity is projected to be lower
than the reformer severity common today because of the projected increase in ethanol use and
the high octane that it provides.
The FCC unit contributes a substantial volume to gasoline. We estimated the utilization
of each refinery FCC unit by adjusting the nameplate capacity of each unit using the utilization
factors listed in Table 9.1-4. Like a number of other gasoline producing units, only a portion of
the feedstock of the FCC unit is converted to naphtha. Again, we used PADD-average estimates
used in the LP refinery model for estimating the portion of the FCC feed volume converted to
naphtha. The conversion percentage to naphtha is affected by the conversion severity of the
individual unit. The PADD-average conversion severity is estimated to be fairly consistent
across the PADDs, so the portion of FCC feedstock converted to naphtha is quite consistent at
about 55 to 57 percent.
Some gasoline blendstocks are purchased and blended into gasoline. The typically
purchased gasoline blendstocks include natural gasoline, alkylate, isooctene and ethanol. We did
not have information on the volume of these gasoline blendstocks purchased and blended into
gasoline by each refinery, so we again relied on the information from EIA, which reports the
consumption of these blendstocks on a PADD basis, and our contractor who estimated the
volume of isooctene which will be available from the conversion of MTBE plants. Based on the
work we conducted for the Renewable Fuels Proposed rule, we provided to our contractor the
volume of ethanol projected to be used in each PADD. We assumed that each refinery in the
PADD purchased a portion of the total amount of gasoline blendstocks purchased in that PADD
in proportion to that refinery's crude oil consumption within the PADD.
Another impact on gasoline volume is the volume of aromatics extracted from gasoline.
Refiners extract aromatics to comply with the RFG toxics standards and also to take advantage
of the higher price of aromatics, such as xylene and benzene, earns over the price of gasoline.
The volume of aromatics, including benzene, extracted from gasoline was initially based on the
nameplate capacity of each refinery's extraction unit listed in the Oil and Gas Journal. Unlike
other refinery units, the extraction unit capacity is based on the volume of aromatics produced
instead of the unit's feed volume. This production volume is estimated based on the unit
capacity and aromatics plant utilization estimated by the LP refinery model as summarized in
Table 9.1-4. This strategy was effective for the few refineries in PADD 2 with extraction units
because it resulted in estimated gasoline benzene levels which closely matched the actual
benzene levels for those refineries. However, this method was ineffective at matching the level
of benzene for individual refineries in PADDs 1 and 3. One reason why the calibration method
did not work so well for the extraction units in PADDs 1 and 3 is because a number of the
refiners there are likely purchasing reformate for other refineries and processing them in their
extraction units. For those PADDs, the degree to which their extraction units were being utilized
was based solely on the need to calibrate each refinery's benzene levels to match year 2003
0 The severity of reformers is measured by the research octane number (RON) of its product. RON together
with motor octane number (MON) makes up the total octane ((R+M)/2) of any gasoline blendstock or the gasoline
pool.
9-20
-------
Final Regulatory Impact Analysis
benzene levels. Each extraction unit had sufficient capacity to supply the needed extraction
estimated, and when averaged across each PADD, this method did match the LP refinery
model's estimated PADD utilization for extractions units reasonably well.
A series of inputs are made to the refinery-by-refinery cost model which are necessary to
estimate the cost for certain aspects of the cost modeling. These inputs are from the LP refinery
model and EIA.
As stated above, hydrogen is necessary to saturate the benzene in the isomerization
reactor when the rerouted benzene precursors are sent there. Similarly, hydrogen is consumed
when benzene is saturated in benzene saturation units. It is also necessary to assign a cost for the
lost hydrogen production in the reformer when the benzene precursors are rerouted around the
reformer. This lost hydrogen production or additional hydrogen consumption must be made up
from somewhere. A price derived from the LP refinery model is assigned for the lost hydrogen
production and/or that consumed for saturating benzene. The LP refinery estimates the cost for
building new hydrogen plant capacity to provide more hydrogen. The cost for this hydrogen
varies somewhat by the region of the country because the typical size of hydrogen plant usually
built in each region varies, which affects the economies of scale for the installed capital.
Hydrogen costs also tend to vary because the feedstocks to hydrogen plants, which is usually
natural gas, also varies by region. To incorporate this variance in regional hydrogen costs, the
hydrogen costs are estimated, and entered into the refinery-by-refinery cost model, by PADD.
These hydrogen prices may be conservative as they do not consider the economies of scale of
producing hydrogen from very large third party hydrogen producers. Conversely, these
hydrogen costs may be optimistic as they were based on EIA energy price projections that are
lower than today's energy prices; for example, crude oil prices are assumed to be $47 dollar per
barrel.15
Another input made to the refinery model is a cost factor used for estimating the cost of
lost octane. When benzene precursors are routed around the reformer, when benzene is saturated
in a benzene saturation unit, or when benzene is extracted from gasoline, the octane of the
resulting gasoline is reduced. Similarly, when the rerouted benzene precursors are sent to the
isomerization unit, the natural benzene from crude oil which is in that stream is saturated and the
high octane of the benzene is lost. However, this resulting low octane stream is then treated in
the isomerization unit which offsets some of the lost octane. For all these cases, the cost for the
net octane loss is accounted for by assigning an octane-barrel cost to the octane change. The
octane-barrel cost is from the LP refinery model which, like for hydrogen, estimates a cost for
making up lost octane. There is a regional variance in the type of octane producing units, in the
economies of scale for designing and constructing these units and in prices for purchased high
octane blendstocks which results in differences in the cost for making up octane loss by PADD.
To account for the regional variance in octane costs, octane barrel costs are estimated, and
entered into the refinery-by-refinery cost model, by PADD.
Gasoline prices are also a necessary input into the refinery-by-refinery cost model to
account for the effects by these various benzene control technologies on changes in gasoline
volume. Extracting benzene from gasoline and selling the benzene into the chemicals market
will result in a small reduction in gasoline produced by the refineries estimated to use this
9-21
-------
Final Regulatory Impact Analysis
technology. When the benzene precursors are routed around the reformer, the reduction in
feedstock to the reformer will increase gasoline supply. This is because the cracking and
aromatization reactions which occur in the reformer reduce the hydrocarbon volume. To account
for the full cost of benzene control, it is necessary to account for the change in gasoline volume.
This loss in gasoline volume supply is accounted for by multiplying the change in gasoline
volume with the gasoline prices from EIA on a PADD basis.
16
The various assumptions associated with estimating gasoline blendstocks and the
volumes of purchased and sold blendstocks and cost factors in 2012 are summarized in Table
9.1-5.
Table 9.1-5. Information used with the Refinery-by-Refinery Cost Model
(Projected Year 2012 Operating Conditions and Year 2004 dollars)
Hydrogen
Cost ($/foeb)
Octane Cost ($/oct-bbl)
RVP Cost ($/rvp-bbl)
Gasoline Price ($/bbl)
Light Straight Run Naphtha (% of Crude Oil)
Medium and Heavy Straight Run Naphtha (% of
Crude Oil)
Reformate
Severity (RON)
Average Reformate Yield (vol%)
Light Coker Naphtha (% of Unit Feed)
Medium and Heavy Coker Naphtha (% of Unit
Feed)
Light Hydrocrackate (% of Unit Feed)
Medium and Heavy Hydrocrackate (% of Unit
Feed)
FCC Naphtha (% of Feed)
Aromatics (% of Unit Capacity)
Inputs
Outputs
Isooctene Purchased (Kbbl/d)
Alkylate Purchased (Kbbl/d)
Natural Gasoline (Kbbl/d)
Ethanol (Kbbl/d)
Naphtha to Petrochem. (Kbbl/d)
Gasoline Blendstocks Kbbl/d)
PADD1
121
0.28
0.36
54
4.5
13.8
94.7
82
5
18.4
28.7
35.4
56.6
As
necessary
20
0
0
73
2
0
PADD 2
108
0.20
0.26
55
5.0
16.2
92.1
83
5
18.4
32.0
43.4
56.9
0.62
0
0
48
203
0
0
PADD 3
82
0.30
0.25
52
4.4
14.0
96.2
82
5
18.4
23.3
50.2
54.9
As
necessary
0
0
117
150
134
0
PADDs 4, 5
93
0.27
0.28
51
4.4
13.6
96.2
81
5
18.4
27.2
33.3
56.4
-
0
0
35
59
1
8
Utility costs are also an input into the refinery-by-refinery cost model. The benzene
reduction technologies consume natural gas, electricity and steam which contribute to the total
cost of using these technologies. The consumption of the utilities is converted to per-gallon costs
using average cost factors for the individual utilities. The utility costs are from EIA, although for
the case of steam are calculated based on fuel oil costs, and are represented on a PADD basis.
Another input into the cost model is a cost factor used for adjusting the installed capital
costs depending on the PADD in which the capital is being installed. Installing capital in
refineries has been shown to vary geographically depending on the region in which the refinery
is located. This difference in cost is primarily due to differences in contractor costs used for
9-22
-------
Final Regulatory Impact Analysis
installing the costs in each region. Installing capital is cheapest in PADD 3 (Gulf Coast), and
most expensive in PADDs 4 and 5 with capital costs 40 percent higher than in PADD 3.
Table 9.1-6 summarizes the various cost factors used in the refinery-by-refinery cost
model by PADD.
Table 9.3-1. 2012 Cost Factors by PADD (2004 dollars)
Natural Gas $/foeb
Electricity $/kw-hr
Steam $/lb
Capital Cost Adjustment
Factors
PADD 1
48.3
0.069
0.010
1.25
PADD 2
43.0
0.044
0.0091
1.15
PADDS
32.9
0.056
0.0070
1.00
PADDs 4 & 5
37.0
0.057
0.0079
1.40
9.1.4.2 Refinery Blendstock Benzene Levels
It is necessary to estimate the benzene levels of individual gasoline blendstocks to model
the benzene levels of gasoline today and for estimating the benzene levels attainable by additions
of benzene control technology. The benzene levels of individual gasoline blendstocks for each
refinery were also not available so they were they were estimated using the average benzene
levels in the LP refinery model. The benzene level of reformate was estimated using average
reformate benzene levels adjusted for the PADD-average severity and also adjusted by the
benzene characteristics of the type of reformer. As the severity of the reformer increases, it
produces a greater concentration of benzene in reformate. The Oil and Gas Journal contains
information on the type of reformer for each refinery in the U.S. The types of reformers are
semi-regenerative (semi-regen) reformers, cyclical reformers, and continuous reformers. Semi-
regen reformers operate the highest pressure of the three and as a result this type of reformer
tends to crack more of the higher molecular weight aromatics to benzene, resulting in a higher
benzene level in reformate. The second type of reformer is the cyclical reformer which operates
at a lower pressure than semi-regen reformers, and therefore causes less cracking of heavier
aromatic compounds to benzene. Continuous reformers are the lowest pressure reformers and as
a result cause relatively little cracking of heavier aromatic compounds to benzene. The benzene
level of heavy reformate varies based on presence of the heaviest portion of straight run naphtha,
which are the nine carbon compounds. Depending on the refinery, the nine carbon hydrocarbons
in straight run is either sent to the reformer, or is blended into jet fuel or diesel fuel. The
inclusion of the nine carbon hydrocarbons in reformer feed depends on the gasoline volume
calibration as described below. The inclusion of the nine carbon hydrocarbons in the feed to the
reformer tends to lower the concentration of benzene in the heavy part of reformate. The
assigned benzene content of gasoline blendstocks, including reformate, is summarized in Table
9.1-7.
9-23
-------
Final Regulatory Impact Analysis
Table 9.1-7. Estimated Benzene Content of Gasoline Blendstocks
Light Straight Run
Light Coker Naphtha
Light Naphtha (rerouted benzene precursors)
Natural Gasoline
Hydrocrackate
Alkylate
FCC Naphtha
Isomerate
Ethanol
Light Reformate (no benzene precursor rerouting)
Light Reformate (with complete benzene precursor rerouting)
Light Reformate (with benzene extraction)
Light Reformate (with benzene saturation)
Heavy Reformate - Semi-Regen (High Press.)
Cyclical (Medium Press.)
Continuous (Low Press.)
Heavy Reformate - High Press.
(with benzene Medium Press.
Extraction) Low Press.
Heavy Reformate - High Press.
(with benzene Medium Press.
Saturation) Low Press
PADDs 1-5 including CA
1.10
2.0
8.10
1.30
2.40
0.05
0.80
0.20
0.05
9.8
0.90
0.58
0.39
1.7-2.2
1.6-2.0
0.78-1.1
0.09-0.11
0.08-0.10
0.040-0.050
0.07-0.09
0.06-0.08
0.03-0.04
9.1.4.3 Calibration of the Refinery-by-Refinery Cost Model
The gasoline volume and benzene levels in the refinery-by-refinery cost model were
calibrated against actual gasoline volume and benzene levels.17 Refiners report their
conventional and reformulated gasoline volumes and benzene levels to EPA to comply with the
reporting provisions of the Reformulated Gasoline program. The 2004 gasoline quality was used
for calibrating the refinery model, which is consistent with the baseyear of the refinery-by-
refinery cost model. However, we could not begin to estimate how the various gasoline
blendstocks were used to blend up RFG and CG for those refineries which produce both, so we
aggregated them together for each refinery and calibrated both the gasoline volume and benzene
levels for each refinery's entire gasoline pool. Also, since most of the information used to
develop the refinery-by-refinery cost model was from summertime refinery modeling runs from
the LP refinery model, summertime gasoline volumes and benzene levels were used to calibrate
the refinery-by-refinery cost model.
Two different adjustments were used to calibrate the gasoline volumes in the refinery-by-
refinery cost model. The first adjustment increased or decreased the utilization of each gasoline
producing unit to adjust the gasoline volume higher or lower, respectively. The second
adjustment factor is applied when the gasoline volume is too high and it is used to reduce the
amount of nine carbon straight run naphtha processed by the reformer. The default in the
refinery model is that the nine carbon straight run naphtha is being sent to the reformer unit.
Therefore, if the initial gasoline volume in the refinery-by-refinery cost model is higher than
actual, adjustment factors are applied to decrease the utilization of each gasoline-producing unit
and reduce the volume of nine carbon feedstock sent to the reformer unit, thus adjusting each
9-24
-------
Final Regulatory Impact Analysis
refinery's estimated volume in the refinery-by-refinery cost model to equal the actual gasoline
volume.
To show the effects of these volumetric calibrations on the PADD volumes, the calibrated
crude oil consumption feed and the gasoline production volumes for each PADD are summarized
in Table 9.1-8.
Table 9.1-8. Calibrated 2004 and Projected 2012 Consumption and Production Volumes
for Crude Oil and Gasoline by PADD (kbbl/day)
Grade Oil
Consumed
(Kbbl/d)
Gasoline
Produced
(Kbbl/d)
Year
2004
2012
2004
2012
PADD1
1590
1574
841
879
PADD 2
3297
3403
1872
2081
PADD 3
7537
7789
3741
4148
PADDs 4, 5
1433
1589
652
718
The initial summertime benzene level of each refinery's gasoline estimated with the
refinery-by-refinery model was also calibrated against the reported annual average benzene
content of gasoline in 2004 from the RFG database. Unlike the straightforward adjustment used
for calibrating gasoline volume, adjusting each refinery's benzene level required one or more of
a series of different methods depending on the level of adjustment needed, the direction of the
adjustment and the processing units in each refinery. If the benzene level for a refinery in the
refinery-by-refinery cost model is higher than actual, and that refinery did not have a benzene
extraction nor a benzene saturation unit, then an adjustment was made to bypass benzene
precursors around the reformer. This is a likely strategy being employed today at refineries
producing RFG. However, we are aware that some conventional gasoline-producing refineries
are also using benzene precursor rerouting to comply with MSAT1. We therefore utilized this
strategy to calibrate the benzene levels for refineries producing either RFG or conventional
gasoline. If routing all the benzene precursors around the reformer did not lower the refinery
benzene level sufficiently to match the actual benzene level, then an additional step was taken
depending on the refinery. Refineries with isomerization units are assumed to route the rerouted
benzene precursor stream to that unit to the extent necessary to reduce the benzene down to the
actual level. The benzene levels of refineries without isomerization units are adjusted lower by
applying an adjustment factor to straight run and FCC naphtha benzene levels, thus lowering the
benzene content of each of these streams until the actual benzene level is achieved. If a refinery
had a benzene saturation or extraction unit and its benzene level is too high, the straight run and
FCC naphtha levels were adjusted lower until the actual benzene level is achieved.
If a refinery's initial benzene level in the refinery-by-refinery model is too low when
compared to its 2004 actual benzene level, two different adjustments were made depending on
the refinery's configuration. For a refinery without a benzene saturation unit or a benzene
extraction unit, its benzene level is adjusted higher by adjusting the straight run and FCC naphtha
benzene levels higher until the refinery's gasoline benzene level matched its actual benzene
level. For a refinery with a benzene saturation unit or a benzene extraction unit, its gasoline
9-25
-------
Final Regulatory Impact Analysis
benzene level is adjusted higher by reducing the utilization of its benzene saturation or its
extraction unit until its refinery gasoline benzene level matched its actual benzene level.
In Table 9.1-9, the refinery-by-refinery 2004 PADD-average benzene levels are
compared to the actual PADD-average benzene levels for 2004. We also show the projected
PADD-average benzene levels for 2012.
Table 9.1-9 Refmery-by-Refmery Model 2004 Calibrated and 2012 Projected
Benzene Levels by PADD versus 2004 PADD-actual Benzene Levels (vol%)
Actual Benzene
Levels
Refinery-by-
Refinery
Benzene Levels
Year
2004
2004
2012
PADD1
0.67
0.68
0.66
PADD 2
1.26
1.23
1.10
PADD 3
0.85
0.84
0.85
PADDs 4, 5
1.68
1.58
1.44
9.2 Cost Inputs for the Benzene Control Technologies
To estimate the cost of reducing refinery benzene levels, it was necessary to identify the
cost inputs of the identified benzene control technologies. This information was obtained from
vendors of these benzene control technologies or from the literature. This information was
updated from the proposed rule reflecting the detailed analysis we conducted to update the
capital costs. Information is presented for routing benzene precursors around the reformer,
routing that rerouted benzene precursor stream to an isomerization unit, and installing either of
two reformer post-treat technologies, which are benzene saturation and benzene extraction.
9.2.1 Benzene Precursor Rerouting
Routing benzene precursors around the reformer requires that a refinery add a naphtha
splitter distillation column, or modify an existing column, to make a distillation separation
between the six carbon and seven carbon hydrocarbons. As discussed in the RIA Section 6.2
above presenting our assessment of the feasibility of complying with this rulemaking, in a
refinery where most of the benzene precursors are not currently being routed around the
reformer, the naphtha splitter would need to be added or modified to be able to make a fairly
clean cut between the six and seven carbon molecules. Making this cut efficiently is important
in separating as much of the six carbon compounds (which include benzene) from the rest of the
heavy straight run naphtha as possible, so that the seven carbon and heavier straight run
hydrocarbons can continue to be sent to the reformer. A new unit would require the addition of a
new naphtha splitter distillation column. Modifying the naphtha splitter distillation column
involves increasing the height of the existing column and adding additional distillation trays or
replacing the distillation tower with a taller unit. The naphtha splitter modification would also
mean that the utility demands of that unit would increase. Conversely, the utility demands of the
reformer decreases as the six carbon compounds are withdrawn from that unit. The estimated
capital cost and increased utility costs for adding a naphtha splitter to facilitate routing benzene
precursors around the reformer is summarized in Table 9.2-1.18 We also summarized the utility
9-26
-------
Final Regulatory Impact Analysis
demands of the reformer in Table 9.2-2 because this information is used to calculate the reduced
utility demands when the benzene precursors are withdrawn from that unit.19
Table 9.2-1. Cost Inputs for Rerouting Benzene Precursors (2004 dollars)
Capital Costs - onsite and offsite ($MM)
Capital Cost Unit Size (bbl/day feedstock)
Catalyst ($/bbl)
Natural Gas (foeb/bbl)
Electricity (kwh/bbl)
9.6
15,000
0.01
0.010
2.80
Table 9.2-2. Cost Inputs and Light Gas Outputs for the Reformer
(Severity 95 RON 2004 dollars)
Catalyst Cost ($/bbl)
Fuel Gas (foeb/bbl)
Electricity (kwh/bbl)
Steam (Ib/bbl)
Hydrogen (foeb/bbl feed)
Plant Gas (foeb/bbl feed)
Propane (bbl/bbl feed)
Isobutane (bbl/bbl feed)
Butane (bbl/bbl feed)
0.354
0.044
2.6
75
0.036
0.029
0.036
0.017
0.028
9.2.2 Isomerizing Rerouted Benzene Precursors
Sending the rerouted benzene precursors to an existing isomerization unit is another
technology identified for further reducing gasoline benzene levels. The rerouted benzene
precursor stream contains naturally occurring benzene from crude oil. The isomerization unit
saturates the benzene in this stream, causing a further reduction in gasoline benzene levels. The
saturation occurs in the isomerization reactor which is designed to convert straight chain
compounds to branched chain compounds. So while the isomerization unit reduces the octane of
this stream by saturating benzene, it also offsets some of the octane loss by producing branched
chain compounds from the saturated benzene. The isomerized six carbon stream is estimated to
have an octane value of 77.4 (R+M)/2, compared to a similar octane value for the rerouted
benzene precursor stream before it. Many refineries have isomerization units today and for this
analysis, refiners are assumed to only rely on these existing units at their present capacity for
benzene reductions and not build a new isomerization unit nor increase an existing unit's
capacity.d In this analysis the rerouted benzene precursors are sent to the isomerization unit
which has been treating five carbon hydrocarbons. If the isomerization unit does not have
sufficient capacity to treat the volume of both the five and six carbon hydrocarbons, the
preference is given to benzene reduction and treating the six carbon hydrocarbons, and the five
carbon hydrocarbons are removed as necessary to make room for the six carbon hydrocarbons.
d Isomerizing straight run naphtha increases its vapor pressure. Many refiners today are vapor pressure
limited and face having to substantially cut its gasoline production volume if its gasoline were to increase in vapor
pressure. Since we do not know which refineries are in this situation, we assume that additional isomerization
capacity beyond that already present in the refinery would not be tolerated.
9-27
-------
Final Regulatory Impact Analysis
Therefore, for some refineries the increased utility costs for treating the rerouted benzene
precursors is based on the capacity of the isomerization unit instead of the total volume of five
and six carbons hydrocarbons fed to the unit, since some of the five carbon hydrocarbons are
backed out of the unit. Table 9.2-3 shows cost figures used in modeling isomerization of
rerouted benzene precursors.20
Table 9.2-3. Cost Inputs for Sending the Rerouted Benzene Precursors to an
Isomerization Unit (2004 dollars)
Catalyst ($/bbl)
Hydrogen (foeb/bbl)
Natural Gas (foeb/bbl)
Plant Gas (bbl/bbl)
Electricity (kwh/bbl)
Steam (Ib/bbl)
0.25
0.002
0.009
-0.024
0.90
50
9.2.3 Benzene Saturation
Benzene saturation is another technology which reduces the benzene content of gasoline.
The advantage that benzene saturation has for benzene reduction is that it treats the naturally
occurring benzene as well as the benzene formed in the reformer. The benzene formed in the
reformer includes the benzene formed from the cracking of heavy aromatics to benzene as well
as that formed by the conversion of six carbon hydrocarbons. The benzene saturation technology
involves the addition of a distillation column called a reformate splitter and then the benzene-rich
stream is sent to a benzene saturation unit.
The distillation column creates a benzene rich stream which prevents other aromatics,
such as toluene, from being sent to the benzene saturation unit. Keeping the toluene and xylenes
out of the benzene saturation unit preserves the octane level of the seven carbon and heavier
reformate. Based on information we received from vendors who are experts on benzene
saturation technology, the reformate splitter is typically optimized to capture 96% of the
benzene, while only capturing 1% of the toluene. We programmed our refinery-by-refinery cost
model so that the reformate splitter captures benzene and toluene consistent with this
information. For those refineries estimated to be currently routing some or all of the benzene
precursors around the reformer, for modeling the cost of benzene saturation, those benzene
precursors are sent to the reformer before the costs of applying benzene saturation are estimated.
The benzene-rich stream is sent to the benzene saturation unit. In the benzene saturation
reactor, hydrogen is reacted with benzene which converts the benzene to cyclohexane. There are
two benzene saturation technologies. One is called Bensat and is licensed by UOP. This
technology maintains the reformate splitter and benzene saturation units as separate discrete
units. The other benzene saturation technology is licensed by CDTech and is called CDHydro.
The CDHydro technology combines the distillation column and benzene saturation reactor
together into a single unit. The advantage of this approach is that it eliminates the need for the
second unit, potentially lowering the capital costs. A review of the capital cost inputs of the two
benzene saturation technologies shows much lower capital costs. When we considered the
9-28
-------
Final Regulatory Impact Analysis
difference in capital costs, the CDHydro capital costs seemed much lower than expected
considering the efficiency provided by the combined units. For this reason, we assigned the
CDHydro unit the same capital costs as the conventional benzene saturation unit. For both
benzene saturation technologies, the capital costs are scaled using a 0.65 scaling factor which
increases the per-barrel capital costs for smaller extraction units than the standard size, and
decreases the per-barrel capital costs for larger extraction units than the standard size. The
capital and utility costs and scaling factor used for both Bensat and CDHydro are summarized in
Table 9.2-4.21 22 23
Table 9.2-4. Cost Inputs for Benzene Saturation (2004 dollars)
Inputs
Capital Cost - onsite and offsite ($MM)
Capital Cost Unit Size (bbl/day feedstock)
Capital Cost Scaling Factor
Hydrogen (foeb/bbl)
Natural Gas (foeb/bbl)
Electricity (kwh/bbl)
Steam (Ib/bbl)
Bensat
20.9
8,400
0.65
0.046
-
2.5
197
CDHydro
20.9
8,400
0.65
0.046
0.016
0.80
-
As discussed below in the summary of costs, benzene saturation is the highest cost
benzene control technology modeled for this final rulemaking. The primary reason for this is
that after processing the straight run naphtha in the reformer to create the benzene for blending
into gasoline as high octane blendstock, this process converts it back to a low octane blendstock.
The process is desirable from the standpoint that it achieves deeper benzene reductions and its
cost is acceptable for larger refineries that can take advantage of their better economies of scale.
9.2.4 Benzene Extraction
Benzene extraction is the final benzene reduction technology used in our cost analysis for
estimating benzene control costs. Benzene extraction physically and chemically separates
benzene from the rest of the hydrocarbons, and then concentrates the benzene into a form
suitable for sale into the chemicals market. Since this process results in a benzene product
stream which must be transported to a buyer, a refiner is unlikely to choose this technology
unless there is economical access to a benzene market.
The first step involved in benzene extraction is the separation of a benzene rich stream
from the rest of the reformate using a reformate splitter. To maximize the removal of benzene
with this technology, any benzene precursor rerouting that is occurring in the basecase is
eliminated prior to costing out this technology, allowing the removal of naturally occurring
benzene. Not only does this further reduce the benzene in the final gasoline, it improves the cost
effectiveness of benzene extraction by improving the economies of scale for the benzene
extraction unit. The benzene-rich stream off the reformate splitter is sent to an extraction unit
which separates the aromatic compounds from other hydrocarbons contained in the benzene-rich
stream using a chemical extraction agent. While the intent is to have benzene as the only
aromatic in the benzene-rich stream, in reality some toluene is also contained in that stream as
well. For this reason, a very precise distillation step is conducted concurrently on the product
9-29
-------
Final Regulatory Impact Analysis
that produces a pure chemical grade benzene product. The desire would be to send only benzene
and no toluene to the benzene extraction unit, however, this would require an unreasonably large
and expensive reformate splitter. Thus, we used the same assumption used for benzene
saturation, which is that 96% of the benzene and 1% of the toluene is captured by the reformate
splitter. The concentration process of benzene for the petrochemicals market also assumes the
use of a clay treater.
The total capital costs for benzene extraction include the capital costs for the installation
of a reformate splitter, a benzene extraction unit and the associated distillation hardware which
concentrates the benzene, including a clay treater. The capital costs for the benzene extraction
unit assumes that the extraction and distillation step occur in one step, which is called extractive
distillation. For new benzene extraction units, additional capital costs are incurred for the
installation of benzene storage and loading equipment. The capital costs for new extraction units
are scaled exponentially using a 0.65 scaling factor. The capital costs for revamped extraction
units are not scaled which provides the same per-barrel capital costs regardless of the size of the
expansion.6 Utility costs are incurred for operating the benzene extraction units. Table 9.2-5
contains the capital and utility cost inputs to the refinery-by-refinery cost model for benzene
extraction.24
Table 9.2-5. Cost Inputs for Benzene Extraction (2004 dollars)
Capital Costs - onsite and offsite ($MM)
Capital Cost Unit Size* (bbl/day product)
Capital Cost Scaling Factor
Catalyst ($/bbl)
Natural Gas (foeb/bbl)
Electricity (kwh/bbl)
Steam (Ib/bbl)
65.9
1800
0.65
0.354
0
9.4
1271
* Capital Cost is based on the volume of benzene produced.
A refiner with an extraction unit in one of their refineries has informed us that they
frequently extract the benzene from benzene-rich reformate streams provided by other U.S.
refineries as well as streams from abroad. This helps offset the high capital costs associated with
these units. Because of the high capital costs, other refiners are hesitant to install an extraction
unit, but have sufficient octane production capacity to sell benzene-rich reformate to a
neighboring refinery which does extract benzene. For our year 2004 basecase analysis, we have
deduced that several refineries without an extraction unit or a benzene saturation unit, but with
already very low benzene levels (which cannot be easily explained on other bases), are selling
benzene-rich reformate to a neighboring refinery with an extraction unit. For modeling the cost
of additional benzene control, we also assume that refineries which already have an extraction
e Typically, the capital costs for revamping an existing refinery unit are not scaled. They are not scaled
because small expansions to existing refinery units require the redesign of only a part of an existing refinery unit to
realize the usually small increase in production capacity. This is in contrast to very small grassroots units of the
same volume as the expansion which requires the design and construction of every piece of equipment involved in
the unit being designed. Thus the small grassroots unit needs to be scaled to capture the higher capital costs while
the capital costs of revamps are estimated consistent with the per-barrel costs of a full sized unit.
9-30
-------
Final Regulatory Impact Analysis
unit would process the benzene rich reformate of other refineries to comply with the benzene
program.
9.3 Benzene Market and Prices
Benzene which is generated by benzene extraction and sold into the chemicals market is
an important output from the refinery-by-refinery cost model. The economics for benzene
extraction are partially dependent on the revenue earned through the sale of chemical grade
benzene. To understand the production and demand for benzene and the projected price of
benzene, we purchased Chemical Market Associates Incorporated (CMAI) 2004 report entitled
the World Benzene Analysis.25 The CMAI report lists the benzene producers and consumers
worldwide and analyzes the economics of benzene production.
Benzene is produced to sell into the chemicals market by 8 different types of benzene
production processes. These include extraction from reformers and pyrolysis gasoline at
refineries and petrochemical plants, selective toluene disproportionation, paraxylene
coproduction, toluene hydrodealkylation and extraction from coke oven naphtha. Except for the
production of benzene from coke ovens, the rest of the benzene is sourced from crude oil. The
World and U.S. production volumes of benzene for 2002, the most recent year that complete
information is available from the CMAI report, are summarized in Table 9.3-1.
Table 9.3-1. 2002 Benzene Supply by Source for U.S. and the World
(thousand metric tons)
U.S.
World
Reformate
3,527
13,213
W
a
Ł
2,086
12,699
S §•
s 1
•Q -S3
H Q
149
353
Selective
Toluene
Disprop.
810
1171
Paraxylene
Coprod.
529
1458
Toluene
Hydrodealk
317
2202
-------
Final Regulatory Impact Analysis
Table 9.3-2. 2002 Benzene Demand by Target Chemical for U.S. and the World (thousand
metric tons)
U.S.
World
g
^ N
*Ł §
w m
4050
18,201
y
g
O
2291
5872
§
,Ł}
P
•g
Ł
752
2200
1
o
o
964
4257
,
-------
Final Regulatory Impact Analysis
prices. Benzene prices remained about the same in 2005, but gasoline prices increased therefore
decreasing benzene's relative price compared to gasoline's price.
CMAI used its economic model to project the benzene market in the medium term during
the future years from 2006 through 2015. CMAI starts by establishing a basecase which was
based on the information on the benzene market in 2005. CMAI then projects the benzene
market based on anticipated supply, demand and energy prices. The benzene supply which
CMAI considers in its cost model includes existing benzene production capacity and announced
and planned new benzene plant construction. The future benzene demand is estimated based on
historical demand, the projected U.S. and world economic conditions, and on the anticipated
changes in the chemical markets which use benzene as a feedstock. After conducting its benzene
market review, CMAI made a series of conclusions. In its 2004 report, CMAI projected that
World benzene and U.S. benzene demand would increase annually at a very robust rate of 3.8
and 2.4 volume percent, respectively. Imports which satisfied just more than 10 percent of U.S.
demand in 2003, is expected to be flat and even decline in the out years. CMAI explains that the
robust world benzene demand coupled with new benzene production, which is expected to be
slow coming on line, will result in continued high benzene prices in 2007. As additional benzene
production capacity comes on line, benzene prices are expected to come down to more moderate
levels. The projected energy prices which CMAI uses in its economic model are nearly identical
with those used by EIA, thus making the two analyses consistent in this regard. Table 9.3-4
summarizes the projected benzene and gasoline prices obtained from CMAI's 2005 benzene
market projections through 2015. For 2011 through 2015, CMAI provided their projected
benzene prices, but not the gasoline prices. We projected the gasoline prices for 2011 through
2015 using the crude oil prices provided by CMAI using the relationship of crude oil prices to
gasoline prices from the previous years.
Table 9.3-4. Projected U.S. Benzene/Gasoline Price Differential
Year
2007
2008
2009
2010
2011
2012
2013
2014
2015
Crude Oil Price
($/bbl)
69
57
52
48
47
48
50
51
52
Benzene Price
($/bbl)
128
102
93
87
85
87
88
90
93
Gasoline Price
($/bbl)
79
65
59
55
54
55
56
58
60
Benzene Price above
Gasoline Price
($/bbl)
50
37
35
33
32
32
32
33
34
The CMAI model estimates that the price of benzene in 2007 will be $50 higher than
gasoline assuming that crude oil prices will stay high through 2007. CMAI projects that the
price of crude oil will decline after 2007. As the projected crude oil price declines, both gasoline
and benzene prices are also expected to decline resulting in benzene's price above gasoline to
decrease to about $32 per barrel above the price of gasoline. CMAI's projected crude oil price is
9-33
-------
Final Regulatory Impact Analysis
$48 per barrel in 2012, which is consistent with the crude oil price assumed for our refinery
modeling.
There may be a concern that the additional benzene that would be extracted from gasoline
and sold into the chemical benzene market in response to this rulemaking could depress the
benzene price below that projected by CMAI. To address this concern we used the projected
volume of benzene extracted from gasoline by the refinery-by-refinery model to evaluate the
impact of the additional benzene supply on benzene price. The refinery-by-refinery cost model
projects that about 12,500 barrels per day, which is 192 million gallons per year, of benzene
would be extracted from gasoline and sold to the petrochemical market under the benzene
program assuming that it took effect in 2012.
Table 9.3-3 above shows that the U.S. demand for chemical grade benzene in 2002 was
8450 metric tons, which is equivalent to 2529 million gallons. Based on an annual growth rate
of 2.4 percent, the U.S. demand for benzene is expected to be 3,000 million gallons in 2010 and
is expected to grow to 3,130 million gallons in 2011. Thus, the increase in U.S. benzene demand
from 2010 to 2011 is projected to be 130 million gallons. We expect the extraction of benzene
would occur over several years due to the effect of the ABT program. Therefore, the increased
production of chemical grade benzene due to extraction would be smaller than the annual growth
over the several years that the program phases in and no significant impact on benzene price
would be expected. Even if all of the benzene extraction capacity were to be installed in a single
year resulting in all 192 million gallons of benzene coming into the benzene market in one year,
the benzene production market could rebalance by the reduced processing of toluene into
benzene, the highest cost process for producing benzene. The toluene would remain in the
gasoline pool helping to maintain the octane and volume lost by benzene extraction. Finally,
refining and petrochemical market experts who evaluated the effect of the benzene extraction
expected to occur in response to the Reformulated Gasoline Program came to a similar
T7 0Ł
conclusion despite the large volume of benzene extracted back then. While we don't expect
a significant impact on benzene price, we rounded the incremental benzene price down to $30
dollars higher than gasoline. This incremental benzene price is slightly lower than CMAI's
projected incremental price to account for a small decrease in benzene price caused by the
increased benzene supply caused by this rulemaking.
9.4 Refinery Modeling of Benzene Control Scenarios
For modeling the benzene program, we addressed the costs and benzene impacts of the
maximum average standard first. Refineries that the model estimates would be above the
maximum average standard are assumed to put in the most cost-effective benzene reduction
technology which the model shows would get them below the maximum average standard.
Under the ABT program, the benzene control units that the model adds to meet the maximum
average standard are assumed to be operated to achieve the maximum possible amount of
benzene reduction. The benzene reductions associated with meeting the maximum average
standard may or may not be sufficient for meeting the average standard depending on how
stringent the maximum average standard is relative to the average standard. If additional
9-34
-------
Final Regulatory Impact Analysis
benzene reduction is necessary, it is achieved in the cost model consistent with the methodology
used to achieve benzene reductions under the average standard only.
If additional benzene reductions are needed after application of the maximum average
standard, or if we were not modeling a maximum average standard, we modeled benzene
reductions to meet the average benzene standard. The national ABT program optimizes the
benzene reduction by allowing the refining industry to collectively choose the most cost-
effective means of benzene reduction. In the refinery-by-refinery modeling, this is accomplished
by ranking the benzene reduction technology available to each refinery and over all the refineries
in order from lowest to highest in benzene reduction cost-effectiveness. Then refineries are
chosen to implement benzene reduction refinery-by-refinery from the lowest to the next lowest
in benzene control cost-effectiveness until the sum of the technologies and refineries chosen
results in the U.S. gasoline being produced meeting the benzene program being modeled, giving
credit to refineries already below the proposed benzene standard.
For the benzene control cases we modeled that do not include an ABT program, all the
refineries that are below the standard are assumed to maintain their current benzene level, while
the refineries with benzene levels above the standard are assumed to take the necessary steps to
reduce their benzene levels down to the standard. If the model shows that capital investments
need to be made to achieve the necessary benzene reduction, we assumed the installation of a full
sized unit is installed to treat the entire stream being treated, but assumed further that the unit is
only operated to the extent necessary to meet the applicable standard.
9.5 Evaluation of the Refinery-by-Refinery Cost Model
As described in the Overview Portion of this section, the refinery-by-refinery cost model
was evaluated to assess its viability by comparing its cost output to the cost output of the LP
refinery cost model. The LP refinery cost model is a good tool for comparison because it has
been used for many years on many different cost studies subjecting it to extensive peer-review.
We evaluated the benzene program with the LP refinery model to estimate the energy and
supply impacts of the benzene program. We specified the mix of benzene control technologies
that the refinery-by-refinery cost model estimates will be used in each PADD to comply with the
benzene program. We trust the refinery-by-refinery cost model's choice of benzene control
technologies because of its ability to estimate benzene control costs for each refinery and choose
the best mix of benzene control technologies across the refining industry. Because we matched
the benzene control technologies and final benzene levels in each PADD a close match in control
costs between the two models would confirm that the refinery-by-refinery cost model is sound in
its construction. Comparing the cost output of the two cost models, the LP refinery cost model
produced very similar costs compared to the refinery-by-refinery cost model, which corroborates
the refinery-by-refinery cost model. Table 9.5-1 summarizes the cost output and estimated
benzene levels for the two refinery modeling analyses.
9-35
-------
Final Regulatory Impact Analysis
Table 9.5-1. Comparison of PADD and National Costs and Benzene Levels for the Benzene
Program (2003 dollars, 7% ROI before taxes)
Refinery -by -Refinery
Cost Model
LP Refinery Cost
Model
Cost (cents/gal)
Bz Level (vol%)
Cost (cents/gal)
Bz Level (vol%)
PADD 1
0.15
0.52
0.16
0.52
PADD 2
0.34
0.63
0.27
0.63
PADD 3
0.16
0.61
0.14
0.61
PADDs4
&5
0.91
0.78
0.92
0.76
U.S.
Average
0.27
0.62
0.24
0.62
9.6 Refining Costs
This subsection summarizes the estimated costs of the benzene program as well as the
other benzene standards considered for this final rulemaking. The estimated cost for the 0.62
vol% benzene standard with 1.3 maximum average standard and ABT program is summarized
first, including the sensitivity cases described above. We next summarize the estimated cost for
the same and higher and lower average benzene standards, with and without various maximum
average standards or which models a benzene program without an ABT program. We adjust our
costs from 2004 dollars back to 2003 dollars to make our costs consistent with the gas can and
vehicle costs. To make this cost adjustment we used a 0.97 inflation cost factor from the
Department of Labor webpage.
The capital costs estimated by the refinery-by-refinery cost model do not include the
capital costs associated with hydrogen production and octane recovery. For this reason we
believe that the capital costs estimated by the refinery-by-refinery cost model are low. We
compared the capital cost estimate by the LP refinery cost model, which includes the hydrogen
and octane capital costs, and found them to be about 20 percent higher than those estimated by
the refinery-by-refinery cost model. For all the capital cost estimates for all the benzene
programs evaluated by the refinery-by-refinery cost model, we adjusted them higher by 20
percent.
9-36
-------
Final Regulatory Impact Analysis
9.6.1 Cost of the Benzene Program
The refinery-by-refinery cost model was used to estimate the cost of the benzene
program, which puts in place a 0.62 vol% average benzene standard with a 1.3 maximum
average standard and an ABT program. For each of the refineries which produce gasoline, the
methodology described above was applied to estimate the cost of reducing the benzene levels.
The projected use of the benzene technologies in the refinery-by-refinery cost model is affected
by the nature of the stringency of the benzene reduction program being modeled. The refinery -
by-refmery cost model indicates that benzene precursor rerouting alone is the most cost effective
benzene control technology, followed by routing the benzene precursors to an isomerization unit
and extraction. Benzene saturation is the least cost-effective benzene control technology, but as
the benzene control stringency is increased, for reasons of technical feasibility benzene
saturation replaces benzene precursor rerouting with or without isomerization as the means for
achieving benzene reductions. We assume that the ABT program would be fully utilized with
credit trading occurring freely within and between refining companies.
The fully phased-in 0.62 vol% benzene standard with 1.3 maximum average standard and
ABT program is estimated to cost 0.27 cents per gallon averaged over all U.S. gasoline and with
capital costs amortized at 7% ROI before taxes. The total capital cost is estimated to be $1110
million; the total annual cost including amortized capital costs is $330 million/yr estimated in the
year 2012.
The 0.27 cents per gallon average cost is calculated by amortizing the costs over all U.S.
produced gasoline including that gasoline volume with benzene levels already at or below 0.62
vol%. When the costs are averaged only over the portion of U.S gasoline which is expected to
be reduced in benzene, the fully phased-in benzene program is expected to cost 0.40 cents per
gallon. For those refineries which are projected to take some action to reduce their benzene
levels, the average capital and total annual operating cost per refinery is $14 million and $4.2
million, respectively. These estimated costs for the benzene program are summarized in Table
9.6-1.
Table 9.6-1. Estimated Costs of the Fully Phased in Benzene Program Evaluated in 2012
(2003 dollars, 7% ROI before taxes)
All Refineries
Refineries
Reducing Their
Gasoline Benzene
Levels
Number of Refineries
Total Capital Cost ($ million)
Total Annual Cost ($ million/year)
Per-Gallon Cost (cents/gallon)
Number of Refineries
Capital Cost per Refinery ($ million)
Operating Cost per Refinery ($ million/year)
Per-Gallon Cost (cents/gallon)
104
1110
330
0.27
79
14
4.2
0.40
Reporting the average per-gallon costs in the above table does not provide any indication
of the range in costs that we project would occur in different refineries. The costs vary by
refinery for a variety of reasons. First, some refineries experience no cost because either the
gasoline produced by those refineries is already below the benzene standards, or (with respect to
9-37
-------
Final Regulatory Impact Analysis
the 0.62 vol% average benzene standard), our modeling shows that these refineries would
experience lower costs by simply purchasing credits. Another reason why refineries are
projected to experience differing costs is due to the range in technologies that they would use and
the extent of benzene reduction achieved by them. The final reason why these refineries are
projected to experience differing costs is due to the different refinery economies of scale and cost
inputs in different refining regions. Figure 9.6-1 summarizes the projected per-gallon costs by
refinery plotted against the cumulative volume of gasoline produced. The figure shows that we
project costs to be low for most refineries, representing most of the gasoline production in the
country; a relatively few higher-cost refineries contribute significantly to the higher average cost
of the program.
Figure 9.6-1. U.S. Refinery Per-Gallon Costs for the Benzene Program
(2003 dollars, 7% ROI before Taxes)
7.00
6.50
6.00
5.50
5.00
— 4.50
|) 4.00
3 3.50
o 3.00
-2- 2.50
-0.50
Cumulative Volume (thousand barrels/day)
To comply with the benzene program, we expect that all of the control technologies
discussed above would be utilized. Of the 79 refineries expected to take steps to reduce their
gasoline benzene levels, 17 are expected to route all of the benzene precursors around the
reformer, and 28 are expected to send that rerouted stream to their isomerization unit. Of the
refineries which take steps to lower their gasoline benzene levels by treating reformate, 16 would
install a grassroots benzene extraction unit or revamp their existing extraction units while the
another 18 would install benzene saturation units. We project that 52 refineries will continue to
produce gasoline with benzene levels greater than the 0.62 vol% average standard and will
choose to purchase credits to comply with that standard. Including the refineries with benzene
levels currently below 0.62, we project that there will be a total of 50 refineries that will produce
gasoline with benzene levels at 0.62 or lower and will generate credits for sale to other refineries.
9-38
-------
Final Regulatory Impact Analysis
Finally, based on our modeling, we project that 26 refineries will not take steps to reduce their
gasoline benzene levels to comply with the 0.62 and 1.3 vol% benzene standards.
While the estimated per-gallon costs are very low, there is a range in costs depending on
the area of the country (again primarily reflecting the degree of benzene reductions as well as the
ability to extract and sell the extracted benzene). The estimated costs in PADDs 1 and 3 are
lowest due to the expected use of extraction (with sale of the recovered benzene). The estimated
benzene control costs are higher for the rest of the PADDs because extraction was not assumed
to be an option due to lack of benzene markets. The average per-gallon benzene control costs for
each PADD are summarized in Table 9.6-2.
Table 9.6-2. Per-Gallon Costs by PADD for the Benzene Program
(cents/gal; 2003 dollars; 7% ROI before taxes)
PADDl
0.15
PADD 2
0.34
PADD 3
0.16
PADD 4
0.69
PADD 5 except CA
1.11
In each PADD, the average costs in Table 9.6-2 represent a wide range in costs across the
refineries in the PADD. However, the nature of the cost range varies in each PADD based on the
factors described above. Figure 9.6-2 depicts the estimated per-gallon costs by refinery in each
PADD plotted against the cumulative gasoline production.
9-39
-------
Final Regulatory Impact Analysis
Figure 9.6-2. U.S. Refinery Per-Gallon Costs by PADD for the
Benzene Program
(2003 dollars, 7% ROI before Taxes)
500 1000 1500 2000 2500 3000 3500 4000 4500
Cumulative Volume (thousand barrels/day)
Figure 9.6-2 shows a significant range in costs by the refineries in each PADD. Costs for
most refineries inPADDs 1 and 3 are similar with most costs being incurred through extraction
which results in near zero (and in a few cases slightly less than zero) costs, as well as zero costs
for refineries which do not need to take any action due to already low gasoline benzene levels.
Most of the refineries in PADDs 4 and 5 face higher costs, and these costs are significantly
higher than the costs for most refineries in the other PADDs due to the generally smaller
refineries there and the inability to use extraction. The refinery costs in PADD 2 are more
moderate for most of the refineries than those in PADDs 4 and 5, but still more severe than the
costs for most of the refineries in PADDs 1 and 3.
In each PADD there are smaller-sized refineries which the model predicts would need to
comply with the maximum average standard using benzene saturation, resulting in high per-
gallon costs. The costs for these refineries are high because of their poorer economies of scale.
The model does not attempt to apply other means likely to be available to these refineries for
9-40
-------
Final Regulatory Impact Analysis
avoiding these high per-gallon costs. We believe that these refiners can avoid resorting to
benzene saturation and their associated high per-gallon costs by installing a reformate splitter.
The reformate splitter is a relatively low capital and operating cost unit that would allow them to
remove a benzene-rich stream from the rest of their reformate, resulting in a final gasoline blend
that would be in compliance with the maximum average standard. The benzene-rich stream can
be sold to another refinery with gasoline benzene levels below the cap standard and so can
absorb this small benzene-rich volume. This sort of trading is similar to the credit trading
program, except that actual benzene is being traded instead of paper credits/
The rule also includes hardship provisions, available to all refineries, to address extreme
hardship situations. The model assumes full compliance without hardship relief, and so may
overstate costs for this reason as well.
Our refinery modeling analysis projects that the ABT program will effectively result in a
phase-in of the benzene program from 2007 through early 2015. Starting in mid-2007 we
believe that using simple operational changes refiners will take the opportunity to achieve
modest benzene reductions to generate early credits. We project that these actions taken in mid-
2007 will result in a reduction of the average U.S. gasoline benzene level from 1.00 to 0.81
volume percent at an average cost of 0.04 cents per gallon averaged over all U.S. gasoline.
To take full advantage of the flexibility provided to refiners by the ABT program to delay
more expensive capital investments, refiners are expected to make additional early benzene
reductions to generate more early credits, requiring modest investments in capital. Because of
the time it takes to assess, design and install the capital equipment, we believe that these
additional early benzene reductions will not occur until the beginning of 2010. These benzene
reductions are expected to further reduce the average benzene level of U.S. gasoline to 0.74
volume percent and cost 0.05 cents per gallon averaged over all U.S. gasoline. Refiners are
expected to make $324 million of capital investments to achieve this benzene reduction. In 2011
when the 0.62 vol% average benzene standard takes effect, we do not anticipate any further
reduction in benzene because we project that the refining industry will be able to comply using
early credits.
In mid-2012, when refineries with high benzene levels need to comply with the 1.3
volume percent maximum average standard, we anticipate that U.S. gasoline benzene levels will
decline further, to 0.73 vol% benzene and cost an additional 0.04 cents per gallon averaged over
all U.S. gasoline. Refiners are expected to make another $153 million in capital investments to
comply with the 1.3 vol% maximum average standard which takes effect in mid-2012. Although
the early credit use period terminates at the end of 2013, refiners will be able to further delay
their most expensive capital investments by using standard credits (which will have been
accruing since the start of 201 l).g Because we expect that refiners will first use their early
credits, the standard credits will be banked and will start to be used in 2014 to show compliance
f Uncertainties in how trading of actual benzene barrels would occur precluded our modeling the cost of this
option. For example, we could not anticipate which refiners would be willing to accept this benzene-rich stream
which affects its transportation costs.
g Early credits generated or obtained and ultimately used by small refiners may be used through 2017. However,
these credits will not affect the overall implementation timeline discussed here.
9-41
-------
Final Regulatory Impact Analysis
with the 0.62 vol% average benzene standard. Our analysis suggests that the U.S. refining
industry will be able to delay their highest capital investments until May 2015 when the standard
credits accumulated since the beginning of 2011 run out. Small refiners must meet the 1.3 vol%
maximum-average standard beginning of July 2016 so they also will be reducing their gasoline
benzene levels to that standard or below.h Small refiners are expected to add an additional 0.01
cents per gallon averaged over all U.S. gasoline when reducing their gasoline benzene levels to
comply with their average and cap standards, although the average U.S. gasoline benzene levels
do not appear to change due to rounding. This additional benzene reduction is estimated to incur
an additional $26 million in capital investments. The nonsmall refiners are projected to fully
complete the transition in May 2015 bringing the average gasoline pool down to 0.62 vol%
benzene, and incurring a 0.13 cents per gallon cost averaged over all U.S. gasoline and $608
million in capital investments. The estimated cost savings of both the early and ongoing aspects
of the ABT program are summarized above in Section 6.5.5.12 where the impacts of the ABT
program are discussed.
We estimated the stream of total annual compliance costs for the U.S. refining industry
complying with the benzene program from 2007 to 2035, including the phase-in of the ABT
program. We used the per-gallon program costs to refiners in 2012 throughout the phase-in
period as well as the fully phased-in program, multiplying these estimated costs times the
projected gasoline demand by the Energy Information Administration (EIA) contained in the
Annual Energy Outlook (AEO) 2006. Since the EIA projections end at 2030, we used the annual
average growth rate over the years 2025 to 2030 to extrapolate the growth in demand to 2035.
The stream of projected gasoline consumption volume and the total annual costs for complying
with the benzene program are summarized in Table 9.6-3.
h Our analysis included values for small additional costs and emission reductions based on an assumption that the
start date for the 1.3 maximum average standard for small refiners would be 2015. Since the final rule sets this date
as July 2016, our 2015 results are slightly over-estimated.
9-42
-------
Final Regulatory Impact Analysis
Table 9.6-3. Stream of Total Compliance Costs for the Benzene Program
(2003 dollars, 7% ROI before Taxes)
Year
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
Gasoline Volume
(million gallons)
123,719
125,315
127,311
129,705
132,233
134,362
136,224
137,953
139,683
141,412
143,142
144,871
146,733
148,463
150,059
151,656
153,119
154,582
156,179
157,775
159,504
161,234
162,830
164,560
166,156
167,885
169,632
171,397
173,180
Cost
(c/gal)
0.02
0.04
0.04
0.09
0.09
0.11
0.13
0.13
0.27
0.27
0.27
0.27
0.27
0.27
0.27
0.27
0.27
0.27
0.27
0.27
0.27
0.27
0.27
0.27
0.27
0.27
0.27
0.27
0.27
Total Program Cost
(million dollars)
28
49
50
101
104
133
164
166
363
379
384
388
393
398
402
406
410
414
419
423
427
432
436
441
445
450
455
459
464
9.6.2 Cost of Alternative Benzene Programs
We used the refinery-by-refinery cost model to estimate the cost of other potential
benzene standards. This includes analyses of different maximum average benzene standards,
different averaging standards, and benzene standards with and without ABT programs.
Table 9.6-4 contains a summary of the national average per-gallon costs and aggregate
capital and total annual costs for maximum average benzene standards which range from 1.1 to
1.5 vol% benzene and average benzene standards which range from 0.50 to 0.71 vol%, with and
without an ABT program. The 0.50 vol% average benzene standard represents the most
stringent benzene standard technically feasible with maximum reformate control assuming that
either benzene extraction or benzene saturation would be used. For comparison, we also
modeled an average standard of 0.71 vo% benzene, but without the full ABT program. Each
refinery would have to average 0.71 vol% benzene across its own gasoline batches with no
ability to average or trade across refineries, or bank credits. This benzene standard is projected
to result in a national average benzene level which would equal the 0.62 vol% benzene standard
9-43
-------
Final Regulatory Impact Analysis
with full ABT - thus it is an interesting case to study relative to the benzene program. However,
the refinery model estimates that two refineries would not be able to achieve the 0.71 vol%
benzene standard based on reformate benzene control alone, thus it is not a perfect comparison.
Table 9.6-4. Cost of Other Benzene Standards
(2003 dollars, 7% ROI before taxes )
Average
Benzene Std.
(vol %)
0.50
0.60
0.60
0.62
0.62
0.62*
0.62
0.62
0.62**
0.65
0.65
0.70
0.70
0.71***
ABT
Program
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
Max-Avg
Std.
(vol %)
None
1.3
None
1.1
1.2
1.3
1.4
1.5
None
1.3
None
1.3
None
None***
Actual In-Use
Benzene Level
(vol %)
0.50
0.60
0.60
0.62
0.62
0.62
0.62
0.62
0.62
0.65
0.65
0.70
0.70
0.62
Per-Gallon
Cost
(cents/gal)
0.74
0.31
0.30
0.34
0.30
0.27
0.26
0.25
0.24
0.21
0.18
0.16
0.11
0.51
Total Annual
Cost
($ million/yr)
900
380
360
410
360
330
320
300
290
250
220
190
140
620
Aggregate
Capital Cost
($ million)
2140
1250
1180
1120
1070
1110
1100
1070
1120
950
960
740
510
1670
* Final Rule
** Proposed Rule
*** The 0.71 volume percent
standard, because without an
an annual average basis.
benzene standard we modeled could also be thought of being a maximum average
ABT program, each refinery would have to meet this level with actual production on
Our refinery model analysis shows that the reduced flexibility of adding a maximum
average benzene standard increases the cost of benzene control over a benzene control program
without a maximum average standard. We estimate that the reduced flexibility will require some
refiners to install a benzene saturation unit instead of routing the benzene precursors around the
reformer or sending that rerouted stream to an isomerization unit and procuring credits to make
up the remaining shortfall. As the table shows, these additional actions by some refiners and the
associated cost increases will not affect the in-use benzene level, which will be driven by the
0.62 vol% average standard regardless of the level of maximum average standard.
The benefit to the program of the 1.3 vol% maximum average standard is the increased
certainty that the benzene reductions projected by our modeling will in fact be achieved
nationwide, especially the significant reductions projected in areas that currently have the highest
benzene levels. Implementing a maximum average standard lower than 1.3 vol% would increase
the number of refineries that need to install the more expensive benzene reduction equipment.
This would diminish the value of the flexibility provided by the ABT program by requiring an
increasing number of refineries to make expenditures in benzene control that could otherwise be
smaller or avoided entirely, thereby increasing the overall costs of the program. Conversely, a
maximum average standard greater than 1.3 vol% would require progressively fewer refineries to
take action to reduce their benzene levels. This would in turn provide less assurance that actual
9-44
-------
Final Regulatory Impact Analysis
benzene levels would be broadly achieved and would undermine the greater degree of
geographic uniformity in benzene reductions achieved by the 1.3 vol% standard.
The 0.71 vol% benzene standard without the ABT program, which results in the same
national average gasoline benzene level as the benzene program, is estimated to cost almost two
times more than the benzene program. Without any ABT program, this standard offers the least
amount of flexibility compared to the benzene program. The lack of flexibility of this benzene
standard results in a larger share of benzene reductions occurring through benzene saturation, the
most expensive benzene control technology.
We plotted the projected per-gallon costs for each refinery producing gasoline (from
lowest to highest cost) versus the cumulative volume of gasoline across the refineries producing
gasoline for several benzene programs of interest. Figure 9.6-3 shows the per-gallon costs for
the final benzene program and a program with the same standard, but without a maximum
average standard. We also included a plot of the 0.50 vol% benzene standard which represents
maximum reformate benzene control technically achievable (albeit at significant higher national
cost, and with significant adverse economic impact on individual refineries).
Figure 9.6-3.
Cost Comparison between Final Benzene Standards and Two Other Options
(2003 dollars, and 7% ROI before taxes)
7
6
6
5
5
4
=- 4
TO
O
O
o
O
Final Rule 0.62 vol% Bz Std.
with ABT and 1.3 Max-Avg
Std.
0.62 vol% Bz Std. with ABT
but no Max-Avg std.;
(Proposed Rule)
0.50 vol%Bz Std.
representing maximum
reformate control
Cumulative Volume (thousand barrels/day)
9-45
-------
Final Regulatory Impact Analysis
Figure 9.6-3 shows that for nearly half the volume of gasoline, the costs for benzene
control are zero or near zero, and for a few extraction refineries even negative. The model
projects that the addition of the maximum average standard will require a small number of
refineries to adopt more expensive benzene control strategies. Comparing the proposed and final
programs, the final rule benzene program would cause 16 refineries to exceed 1 cent per-gallon
compliance cost compared to 8 refineries that would exceed 1 cent per gallon without a
maximum average standard. The 0.50 vol% benzene standard would be much more expensive in
this regard as it is estimated to cause about 60 refineries to exceed 1 cent per gallon in
compliance costs. Although it is difficult to determine this from the above figure, the refinery
with the highest cost of compliance under the final benzene program is estimated to incur about a
6.5 cents per gallon cost (same for the 0.50 vol% standard) while under the benzene program
without the maximum average standard the refinery with the highest cost of compliance would
be about 4 cents per gallon.
Table 9.6-5 below summarizes the number refineries which install or adopt each of the
four different types of benzene control technologies for:
• the final benzene program (0.62 vol% average benzene standard with 1.3
maximum average standard and ABT program,
• a 0.62 vol% benzene standard program with ABT program, but no maximum
average standard (proposed rule),
• a 0.71 vol% benzene standard without an ABT program which results in a 0.62
vol% average benzene level in gasoline, and
• a 0.50 vol% benzene standard with ABT program (maximum reformate benzene
control).
9-46
-------
Final Regulatory Impact Analysis
Table 9.6-5. Projected Number and Type of Benzene Control Technologies Installed for
the Final Benzene Standards and Other Options
Final Rule 0.62 vol%
avgBz std with 1.3
Max- Avg std and ABT
program
0.62 avg Bz std with
ABT Program , no max-
avg std (proposed rule)
0.71avgBzstd, No
ABT Program; 0.62
vol% in-use
0.50 avg Bz std with
ABT (maximum
reformate benzene
control)
Routing Benzene
Precursors Around
Reformer
17
19
1
0
Sending Rerouted
Benzene Precursors to
Isom Unit
28
28
12
0
New and
Revamped
Benzene Extraction
Units
16
17
25
63
Benzene
Saturation
18
8
52
24
Adding a maximum average standard or eliminating the ABT program altogether is
projected to result in a different pattern of benzene reduction across the country. Refineries
which we project will find it economically advantageous to realize only minor benzene
reductions and to primarily purchase credits to comply with the average benzene standard are
primarily located in PADD 4 and PADD 5. The refineries which we project will generate
credits under the ABT program are primarily located in PADDs 1 and 3. The model assumes
perfect trading of credits, so if an alternate program is projected to increase benzene reductions
in one area, the model would project that this increase would be offset by decreased benzene
reductions in other areas. For example, as shown in Table 9.6-6, the model projects that adding a
1.3 vol% maximum average standard should result in significant additional benzene reductions
in PADDs 4 and 5 and a small increased reduction in PADD 1, all of which would appear to be
offset by small decreases in benzene reductions in PADDs 2 and 3.
We note that the design of the refinery model and its inherent trading assumptions is such
that we can be much more certain that large projected changes will actually occur than we can
about small projected changes. Thus, while we are confident that adding the 1.3 vol% maximum
average standard will result in greater benzene reductions in PADDs 4 and 5 than would a
program without the 1.3 vol% standard, we cannot be certain that the small changes projected for
PADDs 1, 2, and 3 will occur or occur in the ways that the model projects. In addition to this
uncertainty about small modeled changes in benzene, some refiners may behave differently than
the model predicts. For example, it is not unlikely that some refiners in PADDs 2 and 3 will
choose to "over-comply" with the 0.62 vol% average standard (to provide a greater margin of
safety for compliance) regardless of the state of the benzene credit market. Yet the model would
tend to project that these refiners would reduce benzene levels as little as necessary. Thus, the
projected benzene levels achieved in PADDs 2 and 3 under a 0.62 vol% benzene standard
without a maximum average standard may well be achieved (or even exceeded) under the final
9-47
-------
Final Regulatory Impact Analysis
rule program with a maximum average standard if refiners choose to comply with a safety
margin. Table 9.6-6 summarizes the estimated benzene level by PADD for several different
benzene programs that would result in the same nationwide benzene level, but differing gasoline
benzene profiles because of the addition of the maximum average standard. We also show the
pattern of benzene control across the country for the 0.50 vol% benzene standard with ABT
program.
Table 9.6-6. Comparison of the 2004 and Modeled Gasoline Benzene Levels by PADD for
the Final Benzene Program and Other Options (vol% benzene)
Current Benzene Levels
Final Rule 0.62 vol% avg
Bz std with 1.3 Max- Avg
std and ABT program
0.62 vol% avg Bz std with
1 . 1 Max- Avg std and ABT
program
0.62 vol% avg Bz std with
1.5 Max- Avg std and ABT
program
0.62 avg Bz std with ABT
Program No Max- Avg
(Proposed Rule)
0.7 1 avg Bz std, No ABT
Program*
0.50 avg Bz std with ABT
(maximum reformate
benzene control)
PADD 1
0.67
0.52
0.55
0.52
0.53
0.53
0.50
PADD 2
1.26
0.63
0.61
0.63
0.61
0.70
0.45
PADD 3
0.85
0.61
0.63
0.60
0.60
0.59
0.52
PADD 4
1.56
0.90
0.83
0.90
0.94
0.71
0.53
PADD 5
excluding CA
1.80
0.69
0.55
0.82
0.88
0.70
0.48
U.S.
Average
1.00
0.62
0.62
0.62
0.62
0.62
0.50
* The cost analysis shows that 2 refineries would not be able to meet a 0.71 vol% benzene standard. These two
refineries would need to achieve the 0.71 vol% standard by reducing benzene levels in another gasoline stream.
To gain a sense of the relative benzene levels among all U.S. refineries, we plotted the
individual refinery benzene levels projected to result from several of the benzene programs with
average national benzene levels of 0.62 vol% benzene. A review of the refinery-by-refinery
output shows that the benzene levels of the refineries in PADD 4 and PADD 5 (excluding
California) are most likely to remain above the average standard with a nationwide ABT
program in place. The plots of the refinery benzene levels against cumulative gasoline
production for all U.S. refineries, and for all refineries in PADDs 4 and 5 (excluding California),
are contained in Figure 9.6-4, and Figure 9.6-5, respectively.
9-48
-------
Final Regulatory Impact Analysis
Figure 9.6-4. U.S. Final Rule Benzene Levels Compared to Benzene Levels
for 2004 and Other Control Options
4.50
g 4.00
| 3.50
| 3.00
1 2.50
j/T 2.00
0 1.50
0)
N
c
0)
DO
1.00
0.50
0.00
-2004 Benzene
Levels
-Final Rule Benzene
Prgm
- 0.62 w/A BT.no
M ax-Avg
-O.SOAvg Std, Max
Reformate Cntrl
O
O
O
O
O
O
CM
O
O
O
co
O
O
O
O
O
O
if)
O
O
O
to
O
O
O
h-
o
O
O
00
Cumulative Volume (thousand barrels/day)
9-49
-------
Final Regulatory Impact Analysis
Figure 9.6-5. PADD 4 and 5 Estimated Final Rule Benzene Levels Compared
to Benzene Levels for 2004 and Other Control Options
Q>
H
Q>
Q.
O
0)
0)
0)
Q>
N
C
0)
CD
4.50
4.00
3.50
3.00
2.50
2.00
1.50
1.00
0.50
0.00
•2004 Benzene Levels
• Final Rule Benzene Lvls
•0.62 w/ABT, no Max-Avg
-0.50 Avg. Sid., Max Reformate Cntrl
100 200 300 400 500 600 700
Cumulative Volume (thousand barrels/day)
800
All of the benzene standards represented in Figure 9.6-4 and Figure 9.6-5 would realize
substantial benzene reductions in all parts of the country compared to today's benzene levels. As
the benzene control standard is tightened or as flexibility is reduced, the curve for gasoline
benzene levels becomes flatter.
9.6.3 Costs Used to Estimate Price Impacts of the Benzene Program
In Chapter 13 of the RIA, we estimate the increase in gasoline prices for the benzene
program. To facilitate that analysis, certain cost information was obtained from the refinery-by-
refinery cost model and presented to the contractor conducting that analysis. The cost
information provided is consistent with specific macroeconomic principles that form the basis for
estimating price impacts.
When modeling macroeconomic effects, the price in any market can be assumed to be
based on the cost for the last, highest cost increment of supply which meets demand. We do not
know which refineries are the highest cost producers of gasoline, so we have estimated three
different cost breakpoints to capture the costs experienced by these price setter refineries. For
the first set of costs provided, we assumed that the highest cost gasoline producers also
experience the highest benzene control costs. The refinery-by-refinery cost model estimates the
compliance cost for individual refineries so we simply sorted through the list of individual
refinery costs and picked the highest cost of compliance in each PADD, which is the market area
we chose to use for evaluating price effects.
9-50
-------
Final Regulatory Impact Analysis
We developed other cost information to capture other ways that this program could
impact prices. Perhaps, the price setting refineries are not experiencing the maximum benzene
control costs, or maybe they are affected by other factors. Refineries produce in a wide range of
markets. Since the products are produced from the same feedstock with limited flexibility for
changing the product slate, market prices for individual products are not independent of each
other. Being the highest cost producer for one product does not mean they are the highest cost
producer for all products, and market prices won't necessarily reflect their costs. To capture
these other possible market effects, two other sets of cost information are provided to our
contractor for estimating price effects.
The second set of costs we developed is based on the maximum variable costs
experienced in each PADD. These costs do not include the capital costs and could also represent
another situation based on claims made by the representatives of the oil industry. They have said
that after complying with the 500 ppm highway diesel fuel sulfur standard, the price increase in
highway diesel fuel after that rule went into effect did not support their recovering their capital
costs. We could not confirm this claim, but providing the maximum variable costs would
attempt to model this situation.
For the third set of costs, we provided the average cost of compliance in each PADD.
Since the highest benzene control costs may not necessarily correlate to the refineries with the
highest overall gasoline production costs this case simply assumes the highest cost gasoline
producer experiences average benzene control costs. Estimating the average cost of compliance
for the fuel consumed is more complicated because the gasoline consumed in any area is a
function of the imports and transfers into the PADD as well as the gasoline produced there. The
methodology for how we generated average compliance costs for the gasoline consumed in a
PADD from the average costs for the gasoline produced in a PADD is summarized in the RIA
Section 6.1.2. Tables 9.6-7,8 and 9 summarize gasoline consumption volumes and average per-
gallon consumption costs and per-gallon maximum total and maximum variable costs for each
PADD for estimating the price impacts of the benzene program.
9-51
-------
Final Regulatory Impact Analysis
Table 9.6-7. Summary of Yearly Volumes and Potential Price Increases by PADD for the
Benzene Program Based on Average Total Costs
(2003 dollars, 7% ROI before taxes)
Year
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
PADD 1
Gasoline
Consumption
(million gals)
49,193
49,517
50,274
50,923
51,734
52,707
53,734
54,599
55,355
56,058
56,761
57,464
58,167
58,869
59,626
60,329
60,978
61,626
62,221
62,816
63,464
64,113
64,816
65,518
66,167
66,870
67,519
68,221
68,931
69,648
70,373
71,105
Cost
(c/gal)
0
0
0
0.008
0.014
0.014
0.027
0.027
0.051
0.048
0.048
0.147
0.147
0.147
0.147
0.147
0.147
0.147
0.147
0.147
0.147
0.147
0.147
0.147
0.147
0.147
0.147
0.147
0.147
0.147
0.147
0.147
PADD 2
Gasoline
Consumption
(million gals)
38,790
39,045
39,642
40,154
40,793
41,560
42,370
43,052
43,649
44,203
44,757
45,311
45,866
46,420
47,016
47,571
48,082
48,594
49,063
49,531
50,043
50,554
51,109
51,663
52,174
52,728
53,240
53,794
54,354
54,919
55,491
56,068
Cost
(c/gal)
0
0
0
0.053
0.091
0.091
0.194
0.194
0.308
0.227
0.227
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
PADD 3
Gasoline
Consumption
(million gals)
20,615
20,751
21,068
21,340
21,680
22,088
22,518
22,881
23,198
23,492
23,787
24,081
24,376
24,670
24,987
25,282
25,554
25,826
26,075
26,324
26,596
26,868
27,162
27,457
27,729
28,023
28,295
28,589
28,887
29,187
29,491
29,798
Cost
(c/gal)
0
0
0
0.013
0.022
0.022
0.042
0.042
0.075
0.065
0.065
0.154
0.154
0.154
0.154
0.154
0.154
0.154
0.154
0.154
0.154
0.154
0.154
0.154
0.154
0.154
0.154
0.154
0.154
0.154
0.154
0.154
PADD 4
Gasoline
Consumption
(million gals)
4542
4572
4642
4702
4777
4867
4962
5042
5111
5176
5241
5306
5371
5436
5506
5571
5631
5691
5745
5800
5860
5920
5985
6050
6110
6175
6235
6299
6365
6431
6498
6566
Cost
(c/gal)
0
0
0
0.019
0.033
0.033
0.099
0.099
0.213
0.227
0.227
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
PADD 5 except CA
Gasoline
Consumption
(million gals)
7918
7971
8092
8197
8327
8484
8649
8788
8910
9023
9137
9250
9363
9476
9598
9711
9815
9920
10,015
10,111
10,215
10,320
10,433
10,546
10,651
10,764
10,868
10,981
11,095
11,211
11,328
11,445
Cost
(c/gal)
0
0
0
0.004
0.007
0.007
0.035
0.035
0.140
0.244
0.244
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
9-52
-------
Final Regulatory Impact Analysis
Table 9.6-8. Summary of Yearly Volumes and Potential Price Increases by PADD for the
Benzene Program Based on Maximum Total Costs
(2003 dollars, 7% ROI before taxes)
Year
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
PADD 1
Gasoline
Consumption
(million gals)
49,193
49,517
50,274
50,923
51,734
52,707
53,734
54,599
55,355
56,058
56,761
57,464
58,167
58,869
59,626
60,329
60,978
61,626
62,221
62,816
63,464
64,113
64,816
65,518
66,167
66,870
67,519
68,221
68,931
69,648
70,373
71,105
Cost
(c/gal)
0
0
0
0.026
0.026
0.026
0.189
0.189
5.67
5.67
5.67
5.80
5.80
5.80
5.80
5.80
5.80
5.80
5.80
5.80
5.80
5.80
5.80
5.80
5.80
5.80
5.80
5.80
5.80
5.80
5.80
5.80
PADD 2
Gasoline
Consumption
(million gals)
38,790
39,045
39,642
40,154
40,793
41,560
42,370
43,052
43,649
44,203
44,757
45,311
45,866
46,420
47,016
47,571
48,082
48,594
49,063
49,531
50,043
50,554
51,109
51,663
52,174
52,728
53,240
53,794
54,354
54,919
55,491
56,068
Cost
(c/gal)
0
0
0
0.243
0.243
0.243
0.473
0.473
3.54
3.54
3.54
5.89
5.89
5.89
5.89
5.89
5.89
5.89
5.89
5.89
5.89
5.89
5.89
5.89
5.89
5.89
5.89
5.89
5.89
5.89
5.89
5.89
PADD 3
Gasoline
Consumption
(million gals)
20,615
20,751
21,068
21,340
21,680
22,088
22,518
22,881
23,198
23,492
23,787
24,081
24,376
24,670
24,987
25,282
25,554
25,826
26,075
26,324
26,596
26,868
27,162
27,457
27,729
28,023
28,295
28,589
28,887
29,187
29,491
29,798
Cost
(c/gal)
0
0
0
0.323
0.323
0.323
0.424
0.424
4.10
4.10
4.10
4.10
4.10
4.10
4.10
4.10
4.10
4.10
4.10
4.10
4.10
4.10
4.10
4.10
4.10
4.10
4.10
4.10
4.10
4.10
4.10
4.10
PADD 4
Gasoline
Consumption
(million gals)
4542
4572
4642
4702
4777
4867
4962
5042
5111
5176
5241
5306
5371
5436
5506
5571
5631
5691
5745
5800
5860
5920
5985
6050
6110
6175
6235
6299
6365
6431
6498
6566
Cost
(c/gal)
0
0
0
0.609
0.609
0.609
0.176
0.176
2.46
2.46
2.46
5.62
5.62
5.62
5.62
5.62
5.62
5.62
5.62
5.62
5.62
5.62
5.62
5.62
5.62
5.62
5.62
5.62
5.62
5.62
5.62
5.62
PADD 5 except CA
Gasoline
Consumption
(million gals)
7918
7971
8092
8197
8327
8484
8649
8788
8910
9023
9137
9250
9363
9476
9598
9711
9815
9920
10,015
10,111
10,215
10,320
10,433
10,546
10,651
10,764
10,868
10,981
11,095
11,211
11,328
11,445
Cost
(c/gal)
0
0
0
0.334
0.334
0.334
0.334
0.334
3.37
3.37
3.37
4.29
4.29
4.29
4.29
4.29
4.29
4.29
4.29
4.29
4.29
4.29
4.29
4.29
4.29
4.29
4.29
4.29
4.29
4.29
4.29
4.29
9-53
-------
Final Regulatory Impact Analysis
Table 9.6-9. Summary of Yearly Volumes and Potential Price Increases by PADD for the
Benzene Program Based on Maximum Operating Costs
(2003 dollars)
Year
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
PADD 1
Gasoline
Consumption
(million gals)
49,193
49,517
50,274
50,923
51,734
52,707
53,734
54,599
55,355
56,058
56,761
57,464
58,167
58,869
59,626
60,329
60,978
61,626
62,221
62,816
63,464
64,113
64,816
65,518
66,167
66,870
67,519
68,221
68,931
69,648
70,373
71,105
Cost
(c/gal)
0
0
0
0.026
0.026
0.026
0.096
0.096
4.56
4.56
4.56
4.56
4.56
4.56
4.56
4.56
4.56
4.56
4.56
4.56
4.56
4.56
4.56
4.56
4.56
4.56
4.56
4.56
4.56
4.56
4.56
4.56
PADD 2
Gasoline
Consumption
(million gals)
38,790
39,045
39,642
40,154
40,793
41,560
42,370
43,052
43,649
44,203
44,757
45,311
45,866
46,420
47,016
47,571
48,082
48,594
49,063
49,531
50,043
50,554
51,109
51,663
52,174
52,728
53,240
53,794
54,354
54,919
55,491
56,068
Cost
(c/gal)
0
0
0
0.243
0.243
0.243
0.351
0.351
3.02
3.02
3.02
4.42
4.42
4.42
4.42
4.42
4.42
4.42
4.42
4.42
4.42
4.42
4.42
4.42
4.42
4.42
4.42
4.42
4.42
4.42
4.42
4.42
PADD 3
Gasoline
Consumption
(million gals)
20,615
20,751
21,068
21,340
21,680
22,088
22,518
22,881
23,198
23,492
23,787
24,081
24,376
24,670
24,987
25,282
25,554
25,826
26,075
26,324
26,596
26,868
27,162
27,457
27,729
28,023
28,295
28,589
28,887
29,187
29,491
29,798
Cost
(c/gal)
0
0
0
0.323
0.323
0.323
0.342
0.342
3.41
3.41
3.41
3.41
3.41
3.41
3.41
3.41
3.41
3.41
3.41
3.41
3.41
3.41
3.41
3.41
3.41
3.41
3.41
3.41
3.41
3.41
3.41
3.41
PADD 4
Gasoline
Consumption
(million gals)
4542
4572
4642
4702
4777
4867
4962
5042
5111
5176
5241
5306
5371
5436
5506
5571
5631
5691
5745
5800
5860
5920
5985
6050
6110
6175
6235
6299
6365
6431
6498
6566
Cost
(c/gal)
0
0
0
0.609
0.609
0.609
0.609
0.609
2.01
2.01
2.01
4.27
4.27
4.27
4.27
4.27
4.27
4.27
4.27
4.27
4.27
4.27
4.27
4.27
4.27
4.27
4.27
4.27
4.27
4.27
4.27
4.27
PADD Sexcept CA
Gasoline
Consumption
(million gals)
7918
7971
8092
8197
8327
8484
8649
8788
8910
9023
9137
9250
9363
9476
9598
9711
9815
9920
10,015
10,111
10,215
10,320
10,433
10,546
10,651
10,764
10,868
10,981
11,095
11,211
11,328
11,445
Cost
(c/gal)
0
0
0
0.334
0.334
0.334
0.334
0.334
2.75
2.75
2.75
3.34
3.34
3.34
3.34
3.34
3.34
3.34
3.34
3.34
3.34
3.34
3.34
3.34
3.34
3.34
3.34
3.34
3.34
3.34
3.34
3.34
9.6.4 Projected Fuel Supply and Energy Impacts of the Benzene Program
EPA has evaluated the potential impact on U.S. fuel supply of the benzene program. As
discussed in detail elsewhere in this chapter, refiners are expected to utilize a variety of
approaches to control benzene. Other than extraction these do not impact gasoline production
appreciably. Extraction physically removes benzene from the refinery reformate stream, usually
for sale into the petrochemical market. In extracting benzene, the volume of reformate available
for gasoline production is reduced.
We estimate that in response to the benzene program, refiners will extract about 12,500
barrels of benzene per day, or 192 million gallons per year, when the benzene program is fully
phased-in. Because benzene has a slightly higher energy density than gasoline (about 7 percent
higher), the projected extracted benzene is equivalent to about 13,375 barrels per day of gasoline,
9-54
-------
Final Regulatory Impact Analysis
or about 0.1 percent of U.S. gasoline production. However, we believe that the net effect on
gasoline supply of the rule will be far less, potentially zero.
This increase in extraction of benzene from gasoline is expected to occur with or without
the benzene program. Using CMAFs estimate of a 2.4 percent annual growth in benzene
demand, we expect that U.S. demand for benzene will increase by 600 million gallons from 2007
to 2015, the years that the benzene program is expected to phase-in. Assuming that reformate
extraction continues to supply about 40 percent of the supply, then reformate extraction is
expected to supply about 250 million gallons additional benzene over the 8 year program phase-
in period. Thus, increased reformate extraction expected to occur to meet increased benzene
demand would exceed the projected benzene extraction expected to occur to comply with the
benzene program, provided that the benzene extraction occurs throughout the entire phase-in
period. If all the benzene extraction occurs to comply with the benzene program in a single
year, then the increased benzene supply would be greater than two times the yearly increase in
total benzene demand.
Even if all the projected benzene extraction occurs in a single year, the benzene market
could adjust to rebalance both the benzene market and the gasoline supply. Selective toluene
disproportionation and toluene hydrodealkylation are benzene production technologies which are
higher benzene production cost technologies. These two marginal benzene production processes
would likely reduce their benzene production which would rebalance the benzene
supply/demand market. Presuming that these two benzene production processes temporally
reduce their output to rebalance benzene supply, the toluene would presumably stay in the
gasoline pool and the effect on gasoline supply would be minimal.
Projected Energy Impacts of the Benzene Program
We used the LP and refinery-by-refinery models to estimate the changes in energy use
that would result from the implementation of the benzene program.29 For this analysis, we used
the refinery-by-refinery model to select the range of technologies we believe would be likely to
be used across the industry by PADD in 2012, both with and without a benzene program. We
then used the resulting array of technologies as input data for the LP model. This data then
became the starting point for runs of the LP model, which we used to produce estimates of the
net change in energy use due to increased refinery processing and changes to inputs into the
refinery. In these runs, the LP model maintains the same volume of gasoline production in the
reference and control cases. The model makes up the loss of gasoline volume due to benzene
extraction by assuming additional purchases of crude oil. To the extent that this benzene
extraction would be made up by swapping gasoline blendstocks or by increases to refinery
intermediate streams that could then be used to produce gasoline, this analysis is somewhat
conservative. Table 9.6-10 presents the results of the energy use evaluation.
9-55
-------
Final Regulatory Impact Analysis
Table 9.6-10. Estimated
(in Thousands of Fuel Oil Eq
Refinery Process Energy Use
Total Benzene Control-Related
Light Naphtha Splitting
Reforming
Isomerization
Benzene Saturation
Benzene Extraction
Hydrogen Production
All Other
Net Process Energy Change
% Change in Process Energy
Net Total Energy Change
% Change in Total Energy
PADD1
0.4
-0.1
0.2
0
0
0.4
-0.2
-0.2
0.2
0.2
0.9
0.05
Changes in Energy Use (2012)
uivalent Barrels per Day (Kfoeb/d)
PADD2
2.0
1.1
-0.6
-0.5
0.2
1.1
0.8
0.5
2.5
0.9
3.2
O.OP
PADD3
3.4
0.1
0.6
0
0.3
1.9
0.5
0.2
3.6
0.4
5.1
0.06
PADDs 4&5
(except CA)
2.1
-0.1
0.4
0.1
0.9
0
0.8
-0.1
2.0
1.8
3.4
0.21
All PADDs
(except CA)
8.0
1.1
0.6
-0.5
1.5
3.4
1.8
0.4
8.3
0.6
12.7
0.08
As shown in the table, our modeling projects that increases refinery process energy (fuel,
steam, and electricity) would contribute most to the total change in energy use (8.3 of the total
increase of 12.7 Kfoeb/d). This process energy increase would represent about 0.6 percent of all
energy used in refinery processes. When all energy involved in refining crude oil is considered,
including the energy in crude oil and other feedstocks, we project that the benzene program
would increase overall energy use by refineries by less than 0.1 percent.
Of the nationwide increase in process energy, most would be due to processes directly
related to benzene control (8 of 8.3 Kfoeb/d). Benzene extraction would be the largest
contributor to this process energy increase (3.4 of 8.3 Kfoeb/d). It is important to note as
discussed above that the increase in benzene production through greater extraction, and thus the
increase in energy used in this process, would likely occur regardless of whether the benzene
program was in place. Thus, the increase in energy used to extract benzene could be attributed to
meeting the increased demand for benzene rather than attributed to the benzene program.
(Projected increases in energy use due to the other benzene-related processes would be
appropriately attributed to the benzene program.)
The variation in energy impacts from PADD to PADD shown in the table results from the
expected differences in the technological approaches refiners would pursue in different parts of
the country, as discussed in Chapter 6. For example, for PADDs 2, 4, and 5, we do not expect
that the benzene program would result in an increase in benzene extraction, and thus the table
shows no increase in energy for this process. However, we project that the largest energy
increases in PADD 1 and PADD 3 would be due to increased benzene extraction. (Refiners in
these regions would be near benzene markets and would tend to invest in benzene extraction
equipment.) Overall, we project that PADD 3 would contribute a significant portion of the
nationwide increase in energy use because of its very large production volume as well as because
of its reliance on extraction. PADDs 4 and 5 provide a significant portion of the energy demand
9-56
-------
Final Regulatory Impact Analysis
despite the lower gasoline production in these two PADDs because of the large reduction in
benzene levels in these two PADDs.
9.7 Refinery Industry Cost Study
The American Petroleum Institute (API) conducted its own refinery modeling study to
evaluate the cost of benzene control.30 The API study, conducted by Baker and O'Brien
Incorporated, analyzed the cost of three different benzene programs, and these were Case A: a
0.60 vol% average benzene standard and 0.90 per-gallon cap standard applicable to RFG, and a
0.95 average vol% benzene standard and 1.30 per-gallon cap standard applicable to CG, but no
credit trading program; Case B: a 0.60 vol% average benzene standard and 0.90 per-gallon cap
standard applicable to both RFG and CG, but no credit trading program; and Case C: a 0.60
vol% average benzene standard for both CG and RFG with no cap standard and with a credit
trading program.1 API made some very conservative assumptions regarding credit generation
and use for Case C. API assumed that when credits are being generated that each refinery will
hold onto 10 percent of the generated credits as a safety margin which resulted in a lower
benzene level than that required.
The API study also assumed that MTBE is no longer blended into the U.S. gasoline pool,
that the Tier 2 gasoline sulfur program is fully implemented, that the renewable fuels standard is
implemented resulting in 7 billion gallons of ethanol blended into the gasoline pool and that
MSAT1 is still in effect. The three cases modeled by API are summarized in Table 9.7-1. We
also included the final U.S. gasoline pool benzene levels for the base case and each case in the
last column of the table. We adjusted the benzene levels to exclude California gasoline because
it is not assumed to be regulated by the API refinery modeling study consistent with our analysis.
Table 9.7-1. Summary of the Three Refinery Modeling Case Studies by API
Basecase
Case A
CaseB
CaseC
Gasoline
Pool
Total Pool
RFG
CG
RFG
CG
RFG
CG
Avg Std
-
0.60
0.95
0.60
0.60
0.60
0.60
Cap Std
-
0.90
1.30
0.90
0.90
None
None
Credit
Trading
-
No
No
Yes
Benzene
Level
(vol%)
1.00
0.70
0.52
0.56
i In the refinery modeling report, Baker and O'Brien states that the benzene program modeled for Case C is a 0.60
vol% benzene standard, and that credits are calculated based on benzene reductions below that value. However, in
its comments to the proposed rule summarizing the results of its refinery modeling study, API stated that it modeled
a 0.62 vol% average benzene standard with a 0.02 vol% compliance margin. It appears that API was trying to adapt
its refinery modeling cost study to mirror the proposed standard, but that the refinery modeling study actually
modeled a 0.60 benzene control standard. For the purposes of our review of the API study we will assume that the
modeled standard was the 0.60 vol% benzene standard, not a 0.62 vol% benzene standard as indicated in its
comments.
9-57
-------
Final Regulatory Impact Analysis
The types of benzene control technologies modeled in the API study include modifying
the cutpoints to remove benzene precursors from reformer feed, build or expand benzene
saturation units, expand aromatics extraction units, and build or expand pentane/hexane
isomerization units. These are the same technologies that we used in our cost study, except that
API did not allow refineries to install grassroot extraction units. Similar to our study, API did
allow refineries with aromatics extraction units to expand their units to extract the aromatics
from the gasoline of other refineries, although in our study we only assumed the extraction of
benzene, not xylene and toluene - the other aromatic compounds that can be extracted from
gasoline.
The total costs for each of the refinery modeling cases analyzed by API were summarized
in their report. The API refinery modeling report did not calculate the per-gallon costs so we
made the necessary calculations based on the total annual capital costs provided by API which
are based on a 10 percent return on investment (ROI). We summarize those costs and adjust
them to a 7 percent ROI - the basis for how we express the per-gallon costs - to express the API
costs on the same basis as ours. The total annual costs, per-gallon and adjusted per-gallon costs
for each case are summarized in Table 9.7-2.
Table 9.7-2. Total and Per-Gallon Costs for API's Refinery Modeling Study
Case#
A
B
C
Total Cost
($MM/yr)
1286
1660
1431
2012
Gasoline
Volume
(Kbbl/day)
8365
8365
8365
Investment
(Million
dollars)
899
1737
1476
Per-
Gallon
Cost
(c/gal)
1.00
1.29
1.12
Capital
Charge
10% after
tax ROI
($MM/yr)
151
293
246
Capital
Charge
7% before
tax ROI
($MM/yr)
97
188
158
Adjusted
Per-
Gallon
Cost
(c/gal)
0.96
1.21
1.05
Of the four cases modeled by API, Case C is the closest to our final benzene program,
thus we will compare API's cost estimate for that case to our estimated benzene program costs.
It is immediately apparent that there is a large difference in estimated cost between our estimated
cost, which is 0.27 cents per gallon, compared to API's Case C, which is 1.05 cents per gallon.
We identified numerous reasons for the most of the difference in cost.
One of the most important differences between the two cost estimates is that API
assumed a much larger benzene reduction than our study. The starting benzene level for the API
study was 1.0 vol% benzene. After control, the API study assumed a slightly more stringent
benzene standard - modeling a 0.60 vol% average standard instead of a 0.62 vol% average
benzene standard - and a much more conservative approach to how refiners use credits. API
assumed that credits would not be traded freely, but instead that refining companies would hold
onto 10 percent of their credits in case they have a future problem with their benzene control
unit. Due to the more stringent benzene standard and the 10 percent credit margin, the API study
estimated that the U.S. refining industry would average 0.56 vol% benzene compared to our 0.62
vol% benzene control level. From the base case to the final benzene level, the API study
analyzed a 0.44 vol% benzene reduction. However, our study estimated the impacts of a 0.33
9-58
-------
Final Regulatory Impact Analysis
vol% benzene reduction. The API study estimated a 33 percent greater benzene reduction than
that analyzed in our analysis.
EPA does not believe that refiners will find it necessary to consistently and significantly
overcomply with the 0.62 vol% average benzene standard and hold onto a significant amount of
credits as assumed by API. This is because this benzene standard is an average standard, not a
cap standard, and can be met by the accumulation of gasoline batches with benzene levels higher
or lower than the standard. Thus, if a refinery tended to produce gasoline with lower or higher
gasoline benzene levels over the first part of the year, the operations could be adjusted to balance
out the gasoline benzene levels for the rest of the year. Also, our program includes several
provisions which give refiners significant flexibility for compliance with average benzene
standard. For example, refiners could overcomply slightly with the standard early on in the
program's implementation and hold onto the credits for up to five years before they expire. If a
refinery's benzene control unit goes down, the refiner would be able to use those accumulated
credits, the refiner could purchase credits from other refineries, or the refiner could create a
benzene reduction deficit at that refinery and make it up the deficit following year. With this
degree of flexibility, there will be little need for a refining company to control its refineries'
benzene level on an ongoing basis at a lower level than the standard to have a substantial supply
of credits on hand. Even if they did feel the need to accumulate some benzene credits, the
company could do so the first year, but then would not likely do so for each year after since the
first year's credits would be sufficient for the next five years. For these reasons, we believe that
the overcompliance modeled by API is unnecessary.
The second reason why the API estimated costs are higher than our estimates is that API
used a more restrictive assumption with respect to benzene extraction - a more cost-effective
benzene control technology than benzene saturation which was the principal benzene control
technology relied upon by the API study. API assumed that no new grassroots benzene
extraction capacity would be installed in the future, but that existing extraction units could be
expanded. We agree that existing units will likely be expanded. However, we also believe that
new grassroots extraction units will be installed as well. Our premise is supported by CMAI
projections of a continued robust benzene market in the future with benzene priced higher than
its historical margin above gasoline. CMAI estimates a benzene price which is $30/bbl higher
than gasoline, which is higher than its historical margin. Higher benzene price margins will
provide an incentive to refiners to add grassroots benzene extraction units, even in areas where
benzene markets have been smaller. For example, one refiner has indicated to us that if the
proposed gasoline benzene standard was to be finalized, it would install a grassroots benzene
extraction unit at one of its refineries in the Midwest, where the benzene market is currently
small. This is a strong indicator that new grassroots benzene extraction units will also be
installed on the Gulf and East Coasts, where benzene markets are already strong.
API's cost of aromatics extraction is likely to be higher than our extraction costs because
of the differences in benzene prices. For the final rule, we used the most recent CMAI benzene
price projection, which prices benzene at $30/bbl above that of gasoline. API used an
incremental benzene price of $20 per barrel above that of gasoline, which is what we used for the
proposed rule. A likely primary reason for CMAI's higher incremental benzene price is that
CMAI is assuming a higher future crude oil price.
9-59
-------
Final Regulatory Impact Analysis
The third reason why the API benzene control costs are higher than ours is the very large
difference in octane control costs. For both studies, the cost associated with the octane loss that
occurs through the use of the various benzene control technologies is accounted for by assigning
a dollar per octane-barrel cost to the octane loss. However, API's costs for restoring octane are
about an order of magnitude higher than the octane recovery costs that we are projecting. The
octane costs used by API and those we use are summarized in Table 9.7-3.
Table 9.7-3. Octane Costs used in the API and EPA Benzene Cost Studies ($/octane-barrel
API
EPA
PADD 1
2.19
0.28
PADD 2
2.11
0.20
PADD 3
1.83
0.30
PADD 4
2.14
0.27
PADD 5
2.58
0.27
The octane costs used by API are high because API used the rack price differential
between premium and regular grade gasoline as summarized by the Energy Information
Administration. Using the rack price differential between premium and regular grade gasoline
results in high octane costs because they reflect a significant amount of profit. For example, the
cost difference to produce premium gasoline is usually only a few cents per gallon more than for
producing regular grade gasoline, yet refiners and marketers usually charge 20 to 30 cents per
gallon higher price for premium gasoline at retail. Much of this marked up price appears at the
rack price differential between regular and premium grades of gasoline. A review of octane
prices shows that the rack price differential between premium and regular grade gasoline is 50%
higher then when estimating octane cost using bulk prices. Bulk prices are closer to the actual
costs incurred by refiners with respect to the cost of octane. However, our linear programming
cost analysis shows that refinery octane costs are much lower than bulk prices.
Another reason why the API octane costs are higher than ours is because they used the
premium-regular grade gasoline price differential for the summer of 2005, when the octane costs
are likely higher than in the future due to the very large volume of ethanol that is expected to
enter the gasoline market by then under the Renewable Fuels Standard. In addition to the large
volume of ethanol, ethanol has very high octane (115 (R + M)/2) which contributes to the large
impact on octane costs. The large impact that ethanol will have on octane costs is reflected in
the octane costs that we use in our analysis.
9-60
-------
Final Regulatory Impact Analysis
References for Chapter 9
1 Refinery Modeling: Legislative and Regulatory Developments: Effects on Gasoline
Supply, Prepared for U.S. Environmental Protection Agency by Abt Associatiates Inc under
subcontract from ICF Incorporated, January 14, 2003; Appendices A, B and D.
2 Evaluation of Refinery Gasoline Benzene Reduction Options - Part 2, Prepared for
U.S. Environmental Protection Agency by Abt Associates Inc. under contract from ICF
Consulting Inc., November 30, 2004.
3 Keesom, William H., et al, Benzene Reduction Alternatives, Technical Paper
presented to the 1991 National Petroleum Refiners Association Annual Meeting.
4DiVita, Vince, Group Manager, Jacobs Consultancy, Peer Review of a Refinery-By-
Refinery Cost Model for the U.S. Refining Industry, January 9, 2006.
5 Hodge, Cal, President, A 2nd Opinion, Peer Review: Cost of Benzene Control Model,
January 8, 2006.
6 Kolb, Jeff, Abt Associates, Capital Costs for Benzene Control, Memorandum provided
to EPA under contract WA 0-01, EP-C-06-094, September 29, 2006.
7 Rees, Conway, Senior Process Director, Fluor; Technical Session: Considerations
when Revamping for ULSD, Hydroprocessing Principles and Practices, National Petrochemical
and Refiners Association Question and Answer Forum, October 2006.
8 Worldwide Refining/Worldwide Production, Oil and Gas Journal, Volume 98.51,
December 18, 2000.
9 Annual Energy Outlook 2006, Energy Information Administration, Department of
Energy.
10 Annual Energy Outlook 2006, Energy Information Administration, Department of
Energy.
11 Worldwide Report, Oil and Gas Journal, www.ogionline.com, December 22, 2005.
10
Capacity of Operable Petroleum Refineries by State as of January 1, 2005, Table 38 of
Petroleum Supply Annual 2004, Volume 1, Energy Information Administration, Department of
Energy.
13 Inputs of Crude Oil and Petroleum Products by Country of Origin, 2004, Tables 22 -
25, Petroleum Supply Annual 2004 Volume 1, Energy Information Administration, Department
of Energy, Department of Energy.
14 Refinery Net Production of Finished Petroleum Products by PAD, and Refining
Districts, 2005, Petroleum Supply Annual, Table 17, Petroleum Supply Annual 2005, Volume 1,
Energy Information Administration, Department of Energy.
15 Annual Energy Outlook 2006, Energy Information Administration, Department of
Energy.
16 Annual Energy Outlook 2006, Energy Information Administration, Department of
Energy.
17 Kolb, Jeff, Abt Associates, Procedures for Calibrating the Benzene Spreadsheet Model,
Memorandum provided to EPA under contract WA 0-01, EP-C-06-094, December 29, 2006.
18 Information in the LP refinery model from literature and provided by vendors of
refining technologies.
19 Information in the LP refinery model from literature and provided by vendors of
refining technologies.
9-61
-------
Final Regulatory Impact Analysis
20 Information in the LP refinery model from literature and provided by vendors of
refining technologies.
21 Information in the LP refinery model from literature and provided by vendors of
refining technologies including new information.
22 Rock, Kerry L., Manager CDTECH, Low Cost Benzene Reduction for RFG,
Technical Paper presented at the 1995 National Petroleum Refiners Association Annual Meeting.
23 Rock, Kerry L., Manager CDTECH, Cost Effective Solutions for Reducing the
Benzene in Gasoline, Technical Paper presented at the 1997 National Petroleum Refiners
Association Annual Meeting.
24 Information in the LP refinery model from literature and provided by vendors of
refining technologies including new information.
25 World Benzene Analysis 2004, Chemical Market Associates Incorporated, Houston
Texas, 2004.
26 Toups, Mackenzie, Client Services Representative - Aromatic Studies
Chemical Market Associates, Inc., e-mail to Lester Wyborny of EPA, October 17, 2006.
27 Fisler, Mark, Vice President, Chemical Market Associates Incorporated, Benzene
Outlook in the Era of Reformulated Gasoline, Presented at the NPRA Annual Meeting, March
1992.
28
Stellman, Richard, President, The Pace Consultants Incorporated, Reformulated
Gasoline and the Petrochemical Industry, Presented at the NPRA Annual Meeting, March 1992.
29 Kolb, Jeff, Abt Associates, Estimated Changes in Energy Use, LP Refinery Model
Output provided to EPA under contract WA 0-01, EP-C-06-094, December 27, 2006.
30 Tamm, David C., Baker & O'Brien Incorporated, An Assessment of the Impact of
Potential Mobile Source Air Toxics II Regulations on Refinery Operations, Costs and Supply, a
cost study completed for the American Petroleum Institute, May 2006.
9-62
-------
Final Regulatory Impact Analysis
Chapter 10: Table of Contents
CHAPTER 10: Portable Fuel Container Costs 2
10.1 Methodology 2
10.2 Costs for Permeation Control 2
10.3 Spout Costs 3
10.4 Certification Costs 4
10.5 Per Container Total Costs 4
10.6 Costs for PFCs Complying with State Programs 5
10.7 Fuel Savings 6
10.8 Annual Total Nationwide Costs and Fuel Savings 6
10-1
-------
Final Regulatory Impact Analysis
CHAPTER 10: Portable Fuel Container Costs
This chapter presents a detailed analysis of the projected average portable fuel container
(PFC) costs related to meeting new emissions standards, which would require the use of "best
available controls." These costs have been developed based on industry information, discussions
with manufacturers (including confidential business information concerning technology costs),
and engineering judgment. These costs include variable costs for improved materials used in
manufacturing PFCs (including improved spouts), and fixed costs for research and development,
tooling, and certification. Finally, this chapter presents estimated fuel savings and aggregate
nationwide costs for PFCs.
10.1 Methodology
The following technology characterization and cost figures reflect our current best
judgment based on engineering analysis, information from manufacturers, and the published
literature. The analysis includes manufacturer markups to the retail level.
Costs of control typically include variable costs (for incremental hardware costs,
assembly costs, and associated markups) and fixed costs (for tooling, R&D, and certification).
Variable costs are marked up at a rate of 29 percent to account for PFC manufacturers' overhead
and profit.l To account for additional warranty costs associated with a change in technology, we
have added 5 percent of the incremental variable cost. We estimated a range of costs for
different size PFCs and also an average per container cost based on the approximate sales
weighting of the three PFC sizes.A All costs are in 2003 dollars.
We are not projecting any additional R&D costs associated with the new EPA PFC
standards. Manufacturers have developed and are continuing to develop control technologies in
response to the California (and other state) programs. EPA's program is very similar to the
California program and we believe the most likely approach for manufacturers will be to use the
technologies developed for state programs nationwide. Manufacturers will incur the R&D costs
even in the absence of EPA emissions standards. Further, the permeation barriers available are
very well understood within the industry. Therefore, we believe manufacturers will use these
same technologies for their nationwide product lines and will not incur significant new R&D
costs due to an EPA program.
We estimate that tooling and certification costs will be incurred one year prior to
production, on average. These fixed costs were increased by seven percent to reflect the time
value of money over the one year period. The fixed costs then were recovered over the first five
years of production at a rate of seven percent.
10.2 Costs for Permeation Control
Multi-layered designs
1 PFC sales for 1,2, and 5 gallon containers are weighted at 33%, 33%, and 34% of total sales, respectively.
10-2
-------
Final Regulatory Impact Analysis
Manufacturers have indicated that most are likely to switch to multi-layer designs to meet
permeation requirements. For this analysis, we considered a PFC design with a material
composition of 3% ethylene vinyl alcohol (EVOH) at $3.50/lb, 4% adhesive layer at $l/lb and
the remainder HDPE.2 This resulted in materials costs ranging from $0.29 to $0.58 for 1 to 5
gallon containers, with an average materials cost of $0.41.B
In some cases, blow-molding machines can be retrofitted for multi-layer operation. The
total cost of such a retrofit, including supporting equipment, would be about $1,000,000 per
machine. In other cases, a new blow-molding machine would be required. A machine that could
blow-mold multi-layer tanks would approximately double the price of the blow-molding
machine. For this analysis, we use a machine cost increase of $2,000,000, including all molds
and related set-up. For our analysis, we've projected that half the machines would be retrofit and
half would be new, for an average cost of about $1,500,000 per machine. Our analysis uses an
average total annual production of 350,000 blow-molded tanks per machine and an amortization
of the capital costs over 5 years. This results in an average fixed cost per container of $1.12.
Adding the fixed costs to the variable costs described above gives an average per container cost
for multi-layered cans of about $1.53.
Non-continuous Barrier Platelets
Manufacturers may reduce permeation from blow-molded PFCs by blending in a low
permeation material such as EVOH with the HDPE. This is typically known by its trade name,
Selar. The EVOH in the plastic forms non-continuous barrier platelets in the PFC during blow-
molding that make it harder for fuel to permeate through the walls of the tank. Using this
approach, no changes should be necessary in the blow-molding equipment, so the costs are based
on increased material costs. We used 10 percent EVOH, which costs about $3-4 per pound, and
90 percent HDPE, which costs about $0.65-0.75 per pound. This equates to a price increase of
about $0.35 per pound. The increased cost for PFCs would range from $0.69 to $1.38, with an
average cost increase of $1.00 per container.
Fluorination
We have also estimated costs for fluorination since some PFC manufacturers have used
this approach to meet current California standards. Our surface treatment cost estimates are
based on price quotes from a company that specializes in this fluorination.3 We estimate that
PFC costs would range from $0.86 to $3.30, with an average cost of $1.84. These prices do not
include the cost of transporting the PFCs; we estimated that shipping, handling and overhead
costs would be an additional $0.30 per PFC.4
10.3 Spout Costs
Manufacturers will need to move from a simple pouring spout to an automatic closing
spout in order to meet evaporative emissions standards. The automatic closing spouts would
include a spring closing mechanism. For this analysis, we estimated an average variable cost
B This analysis was done using container weights of 1.5, 2.0, and 3.0 pounds for 1,2, and 5 gallon containers,
respectively.
10-3
-------
Final Regulatory Impact Analysis
increase for spouts of about $0.85 including assembly costs, based on discussions with PFC
manufacturers. We have also estimated $200,000 for tooling per 1 million spouts. This results
in a fixed cost for tooling of about $0.05 per spout, for a total spout cost of $0.90. The spout
costs would not likely vary by PFC size.
10.4 Certification Costs
Manufacturers will need to integrate the emission control technology into their designs
and there will be some engineering and clerical effort needed to submit the required information
for certification. We expect that in the early years, PFC manufacturers will perform durability
and permeation testing for certification. They will be able to carry over this data in future years
and to PFCs that are made of similar materials and have the same permeation control strategy
regardless of PFC size.
Manufacturers will need to run certification testing for their PFCs and then submit the
data and supporting information to EPA for certification. Based on the current approach used by
manufacturers, we've estimated that each manufacturer will contract out testing at a cost of about
$7,500 per manufacturer. We've included an additional cost of $5,000 for staff time for the
certification process, for a total certification cost of $12,500 per manufacturer.
To calculate a per PFC certification cost, we calculated a total industry cost for
certification of $125,000 and spread this cost over industry-wide sales of 26,000,000 units. As
with other fixed costs, we amortized the cost over five years of sales to calculate per unit
certification costs. Due to the large sales volumes, the analysis results in an average per can cost
for certification of less than one cent.
10.5 Per Container Total Costs
We based our cost analysis on costs associated with multi-layer PFCs. We believe most
manufacturers will continue down the path of using this technology since it is robust, has well-
understood emissions performance, and appears to have the lowest cost once the capital costs are
recovered. Other options for permeation barriers have similar overall costs, especially in the
near term. If manufacturers select a different permeation barrier approach such as non-
continuous barrier platelets or fluorination, tooling costs would be lower, but would be offset by
higher variable costs. Our estimated per container costs are shown in Table 10.5-1. The
weighted average costs would be $2.69. These costs are similar to cost data shared with us by
manufacturers on a confidential basis.
10-4
-------
Final Regulatory Impact Analysis
Table 10.5-1. Costs per PFC
Variable costs
- Permeation Barrier
- Spout
Total Variable Costs
Total Variable costs w/ OEM
Mark-up and warranty
Tooling
Certification
Total
1 gallon
$0.22
$0.85
$1.07
$1.40
$1.17
Less than $0.01
$2.57
2 gallon
$0.28
$0.85
$1.13
$1.48
$1.17
Less than $0.01
$2.65
5 gallon
$0.44
$0.85
$1.29
$1.69
$1.17
Less than $0.01
$2.86
Costs are well understood due to the experience manufacturers have had previously with
permeation emissions control technologies and with the California PFC program. We are
estimating costs based on the likely technology path manufacturers will take to meet the
standards. Costs could be somewhat higher or lower if manufacturers use a different mix of
control technologies or use multiple technologies across their product lines. Other sources of
potential uncertainty include whether costs might be lower on a nationwide basis due to
economies of scale or due to additional learning by the manufacturers.
10.6 Costs for PFCs Complying with State Programs
The above costs are for currently uncontrolled PFCs. Some states have adopted PFC
programs, based on the original California program which took effect in 200l.c The original
California program contained permeation requirements that would be significantly less stringent
than the standards considered in this cost analysis (about a 50 percent emission reduction
compared to an 80 to 90 percent emission reduction). Because the standards considered in this
cost analysis are more stringent than those currently in place in states with programs, we have
estimated costs associated with the difference. For purposes of the cost analysis, we have
estimated that the costs associated with meeting the state programs would be half those for the
permeation requirements considered here, resulting in a cost difference of $0.77 per container.
Although there technically is a difference in stringency between current state programs
and the potential EPA requirements and we are including costs associated with the difference, it
is unlikely that these costs would be realized. California has adopted revised program
requirements that are essentially equivalent to those being considered by EPA. Manufacturers
are in the process of incorporating more robust permeation controls in response to the new
California program. Manufacturers would want to avoid carrying two different products and
would likely use the more robust permeation controls in all states with programs. Also, in the
absence of an EPA program, states would likely adopt the new California requirements
eventually.
c Delaware, Maine, Maryland, Pennsylvania, New York, Connecticut, Massachusetts, New Jersey, Rhode Island,
Vermont, Virginia, Washington DC, Ohio, New Hampshire, and Texas
10-5
-------
Final Regulatory Impact Analysis
10.7 Gasoline Savings
The emissions reductions due to reduced evaporative losses and reduced spills from PFCs
filled with gasoline translate into gasoline savings. As described in Chapter 2, we have
estimated the annual HC reductions due to new standards. By dividing the tons reduced by the
number of PFCs in use with gasoline we can estimate the annual tons reduction per PFC. In
2015, after the program is fully implemented, we estimated that there would be 88,023,896 PFCs
in use with gasoline nationwide and that those cans would be responsible for about 202,347 tons
of HC reduction. We can then translate the tons reduction per can per year (0.002 tons, or 4.1
pounds) to gallons using a fuel density of six Ibs/gallon (for lighter hydrocarbons which
evaporate first). We used an average life of five years for PFCs and used a discount rate of seven
percent to estimate total average undiscounted and discounted fuel savings per PFC, provided
below. We calculated the savings using $1.52 per gallon of gasoline.5 These savings would
offset the cost of the PFC controls.
Table 10.7-1. Average Gasoline Savings Over Life of PFC
HC reduced (pounds)
Fuel Savings (gallons)
Undiscounted Savings
Discounted Savings
20.5
3.4
$5.17
$4.24
10.8 Annual Total Nationwide Costs and Fuel Savings
The above analyses provide incremental per unit PFC cost estimates. Using these per
unit costs and projections of future annual sales, we have estimated total aggregate annual costs.
The aggregate costs are presented on a cash flow basis, with hardware and fixed costs incurred in
the year the PFCs are sold and fuel savings occurring over the life of the PFC. To project annual
sales into the future, we started with an estimated 26 million PFCs sold nationwide in 2002 and
then grew sales by two percent per year.6'7 The resulting sales estimates for select years are
shown in Table 10.8-1 below. To estimate sales in states with and without existing PFC
programs, we projected that 39 percent of overall sales would be in states with existing PFC
programs. This estimate is based on current estimated PFC populations by state provided in
Chapter 2 of the RIA.
Table 10.8-1. Projected Annual PFC Sales
Projected sales
2009
29,866,000
2015
33,634,000
2020
37,134,000
2030
45,267,000
For total fuel savings, we used the nationwide HC reductions estimated in Chapter 2 of
the RIA and the methodology described above to convert to gallons of fuel saved nationwide,
and then to savings in dollars. We estimate that fuel savings ramp up as new PFCs replace old
ones and would more than offset the aggregate costs in the long term, for an overall savings.
Table 10.8-2 presents the results of this analysis. As shown in the table, aggregate costs start out
10-6
-------
Final Regulatory Impact Analysis
at about $58 million and then drop to $33 million in 2014 when the fixed costs have been
recovered. Fuel savings start out at about $15 million per year and reach $101 million in 2014.
After 2014, increases in costs and savings are due to PFC sales and population growth.
As noted above, fixed costs due to certification and tooling are expected to actually be
incurred on average one year prior to the start of the program. We estimate that the total fixed
costs in that year would be about $107 million.
10-7
-------
Final Regulatory Impact Analysis
Table 10.8-2. Annual Nationwide PFC Costs and Fuel Savings
| Calendar Year | Variable Costs | Fixed Costs | Total Costs | Fuel Savings | Net Cost |
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
0
$ 30,194,245
$ 30,798,130
$ 31,414,092
$ 32,042,374
$ 32,683,222
$ 33,336,886
$ 34,003,624
$ 34,683,696
$ 35,377,370
$ 36,084,918
$ 36,806,616
$ 37,542,748
$ 38,293,603
$ 39,059,475
$ 39,840,665
$ 40,637,478
$ 41,450,228
$ 42,279,232
$ 43,124,817
$ 43,987,313
$ 44,867,059
$ 45,764,401
$ 46,679,689
$ 47,613,282
$ 48,565,548
$ 49,536,859
$ 50,527,596
0
$ 27,875,926
$ 27,875,926
$ 27,875,926
$ 27,875,926
$ 27,875,926
$
0
$58,070,171
$58,674,056
$59,290,018
$59,918,300
$60,559,148
$33,336,886
$34,003,624
$34,683,696
$35,377,370
$36,084,918
$36,806,616
$37,542,748
$38,293,603
$39,059,475
$39,840,665
$40,637,478
$41,450,228
$42,279,232
$43,124,817
$43,987,313
$44,867,059
$45,764,401
$46,679,689
$47,613,282
$48,565,548
$49,536,859
$50,527,596
0
$15,346,933
$30,693,867
$48,298,000
$65,901,627
$83,505,760
$101,109,387
$102,522,480
$103,935,898
$105,349,189
$106,762,481
$108,175,772
$109,589,064
$111,056,401
$112,523,738
$113,991,075
$115,458,412
$116,925,749
$118,393,086
$119,860,423
$121,327,760
$122,795,097
$124,262,434
$125,675,726
$127,089,018
$128,502,309
$129,915,601
$131,328,892
0
$42,723,237
$27,980,189
$10,992,018
-$5,983,327
-$22,946,612
-$67,772,501
-$68,518,856
-$69,252,201
-$69,971,819
-$70,677,563
-$71,369,156
-$72,046,316
-$72,762,798
-$73,464,263
-$74,150,410
-$74,820,934
-$75,475,522
-$76,113,854
-$76,735,606
-$77,340,447
-$77,928,038
-$78,498,034
-$78,996,037
-$79,475,735
-$79,936,761
-$80,378,742
-$80,801,296
10-8
-------
Final Regulatory Impact Analysis
References for Chapter 10
1 ."Update of EPA's Motor Vehicle Emission Control Equipment Retail Price Equivalent (RPE)
Calculation Formula," Jack Faucett Associates, Report No. JACKFAU-85-322-3, September
1985.
2"Plastic News," Resin Pricing for November 8, 2004, www.plasticsnews.com.
3 "Information on Costs and Effectiveness of Fluorination Received from Fluoroseal,"
Memorandum from Mike Samulski to Docket A-2000-1, March 27, 2002.
4"Shipping Costs," Memorandum from Glenn Passavant, U.S. EPA to Docket A-2000-01, March
27, 2002.
5 Energy Information Administration, Annual Energy Outlook 2005, Table 12, Petroleum
Product Prices, January 2005, DOE/EIA-0383(2005). EIA projected average post-tax gasoline
costs for 2010.
6 "Characterizing Gas Can Markets: A Profile," RTI International, Final Report, August 2004.
7 "Gas Can Industry Profile Updates", Memorandum from RTI to EPA, January 4, 2007.
10-9
-------
Final Regulatory Impact Analysis
Chapter 11: Table of Contents
Chapter 11: Cost per Ton of Emissions Reduced 2
11.1 Cost per Ton for Vehicle Standards 2
11.2 Cost Per Ton for Fuel Benzene Standard 4
11.3 Cost Per Ton for PFCs 6
11.4 Cost Per Ton for the Overall Proposal 7
11-1
-------
Final Regulatory Impact Analysis
Chapter 11: Cost per Ton of Emissions Reduced
We have calculated the cost per ton for the rule based on the net present value of all costs
incurred and all emission reductions generated from 2009 out to 2030. The time window is
meant to capture both the early period of the program when there are a small number of
compliant vehicles and portable fuel containers (PFCs) in use, and the later period when there is
nearly complete turnover to compliant vehicles and PFCs. For the fuel benzene standards, which
begin in 2011, the cost per ton estimates include costs and emission reductions that will occur
from all vehicles and nonroad engines fueled with gasoline, PFCs, and gasoline distribution. We
have also calculated the cost per ton of emissions reduced in the year 2030 using the annual costs
and emissions reductions in that year alone. This number represents the long-term cost per ton
of emissions reduced. All costs are in 2003 dollars.
To calculate the cost per ton for each pollutant reduced under the program, we divided
the net present value of the annual costs by the net present value of the annual emissions
reductions. We have not attempted to apportion costs across these various pollutants for purposes
of the cost per ton calculations since there is no distinction in the technologies, or associated
costs, used to control the pollutants. Instead, we have calculated costs per ton by assigning all
costs to each individual pollutant. If we apportioned costs among the pollutants, the costs per ton
presented here would be proportionally lowered depending on what portion of costs were
assigned to the various pollutants. Results are presented using both a 3 percent and 7 percent
discount rate.
This analysis uses the aggregate costs presented in Chapters 8 through 10 for vehicles,
fuels, and PFCs as well as the emissions reductions presented in Chapter 2. In Section 11.1
through 11.3 we present the cost per ton estimates for vehicles, fuels, and PFCs separately. In
Section, 11.4, we present the cost per ton estimates for the combined rule.
11.1 Cost per Ton for Vehicle Standards
We are establishing new cold temperature NMHC standards for light-duty vehicles,
including medium-duty passenger vehicles. The new standard will be phased in from 2010
through 2015. As discussed in Chapter 8, we are projecting costs for R&D and facilities
upgrades. For our cost estimates, we projected that these fixed costs would be recovered over
the first five years of production for R&D and the first ten years of production for facilities
upgrades. We are not projecting any variable costs, so after the first ten years of production, the
overall annualized costs for the new standards are reduced to $0. For vehicles, we are
establishing NMHC standards which would also VOC-based toxics including benzene. We are
also expecting direct PM reductions due to the new NMHC standards. We have estimated
NMHC, total MSATs, benzene, and PM emissions reductions associated with the cold
temperature NMHC standards, as provided in Chapter 2. We have interpolated to estimate the
emissions reductions for intermediate years not modeled. The annualized costs and emissions
reduction estimates in tons for 2009 through 2030 are provided in Table 11.1-1 below.
11-2
-------
Final Regulatory Impact Analysis
Table 11.1-1 Aggregate Annualized Vehicle Costs and Emissions Reductions
Calendar Year
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
Cost
$0
$11,118,971
$11,772,829
$12,535,232
$13,297,635
$13,406,181
$12,860,869
$12,207,011
$11,444,608
$10,682,205
$10,573,659
$0
$0
$0
$0
$0
$0
$0
$0
$0
$0
$0
NMHC
Reduction (tons)
0
151,748
185,655
219,562
253,470
287,377
321,284
362,900
404,516
446,131
487,747
529,363
564,703
600,043
635,383
670,723
706,063
741,402
776,742
812,082
847,422
882,762
Benzene
Reduction (tons)
0
7,939
9,665
11,391
13,118
14,844
16,570
18,675
20,781
22,886
24,992
27,097
28,891
30,685
32,479
34,273
36,067
37,861
39,655
41,449
43,243
45,037
MSAT
Reduction (tons)
0
51,987
63,136
74,285
85,433
96,582
107,731
121,586
135,441
149,297
163,152
177,007
188,789
200,570
212,352
224,134
235,916
247,697
259,479
271,261
283,042
294,824
PM
Reduction (tons)
0
1,414
2,544
3,675
4,806
5,937
7,068
7,984
8,899
9,815
10,730
11,646
12,424
13,201
13,979
14,756
15,534
16,311
17,089
17,866
18,644
19,421
We have calculated the costs per ton using the net present value of the annualized costs of
the program from 2009 through 2030 and the net present value of the annual emission reductions
through 2030. We have also calculated the cost per ton of emissions reduced in the year 2030
using the annual costs and emissions reductions in that year alone. This number represents the
long-term cost per ton of emissions reduced. As noted above, we have calculated costs per ton by
assigning all costs to each individual pollutant. The results for each pollutant are provided in
Table 11.1-2.
-------
Final Regulatory Impact Analysis
Table 11.1-2. Vehicle Aggregate Cost per Ton and Long-Term Annual Cost Per Ton
($2003)
NMHC
Benzene
Total MSATs
Direct PM
Discounted
Lifetime
Cost per ton at 3%
$14
$270
$42
$650
Discounted
Lifetime
Cost per ton at 7%
$18
$360
$54
$870
Long-Term Cost
per Ton in 2030
$0
$0
$0
$0
11.2 Cost Per Ton for Fuel Benzene Standard
We are adopting a new benzene fuel content standard which will go into effect in 2011.
We have estimated the costs and benzene reductions for the new standards, which are provided
in Chapters 9 and 2, respectively. Table 11.2-1 provides the estimated annualized aggregate
costs and emissions reductions associated with the standard through 2030. The cost per ton
estimates include costs and emission reductions that will occur from all vehicles and nonroad
engines fueled with gasoline, as well as reductions from PFCs and gasoline distribution.
11-4
-------
Final Regulatory Impact Analysis
Table 11.2-1 Aggregate Annualized Fuels Costs and Benzene Reductions
Calendar Year
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
Cost
$354,384,659
$360,089,040
$365,080,373
$369,715,182
$374,349,992
$378,984,801
$383,619,610
$388,254,420
$393,245,753
$397,880,563
$402,158,848
$406,437,134
$410,358,896
$414,280,657
$418,558,943
$422,837,229
$427,472,038
$432,106,847
$436,385,133
$441,019,943
Benzene
Reduction (tons)
18,095
17,975
17,855
17,735
17,615
17,616
17,616
17,617
17,617
17,618
17,821
18,023
18,226
18,428
18,631
18,833
19,036
19,238
19,441
19,643
The cost per ton of benzene reductions for fuels are shown in Table 11.2-2 using this
same methodology as noted above.
Table 11.2-2. Fuel Benzene Aggregate Cost per Ton and Long-Term Annual Cost Per Ton
($2003)
Benzene
Discounted
Lifetime
Cost per ton at 3%
$22,400
Discounted
Lifetime
Cost per ton at 7%
$23,100
Long-Term Cost
per Ton in 2030
$22,500
11-5
-------
Final Regulatory Impact Analysis
11.3 Cost Per Ton for PFCs
We are adopting an HC standard for PFCs that will go into effect beginning in 2009. The
estimated costs for the standard, and gasoline fuel savings, are presented in Chapter 10 and the
emissions reductions are provided in Chapter 2. The new HC standard will also reduce VOC-
based toxics including benzene. The stream of annualized costs, gasoline fuel savings, and
emissions reduction estimates in tons for HC, benzene, and total MSATs for PFCs are provided
in Table 11.3-1.
Table 11.3-1 Aggregate Annualized Portable Fuel Container Costs and Emissions Reductions
Calendar Year
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
Cost
$58,070,171
$58,674,056
$59,290,018
$59,918,300
$60,559,148
$33,336,886
$34,003,624
$34,683,696
$35,377,370
$36,084,918
$36,806,616
$37,542,748
$38,293,603
$39,059,475
$39,840,665
$40,637,478
$41,450,228
$42,279,232
$43,124,817
$43,987,313
$44,867,059
$45,764,401
Fuel
Savings
$15,346,933
$30,693,867
$48,298,000
$65,901,627
$83,505,760
$101,109,387
$102,522,480
$103,935,898
$105,349,189
$106,762,481
$108,175,772
$109,589,064
$111,056,401
$112,523,738
$113,991,075
$115,458,412
$116,925,749
$118,393,086
$119,860,423
$121,327,760
$122,795,097
$124,262,434
HC
Reduction (tons)
30,290
60,580
95,325
130,069
164,814
199,558
202,347
205,137
207,926
210,715
213,505
216,294
219,190
222,086
224,982
227,878
230,775
233,671
236,567
239,463
242,359
245,255
Benzene
Reduction (tons)
100
200
294
389
483
578
672
681
690
700
709
718
728
737
747
756
766
776
785
795
804
814
MSAT
Reduction (tons)
2,590
5,179
8,149
11,120
14,090
17,060
17,357
17,596
17,835
18,075
18,314
18,553
18,801
19,050
19,298
19,546
19,795
20,043
20,291
20,539
20,788
21,036
Table 11.3-2 provides estimated cost per ton for both overall HC reductions, overall
MSAT reductions, and for benzene reductions. As with vehicles, we have calculated costs per
ton by assigning all costs to each individual pollutant. If we apportioned costs among the
pollutants, the costs per ton presented here would be proportionally lowered depending on what
portion of costs were assigned to the various pollutants. The cost per ton estimates are presented
11-6
-------
Final Regulatory Impact Analysis
with and without gasoline fuel savings. Where the fuel savings outweigh the costs, the table
presents cost per ton as $0, rather than calculating a negative value that has no clear meaning.
Table 11.3-2. PFC Aggregate Cost per Ton and Long-Term Annual Cost Per Ton
($2003)
HC without fuel savings
HC with fuel savings
Total MSATs without
fuel savings
Total MSATs with fuel
savings
Benzene without fuel
savings
Benzene with fuel
saving
Discounted Lifetime
Cost per ton at 3%
$240
$0
$2,800
$0
$74,500
$0
Discounted
Lifetime
Cost per ton at 7%
$270
$0
$3,100
$0
$82,900
$0
Long-Term Cost
per Ton in 2030
$190
$0
$2,200
$0
$56,200
$0
11.4 Cost Per Ton for the Overall Proposal
The cost per ton estimates for each individual program are presented separately in the
sections and tables above, and are part of the justification for each of the programs. For
informational purposes, we also present below the cost per ton for the three programs combined.
For MSATs and benzene, we have estimated overall costs by summing the cost shown above for
fuels, vehicles, and PFCs, including gasoline fuel savings. For MSAT and benzene reductions,
we have accounted for the interaction between reduced fuel benzene content due to the new
standard and the reductions in benzene that are provided by the vehicle and PFC standards.
These emissions reduction estimates are provided in Chapter 2. For HC, we have added the costs
and HC reductions shown above for vehicles and PFCs, including fuel savings. Tables 11.4-1
and 11.4-2 provide the streams of costs and emissions reductions in tons for benzene and HC,
respectively.
11-7
-------
Final Regulatory Impact Analysis
Table 11.4-1 Aggregate Annualized Overall Costs, and Benzene and MSAT Emissions Reductions *
Calendar Year
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
Cost Including
Fuel Savings
$42,723,237
$39,099,160
$377,149,506
$366,640,945
$355,431,396
$315,348,863
$318,692,004
$321,939,611
$325,092,399
$328,259,062
$332,450,256
$325,834,247
$329,396,050
$332,972,871
$336,208,485
$339,459,723
$343,083,421
$346,723,375
$350,736,432
$354,766,400
$358,457,095
$362,521,909
Benzene
Reduction (tons)
100
8,139
26,708
28,327
29,946
31,565
33,206
35,117
37,028
38,938
40,849
42,760
44,588
46,415
48,243
50,070
51,898
53,725
55,553
57,380
59,208
61,035
MSAT
Reduction (tons)
2590
57,166
88,034
101,951
115,869
129,786
140,837
154,730
168,623
182,517
196,410
210,303
234,411
234,613
246,667
258,721
270,775
282,828
294,882
306,936
318,990
330,844
includes fuels, vehicles, and portable fuel containers
11-8
-------
Final Regulatory Impact Analysis
Table 11.4-2 Aggregate Annualized Overall Costs and HC Emissions Reductions*
Calendar Year
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
Cost Including
Fuel Savings
$42,723,237
$39,099,160
$22,764,847
$6,551,906
-$9,648,977
-$54,366,320
-$55,657,987
-$57,045,190
-$58,527,211
-$59,995,358
-$60,795,497
-$72,046,316
-$72,762,798
-$73,464,263
-$74,150,410
-$74,820,934
-$75,475,522
-$76,113,854
-$76,735,606
-$77,340,447
-$77,928,038
-$78,498,034
HC
Reduction
30,290
212,328
280,980
349,631
418,284
486,935
523,631
568,036
612,442
656,847
701,252
745,657
783,893
822,129
860,365
898,601
936,837
975,073
1,013,309
1,051,545
1,089,781
1,128,017
* includes vehicles and gas cans
Table 11.4-3 provides the estimated combined cost per ton estimates for benzene,
MSATs and HC. The HC estimates are reported as $0 because the gasoline fuel savings from
PFCs offsets the combined costs of the vehicle and PFC programs.
11-9
-------
Final Regulatory Impact Analysis
Table 11.4-3. Overall Aggregate Cost per Ton and Long-Term Annual Cost Per Ton
($2003)
Benzene for fuels,
vehicles, and PFCs
combined
Total MSATs for fuels,
vehicles, and PFCs
combined
HC for vehicles and
PFCs combined
Discounted Lifetime
Cost per ton at 3%
$8,200
$1,700
$0
Discounted
Lifetime
Cost per ton at 7%
$8,600
$1,800
$0
Long-Term Cost
per Ton in 2030
$5,900
$1,100
$0
11-10
-------
Final Regulatory Impact Analysis
Chapter 12: Table of Contents
Chapter 12: Cost-Benefit Analysis 2
12.1 Overview 2
12.2 Air Quality Impacts 10
12.2.1 PM Air Quality Impact Estimation 10
12.3 PM-Related Health Benefits Estimation - Methods and Inputs 12
12.4 Benefits Analysis Results for the Final Cold Temperature Vehicle Standards 19
12.5 Unquantified Health and Welfare Effects 22
12.5.1 Human Health Impact Assessment 23
12.5.2 Welfare Impact Assessment 24
12.5.2.1 Visibility Benefits 24
12.5.2.2 Agricultural and Forestry Benefits 24
12.5.2.2.1 Agricultural Benefits 24
12.5.2.2.2 Forestry Benefits 25
12.5.2.3 Benefits from Reductions in Materials Damage 25
12.5.3 UVb Exposure 25
12.6 Methods for Describing Uncertainty 26
12.6.1 Analysis of Statistical Uncertainty 27
12.6.1.1 Monte Carlo Approach 29
12.6.1.2 Monte Carlo Results 31
12.6.2 Additional Approaches to Characterizing Uncertainty Related to PM-Mortality.. 32
12.6.2.1 Uncertainty Associated with the Concentration-Response Function 33
12.6.2.2 PM2.5-Mortality Cutpoint/Threshold Analysis 34
12.7 Health-Based Cost Effectiveness Analysis 36
12.8 Comparison of Costs and Benefits 38
12-1
-------
Final Regulatory Impact Analysis
Chapter 12: Cost-Benefit Analysis
12.1 Overview
Mobile sources are significant contributors to hazardous air pollutant emissions ("air
toxics") across the country and into the future. The Agency has determined that these emissions
cause or contribute to air pollution which may reasonably be anticipated to endanger public
health or welfare, and is therefore establishing standards to control these emissions. The health-
and environmentally-related effects associated with these emissions are a classic example of an
externality-related market failure. An externality occurs when one party's actions impose
uncompensated costs on another party. The final MS AT standards will help correct this market
failure.
EPA is required by Executive Order (E.O.) 12866 to estimate the benefits and costs of
major new pollution control regulations. Accordingly, the analysis presented here attempts to
answer three questions: (1) what are the physical health and welfare effects of changes in
ambient paniculate matter (PM) resulting from direct PM emission reductions related to the cold
temperature standards? (2) what is the monetary value of the changes in effects attributable to the
final rule? and (3) how do the monetized benefits compare to the costs? It constitutes one part of
EPA's thorough examination of the relative merits of this regulation. At the same time, EPA
notes that this analysis is for purposes of Executive Order 12866, rather than for purposes of
showing that the final rule satisfies the requirements of section 202(1)(2) of the Act. That
provision requires that emission reductions of mobile source air toxics be reduced to the greatest
amount achievable with available technologies, considering cost among other factors. Section
202(1)(2) thus does not require a weighing of costs and benefits in determining what standards
are achievable, and EPA did not do so in determining what standards to adopt.
This chapter reports EPA's analysis of a subset of the public health and welfare impacts
and associated monetized benefits to society associated with the final standards. In terms of
emission benefits, we expect to see significant reductions in mobile source air toxics (MSATs)
from the vehicle, fuel and PFC standards; reductions in VOCs (an ozone and PM2.5 precursor)
from the cold temperature vehicle and PFC standards; and reductions in direct PM2.s from the
cold temperature vehicle standards. When translating emission benefits to health effects and
monetized values, however, we have chosen to quantify only the PM-related benefits associated
with the cold temperature vehicle standards.
We estimate that the final standards will reduce cancer and noncancer risk from reduced
exposure to MSATs (as described in Chapter 3). However, we do not translate this risk
reduction into benefits. We also do not quantify the benefits related to ambient reductions in
ozone or PM2.s due to the VOC emission reductions that will occur as a result of the final
standards. We describe in more detail below why these benefits are not quantified.
The analysis presented in this chapter uses a methodology generally consistent with
benefits analyses performed for the recent analysis of the Clean Air Interstate Rule (CAIR)
12-2
-------
Final Regulatory Impact Analysis
standards and the Clean Air Nonroad Diesel Rule (CAND).1:2 For this reason, the current
chapter avoids repeating this information and refers to the appropriate sections of each RIA. The
benefits analysis relies on two major components:
1) Calculation of the impact of the cold temperature vehicle standards on the national direct
PM emissions inventory for two future years (2020 and 2030).A
2) A benefits analysis to determine the changes in human health, both in terms of physical
effects and monetary value, based on a PM benefits transfer approach that scales CAND
results (see Section 12.2.).
A wide range of human health and welfare effects are linked to the emissions of direct
PM and its resulting impact on ambient concentrations of PM2.5. Potential human health effects
associated with PM2 5 range from premature mortality to morbidity effects linked to long-term
(chronic) and shorter-term (acute) exposures (e.g., respiratory and cardiovascular symptoms
resulting in hospital admissions, asthma exacerbations, and acute and chronic bronchitis [CB]).
Welfare effects potentially linked to PM include materials damage and visibility impacts.
Table 12.1-1 summarizes the annual monetized health and welfare benefits associated
with the cold temperature standards for two years, 2020 and 2030. The PM2.s benefits are scaled
based on relative changes in direct PM emissions between this rule and the proposed Clean Air
Nonroad Diesel (CAND) rule.8 As explained in Section 12.2.1 of this chapter, the PM2.s
benefits scaling approach is limited to those studies, health impacts, and assumptions that were
used in the proposed CAND analysis. As a result, PM-related premature mortality is based on
the updated analysis of the American Cancer Society cohort (ACS; Pope et al., 2002). However,
it is important to note that since the CAND rule, EPA's Office of Air and Radiation (OAR) has
adopted a different format for its benefits analysis in which characterization of the uncertainty in
the concentration-response function is integrated into the main benefits analysis. Within this
context, additional data sources are available, including a recent expert elicitation and updated
analysis of the Six-Cities Study cohort (Laden et al., 2006). Please see the PM NAAQS RIA for
an indication of the sensitivity of our results to use of alternative concentration-response
functions.
The analysis presented here assumes a PM threshold of 3 ug/m3, equivalent to
background. Through the RIA for CAIR, EPA's consistent approach had been to model
premature mortality associated with PM exposure as a nonthreshold effect; that is, with harmful
effects to exposed populations modeled regardless of the absolute level of ambient PM
concentrations. This approach had been supported by advice from EPA's technical peer review
panel, the Science Advisory Board's Health Effects Subcommittee (SAB-HES). However,
A We consider two future years for analysis (2020 and 2030). Gas can, vehicle, and fuels controls will be fully
implemented by 2020. However, for vehicles, the in-use fleet will not be fully turned over to vehicles meeting the
new standards by 2020. Therefore, we have analyzed 2030 to represent a more fully turned over fleet.
B Due to time and resource constraints, EPA scaled the final CAND benefits estimates from the benefits estimated
for the CAND proposal. The scaling approach used in that analysis, and applied here, is described in the RIA for the
final CAND rule.2
12-3
-------
Final Regulatory Impact Analysis
EPA's most recent PM2 5 Criteria Document concludes that "the available evidence does not
either support or refute the existence of thresholds for the effects of PM on mortality across the
range of concentrations in the studies," (p. 9-44).3 Furthermore, in the RIA for the PM NAAQS
we used a threshold of 10 ug/m3 based on recommendations by CAS AC for the Staff Paper
analysis. We consider the impact of a potential, assumed threshold in the PM-mortality
concentration response function in Section 12.6.2.2 of the RIA.
Table 12.1-1. Estimated Monetized PM-Related Health Benefits of the Final Mobile Source
Air Toxics Standards: Cold Temperature Controls
Using a 3% discount rate
Using a 7% discount rate
Total Benefits3' b'c (billions 2003$)
2020
$3.3 + B
$3.0 + B
2030
$6.3 + B
$5.7 + B
Benefits include avoided cases of mortality, chronic illness, and other morbidity health endpoints. PM-related
mortality benefits estimated using an assumed PM threshold at background levels (3 ug/m3). There is
uncertainty about which assumed threshold to use and this may impact the magnitude of the total benefits
estimate. For a more detailed discussion of this issue, please refer to Section 12.6.2.2 of the RIA.
For notational purposes, unqualified benefits are indicated with a "B" to represent the sum of additional
monetary benefits and disbenefits. A detailed listing of unqualified health and welfare effects is provided in
Table 12.1-2 of the RIA.
Results reflect the use of two different discount rates: 3 and 7 percent, which are recommended by EPA's
Guidelines for Preparing Economic Analyses4 and OMB Circular A-4.5 Results are rounded to two significant
digits for ease of presentation and computation.
This chapter specifically assesses the direct PM-related benefits of the cold temperature
vehicle standards. Other standards in this rulemaking, such as the cold temperature vehicle and
PFC standards, will also reduce the national emissions inventory of precursors to ozone, such as
VOCs. Exposure to ozone has been linked to a variety of respiratory effects including hospital
admissions and illnesses resulting in school absences. In addition, recent analyses (reflected in
the 2006 Ozone Criteria Document for the current ozone review cycle under section 109(d) of
the Act) provide evidence that short-term ozone exposure is associated with increased premature
mortality independent of exposure to PM. Ozone can also adversely affect the agricultural and
forestry sectors by decreasing yields of crops and forests. Although ozone benefits are typically
quantified in regulatory impact analyses, we do not evaluate them for this analysis.
We estimate that there will be demonstrable VOC reductions as a result of the cold
temperature vehicle standards. However, we assume that these emissions will not have a
measurable impact on ozone formation since the standards seek to reduce VOC emissions at cold
ambient temperatures and ozone formation is primarily a warm ambient temperature issue.
There will, however, likely be benefits associated with VOC emission reductions resulting from
the PFC standards. In Chapter 3, we discuss that the ozone modeling conducted for the PFC
standards results in a net reduction in the average population-weighted ozone design value metric
measured within the modeled domain (37 Eastern states and the District of Columbia). The net
12-4
-------
Final Regulatory Impact Analysis
improvement is very small, however, and will likely lead to negligible monetized benefits. We
therefore do not estimate ozone benefits for the PFC standards due to the magnitude of this
change and the uncertainty present in the modeling. Instead, we acknowledge that this analysis
may underestimate the benefits associated with reductions in ozone precursor emissions achieved
by the various standards, and we will discuss them qualitatively within this chapter.
The VOC reductions resulting from the cold temperature vehicle standards and PFC
standards will also likely reduce secondary PM2.5 formation. However, we did not quantify the
impacts of these reductions on ambient PM2 5 or estimate any resulting benefits. As described
further below, we estimated PM benefits by scaling from a previous analysis, and this analysis
did not examine the relationship between VOC reductions and ambient PM. As a result, we did
not quantify PM benefits associated with this rule's VOC reductions, and we acknowledge that
this analysis may therefore underestimate benefits.
There will also be significant reduction in emissions of mobile source-related air toxics
with the final standards in place (including benzene, 1,3-butadiene, formaldehyde, acetaldehyde,
acrolein, naphthalene, and other toxic air pollutants). While there will be substantial benefits
associated with air toxic pollutant reductions, notably with regard to reductions in exposure and
risk (see Chapter 3), we do not attempt to extrapolate this risk reduction to monetize those
benefits. This is primarily because available tools and methods to assess air toxics risk from
mobile sources at the national scale are not adequate for extrapolation to benefits assessment.
The best suite of tools and methods currently available for assessment at the national
scale are those used in the National-Scale Air Toxics Assessment (NATA; these tools are
discussed in Chapter 3). The EPA Science Advisory Board specifically commented in their
review of the 1996 NATA that these tools were not yet ready for use in a national-scale benefits
analysis, because they did not consider the full distribution of exposure and risk, or address sub-
chronic health effects.6 While EPA has since improved the tools, there remain critical
limitations for estimating incidence and assessing monetized benefits of reducing mobile source
air toxics.
In addition to inherent limitations in the tools for national-scale modeling of air quality
and exposure, there is a lack of epidemiology data for air toxics in the general population.
Therefore, we must rely on health endpoints estimated from occupational or animal exposure
studies. There are several limitations in our ability to quantify and value changes in incidence of
health effects. For the MSATs of greatest concern, we are currently unable to estimate cessation
lag, which is the time between reduction in exposure and decline in risk to "steady state level."
We have not resolved the analytical challenges associated with quantifying partial lifetime
probabilities of cancer for different age groups or estimating changes in survival rates over time.
In addition, we are currently unable to estimate the premium people are willing to pay to avoid
cancer. There is also no data on the cost of treating leukemia cases and little data on how to
value non-fatal leukemias. Given all the limitations in our ability to develop incidence estimates
and to monetize willingness to pay or treatment costs, a quantitative benefits analysis for
benzene would not be meaningful or informative. We continue to work to address these
12-5
-------
Final Regulatory Impact Analysis
limitations, and we are exploring the feasibility of a quantitative benefits assessment for air
toxics through a benzene case study as part of the revised study of "The Benefits and Costs of the
Clean Air Act" (also known as the "Section 812" report). c In this case study, we are attempting
to monetize the benefits of reduced cancer incidence, specifically leukemia, and are not
addressing other cancer or noncancer endpoints.
Table 12.1-2 lists the full complement of human health and welfare effects associated
with PM, ozone and air toxics, and identifies those effects that are quantified for the primary
estimate and those that remain unquantified because of current limitations in methods or
available data.
Table 12.1-2. Human Health and Welfare Effects of Pollutants Affected by the Final
MSAT Standards
Pollutant/Effect
Quantified and Monetized in Base
Estimates"
Unquantified Effects - Changes in:
PM/Healthb
Premature mortality based on cohort
study estimates0
Bronchitis: chronic and acute
Hospital admissions: respiratory
and cardiovascular
Emergency room visits for asthma
Nonfatal heart attacks (myocardial
infarction)
Lower and upper respiratory illness
Minor restricted-activity days
Work loss days
Asthma exacerbations (asthmatic
population)
Respiratory symptoms (asthmatic
population)
Infant mortality
Premature mortality: short-term exposures'1
Subchronic bronchitis cases
Low birth weight
Pulmonary function
Chronic respiratory diseases other than chronic bronchitis
Nonasthma respiratory emergency room visits
UVb exposure (+/-)e
PM/Welfare
Visibility in Southeastern Class I areas
Visibility in northeastern and Midwestern Class I areas
Household soiling
Visibility in we stern U.S. Class I areas
Visibility in residential and non-Class I areas
UVb exposure (+/-)e
The analytic blueprint for the Section 812 benzene case study can be found at
http://www.epa.gov/air/sect812/appendixi51203 .pdf.
12-6
-------
Final Regulatory Impact Analysis
Pollutant/Effect
Quantified and Monetized in Base
Estimates"
Unquantified Effects - Changes in:
Ozone/Healthf
Premature mortality: short-term exposures8
Hospital admissions: respiratory
Emergency room visits for asthma
Minor restricted-activity days
School loss days
Asthma attacks
Cardiovascular emergency room visits
Acute respiratory symptoms
Chronic respiratory damage
Premature aging of the lungs
Nonasthma respiratory emergency room visits
UVb exposure (+/-)e
Ozone/Welfare
Decreased outdoor worker productivity
Yields for:
- Commercial forests
- Fruits and vegetables, and
- Other commercial and noncommercial crops
Damage to urban ornamental plants
Recreational demand from damaged forest aesthetics
Ecosystem functions
UVb exposure (+/-)e
MSAT Health11
Cancer (benzene, 1,3-butadiene, formaldehyde,
acetaldehyde, naphthalene)
Anemia (benzene)
Disruption of production of blood components (benzene)
Reduction in the number of blood platelets (benzene)
Excessive bone marrow formation (benzene)
Depression of lymphocyte counts (benzene)
Reproductive and developmental effects (1,3-butadiene)
Irritation of eyes and mucus membranes (formaldehyde)
Respiratory irritation (formaldehyde)
Asthma attacks in asthmatics (formaldehyde)
Asthma-like symptoms in non-asthmatics (formaldehyde)
Irritation of the eyes, skin, and respiratory tract
(acetaldehyde)
Upper respiratory tract irritation and congestion (acrolein)
Neurotoxicity (n-hexane, toluene, xylenes)
MSAT Welfare11
Direct toxic effects to animals
Bioaccumulation in the food chain
Damage to ecosystem function
Odor
a Primary quantified and monetized effects are those included when determining the primary estimate of total
monetized benefits of the final standards.
b In addition to primary economic endpoints, there are a number of biological responses that have been associated
with PM health effects including morphological changes and altered host defense mechanisms. The public health
impact of these biological responses may be partly represented by our quantified endpoints.
0 Cohort estimates are designed to examine the effects of long term exposures to ambient pollution, but relative risk
estimates may also incorporate some effects due to shorter-term exposures (see Kunzli, 2001 for a discussion of this
issue).7
12-7
-------
Final Regulatory Impact Analysis
d While some of the effects of short-term exposure are likely to be captured by the cohort estimates, there may be
additional premature mortality from short-term PM exposure not captured in the cohort estimates included in the
primary analysis.
e May result in benefits or disbenefits. See Section 12.5.3. for more details.
f In addition to primary economic endpoints, there are a number of biological responses that have been associated
with ozone health including increased airway responsiveness to stimuli, inflammation in the lung, acute
inflammation and respiratory cell damage, and increased susceptibility to respiratory infection. The public health
impact of these biological responses may be partly represented by our quantified endpoints.
gEPA sponsored a series of meta-analyses of the ozone mortality epidemiology literature, published in the July 2005
volume of the journal Epidemiology, which found that short-term exposures to ozone may have a significant effect
on daily mortality rates, independent of exposure to PM. EPA is currently considering how to include an estimate of
ozone mortality in its benefits analyses.
h The categorization of unqualified toxic health and welfare effects is not exhaustive.
Figure 12.1-1 illustrates the major steps in the PM benefits analysis. Given the change in
direct PM emissions modeled for the cold temperature vehicle standards, we use a benefits
transfer approach to scale PM benefits estimated for the CAND analysis (see Section 12.2 for a
description of the scaling approach). For the CAND analysis, EPA ran a sophisticated
photochemical air quality model, the Regional Modeling System for Aerosols and Deposition
(REMSAD), to estimate baseline and post-control ambient concentrations of PM for each future
year (2020 and 2030). The estimated changes in ambient concentrations were then combined
with population projections to estimate population-level potential exposures to changes in
ambient concentrations. Changes in population exposure to ambient air pollution were then
input to impact functions0 to generate changes in the incidence of health effects. The resulting
changes in incidence were then assigned monetary values, taking into account adjustments to
values for growth in real income out to the year of analysis (values for health and welfare effects
are in general positively related to real income levels). Values for individual health and welfare
effects were summed to obtain an estimate of the total monetary value of the changes in
emissions. Finally, we scale the CAND results to reflect the magnitude of the direct PM
emissions changes we estimate will occur as a result of the cold temperature standards.
Benefits estimates calculated for the CAND analysis, and scaled for the cold temperature
standards, were generated using the Environmental Benefits Mapping and Analysis Program
(BenMAP). BenMAP is a computer program developed by EPA that integrates a number of the
modeling elements used in previous RIA's (e.g., interpolation functions, population projections,
health impact functions, valuation functions, analysis and pooling methods) to translate modeled
air concentration estimates into health effect incidence estimates and monetized benefit
D The term "impact function" as used here refers to the combination of a) an effect estimate obtained from the
epidemiological literature, b) the baseline incidence estimate for the health effect of interest in the modeled
population, c) the size of that modeled population, and d) the change in the ambient air pollution metric of interest.
These elements are combined in the impact function to generate estimates of changes in incidence of the health
effect. The impact function is distinct from the C-R function, which strictly refers to the estimated equation from
the epidemiological study relating incidence of the health effect and ambient pollution. We refer to the specific
value of the relative risk or estimated coefficients in the epidemiological study as the "effect estimate." In
referencing the functions used to generate changes in incidence of health effects for this RIA, we use the term
"impact function" rather than C-R function because "impact function" includes all key input parameters used in the
incidence calculation.
12-8
-------
Final Regulatory Impact Analysis
estimates. Interested parties may wish to consult the webpage
http://www.epa.gov/ttn/ecas/benmodels.html for more information.
Figure 12.1-1. Key Steps in Air Quality Modeling Based Benefits Analysis
Emissions inventories
(CAND)
PROCESSES
Model CAND baseline and
post-control ambient PM23
(REMSAD)
Interpolation of projected air
concentration surfaces (base and
control)
BenMAP
integrated
model
Model population exposure to
changes in ambient concentrations
Estimate expected changes in
human health outcomes
Estimate monetary value of
changes in human health
Adjust monetary values for growth
in real income to year of analysis
Sum health and we If are monetary
values to obtain total monetary
benefits
All of the benefit estimates for the final control options in this analysis are based on an
analytical structure and sequence similar to that used in the benefits analyses for the CAND final
rule, the CAIR rule, and, when feasible, the final PM NAAQS analysis.E By adopting the major
design elements, models, and assumptions developed in recent RIAs, we rely on methods that
have already received extensive review by the independent Science Advisory Board (SAB), by
the public, and by other federal agencies. In addition, we will be working through the next
section 812 prospective study to enhance our methods.F
This chapter is organized as follows. In Section 12.2, we provide an overview of the air
quality impacts modeled for the final standards that are used as inputs to the benefits analysis. In
E See: Clean Air Nonroad Diesel final rule (69 FR 38958, June 29, 2004); Clean Air Interstate final rule (70 FR
25162, May 12, 2005); PM NAAQS (71 FR61144, Oct. 17, 2006).
F Interested parties may want to consult the webpage: http://www.epa.gov/sciencel regarding components of the
812 prospective analytical blueprint.
12-9
-------
Final Regulatory Impact Analysis
Section 12.3, we document key differences between this benefits analysis and the benefits
analysis completed for the final CAIR and CAND rules. This section also presents and discusses
the key inputs and methods used in the benefits analysis. In Section 12.4, we report the results of
the analysis for human health and welfare effects. Section 12.5 qualitatively describes benefits
categories that are omitted from this analysis, due either to inadequate methods or resources.
Section 12.6 discusses how we incorporate uncertainty into our analysis. Section 12.7 discusses
the health-based cost-effectiveness analysis for the final standards. Finally, in Section 12.8, we
present a comparison of the costs and benefits associated with the final standards.
12.2 Air Quality Impacts
This section summarizes the methods for and results of estimating air quality for the 2020
and 2030 base case and final control scenario for the purposes of the benefits analysis. EPA has
focused on the health, welfare, and ecological effects that have been linked to ambient changes
in PM2.5 related to direct PM emission reductions estimated to occur due to the cold temperature
vehicle standards. We do this by scaling the modeled relationship between emissions and
ambient PM concentrations observed for the CAND analysis.8
12.2.1 PM Air Quality Impact Estimation
To estimate PM2.5 benefits resulting from the cold temperature vehicle standards, we rely
on a benefits transfer technique. The benefits transfer approach uses as its foundation the
relationship between emission reductions and ambient PM2.5 concentrations modeled for the
Clean Air Nonroad Diesel (CAND) proposal.0 For a given future year, we first calculate the
ratio between CAND direct PM2.5 emission reductions and direct PM2.5 emission reductions
associated with the final standards (final emission reductions/CAND emission reductions,
displayed in Table 12.2-1). We multiply this ratio by the percent that direct PM2.5 contributes
towards population-weighted reductions in total PM2 5 due to the CAND standards (displayed in
Table 12.2-2). This calculation results in a "benefits apportionment factor" for the relationship
between direct PM emissions and primary PM2 5 (displayed in Table 12.2-3), which is then
applied to the BenMAP-based incidence and monetized benefits from the CAND proposal. In
this way, we apportion the results of the proposed CAND analysis to its underlying direct PM
emission reductions and scale the apportioned benefits to reflect differences in emission
reductions between the two rules.H This benefits transfer method is consistent with the approach
used in other recent mobile and stationary source rules.1 We refer the reader to the final CAND
RIA for more details on this benefits transfer approach.9
G See 68 FR 28327, May 23, 2003.
H Note that while the final MS AT standards also control VOCs, which contribute to PM formation, the benefits
transfer scaling approach only scales benefits based on NOx, SO2, and direct PM emission reductions. PM benefits
will likely be underestimated as a result, though we are unable to estimate the magnitude of the underestimation.
1 See: Clean Air Nonroad Diesel final rule (69 FR 38958, June 29, 2004); Nonroad Large Spark-Ignition Engines
and Recreational Engines standards (67 FR 68241, November 8, 2002); Final Industrial Boilers and Process Heaters
NESHAP (69 FR 55217, September 13, 2004); Final Reciprocating Internal Combustion Engines NESHAP (69 FR
33473, June 15, 2004); Final Clean Air Visibility Rule (EPA-452/R-05-004, June 15, 2005); Ozone Implementation
Rule (70 FR 71611, November 29, 2005).
12-10
-------
Final Regulatory Impact Analysis
Table 12.2-1. Comparison of 48-state Emission Reductions in 2020 and 2030 Between the
CAND and Final Cold Temperature Standards
Emissions Species
2020
Direct PM2 5
2030
Direct PM2 5
Reduction from Baseline (tons)
CAND Modeling
Inputs3
98,121
138,208
Cold Temperature
Emissions
Changes'3
11,646
19,421
Ratio of Reductions
(MSAT/ CAND)
0.119
0.141
a Includes all affected nonroad sources: land-based, recreational marine, commercial
marine, and locomotives. See the CAND RIA for more information regarding the
CAND emission inventories.
b Includes changes to the light duty onroad vehicles inventory.
Table 12.2-2. Apportionment of Modeled CAND Preliminary Control Option Population-
weighted Change in Ambient PMi.s to Nitrate, Sulfate, and Primary Particles
Total PM2 5
Sulfate
Nitrate
Primary PM
2020
Population-weighted
Change (ug/m3)
Percent of Total
Change
0.316
0.071
0.041
0.203
22.5%
13.1%
64.4%
2030
Population-weighted
Change (ug/m3)
Percent of Total
Change
0.438
0.090
0.073
0.274
20.5%
16.8%
62.7%
Source: CAND RIA, Chapter 9.
12-11
-------
Final Regulatory Impact Analysis
Table 12.2-3. Calculation of PM2.5 Benefits Apportionment Factor for Final
Cold Temperature-Related Direct PM Emission Reductions
Direct PM
Emissions
2020
Ratio of
Emission
Reductions3
(1)
0.119
% of Total
Ambient
Changeb
(2)
0.644
Benefits
Apportionment
Factor
(1*2)
0.088
2030
Ratio of
Emission
Reductions3
(3)
0.141
% of Total
Ambient
Changeb
(4)
0.627
Benefits
Apportionment
Factor
(3*4)
0.076
3 Calculated by dividing cold temperature vehicle emission reductions by CAND emission reductions. See Table
12.2-1.
b See Table 12.2-2.
12.3 PM-Related Health Benefits Estimation - Methods and Inputs
The analytical approach used in this benefits analysis is largely the same approach used
in the Final CAIR and Final CAND benefits analyses and the reader is referred to each RIA for
details on the benefits methods and inputs. This analysis, however, also reflects advances in data
and methods in epidemiology, economics, and health impact estimation. Updates to the
assumptions and methods used in estimating PM2.5-related benefits since the analysis for the
CAIR and CAND rules include the following:
• We have updated our projections of mortality incidence rates to be consistent with the
U.S. Census population projections that form the basis of our future population
estimates. This approach combines Centers for Disease Control (CDC) county-level
mortality rate data for the years 1996-1998 with US Census Bureau mortality
projections out to 2050. To estimate age- and county-specific mortality rates in years
2020 and 2030, we calculated adjustment factors, based on a series of Census Bureau
projected national mortality rates, to adjust the CDC Wonder age- and county-specific
mortality rates in 1996-1998 to corresponding rates for each future year. This
approach is different than the fixed 1996-1998 CDC mortality rate data used in the
CAND and CAIR analyses, and results in a reduction in mortality impacts in future
years as overall mortality rates are projected to decline for most age groups. A
memorandum drafted by Abt Associates (Abt Associates, 2005) contains complete
details regarding the derivation of mortality rate adjustment factors, and estimation of
future-year mortality rates used in the analysis.10 The scaled mortality benefits for
the final standards have been updated accordingly.
• Use of a revised mortality lag assumption. In the Final CAND, we used a five-year
segmented lag. Since that analysis, upon which the PM benefits transfer scaling
approach is based, the SAB Health Effects Subcommittee (HES) recommended that
until additional research has been completed, EPA should assume a segmented lag
12-12
-------
Final Regulatory Impact Analysis
structure characterized by 30 percent of mortality reductions occurring in the first
year, 50 percent occurring evenly over years 2 to 5 after the reduction in PM2.5, and
20 percent occurring evenly over the years 6 to 20 after the reduction in PM2 5. The
distribution of deaths over the latency period is intended to reflect the contribution of
short-term exposures in the first year, cardiopulmonary deaths in the 2- to 5-year
period, and long-term lung disease and lung cancer in the 6- to 20-year period. For
future analyses, the specific distribution of deaths over time will need to be
determined through research on causes of death and progression of diseases
associated with air pollution. It is important to keep in mind that changes in the lag
assumptions do not change the total number of estimated deaths but rather the timing
of those deaths. This approach is different than the 5-year segmented lag used in the
CAND analysis, and the scaled benefits analysis of the final standards has been
updated accordingly.
For the purposes of this RIA, the health impacts analysis is limited to those health effects
that are directly linked to ambient levels of air pollution and specifically to those linked to PM.
The specific studies from which effect estimates for the primary analysis are drawn are included
in Table 12.3-1. The specific unit values used for economic valuation of health endpoints are
included in Table 12.3-2.
12-13
-------
Final Regulatory Impact Analysis
Table 12.3-1. Endpoints and Studies Used to Calculate Total Monetized Health Benefits"
Endpoint
Pollutant
Study
Study
Population
Premature Mortality
Premature mortality
— ACS cohort study,
all-cause
Premature mortality
— all-cause
PM25
PM25
Pope etal. (2002) n
Woodruff etal. (1997)12
>29 years
Infant (<1 year)
Chronic Illness
Chronic bronchitis
Nonfatal heart attacks
PM25
PM25
Abbey etal. (1995)13
Peters etal. (200 1)14
>26 years
Adults
Hospital Admissions
Respiratory
Cardiovascular
Asthma-related ER
visits
PM25
PM25
PM25
PM25
PM25
PM25
PM25
Pooled estimate:
Moolgavkar (2003)15— ICD 490-496 (COPD)
Ito (2003) 16— ICD 490-496 (COPD)
Moolgavkar (2000) 17— ICD 490-496 (COPD)
Ito (2003)— ICD 480-486 (pneumonia)
Sheppard (2003)18— ICD 493 (asthma)
Pooled estimate:
Moolgavkar (2003)— ICD 390-429 (all cardiovascular)
Ito (2003)— ICD 410-414, 427-428 (ischemic heart
disease, dysrhythmia, heart failure)
Moolgavkar (2000)— ICD 390-429 (all cardiovascular)
Norrisetal. (1999)19
>64 years
20-64 years
>64 years
<65 years
>64 years
20-64 years
0-18 years
Other Health Endpoints
Acute bronchitis
Upper respiratory
symptoms
Lower respiratory
symptoms
Asthma
exacerbations
Work loss days
MRADs
PM25
PM25
PM25
PM25
PM25
PM25
Dockeryetal. (1996)20
Pope etal. (1991)21
Schwartz and Neas (2000) 22
Pooled estimate:
Ostro et al. (200 1)23 (cough, wheeze and shortness of
breath)
Vedal et al. (1998)24 (cough)
Ostro (1987)25
Ostro and Rothschild (1989)26
8-12 years
Asthmatics, 9-
1 1 years
7-14 years
6-18yearsb
18-65 years
18-65 years
a The endpoints and studies used for the primary estimate of benefits associated with the final rule have been
subject to external technical guidance and review, including the Health Effects Subgroup (HES) of the EPA's
Science Advisory Board (SAB) and the Office of Management and Budget (OMB).
b The original study populations were 8 to 13 for the Ostro et al. (2001) study and 6 to 13 for the Vedal et al.
(1998) study. Based on advice from the SAB-HES, we extended the applied population to 6 to 18, reflecting the
common biological basis for the effect in children in the broader age group.
12-14
-------
Final Regulatory Impact Analysis
Table 12.3-2. Unit Values Used for Economic Valuation of Health Endpoints (2000$)a
Health Endpoint
Premature Mortality (Value of a
Statistical Life)
Chronic Bronchitis (CB)
Nonfatal Myocardial Infarction
(heart attack)
3% discount rate
Age 0-24
Age 25^4
Age 45-54
Age 55-65
Age 66 and over
7% discount rate
Age 0-24
Age 25^4
A OP 4S S4
Age 55-65
Age 66 and over
Central Estimate of Value Per Statistical Incidence
1990 Income
Level
$5,500,000
$340,000
$66,902
$74,676
$78,834
$140,649
$66,902
$65,293
$73,149
$76,871
$132,214
$65,293
2020 Income
Level"
$6,600,000
$420,000
$66,902
$74,676
$78,834
$140,649
$66,902
$65,293
$73,149
$76,871
$132,214
$65,293
2030 Income
Level"
$6,800,000
$430,000
$66,902
$74,676
$78,834
$140,649
$66,902
$65,293
$73,149
$76,871
$132,214
$65,293
Derivation of Estimates
Point estimate is the mean of a normal distribution with a 95 percent
confidence interval between $1 and $10 million. Confidence interval is
based on two meta-analyses of the wage-risk VSL literature: $1 million
represents the lower end of the interquartile range from the Mrozek and
Taylor (2002)27 meta-analysis and $10 million represents the upper end of
the interquartile range from the Viscusi and Aldy (2003)28 meta-analysis.
The VSL represents the value of a small change in mortality risk aggregated
over the affected population.
Point estimate is the mean of a generated distribution of WTP to avoid a case
of pollution-related CB. WTP to avoid a case of pollution-related CB is
derived by adjusting WTP (as described in Viscusi et al., [1991]29) to avoid
a severe case of CB for the difference in severity and taking into account the
elasticity of WTP with respect to severity of CB.
Age-specific cost-of-illness values reflect lost earnings and direct medical
costs over a 5-year period following a nonfatal MI. Lost earnings estimates
are based on Cropper and Krupnick (1990). 30 Direct medical costs are based
on simple average of estimates from Russell et al. (1998)31 and Wittels et al.
(1990). 32
Lost earnings:
Cropper and Krupnick ( 1 990). Present discounted value of 5 years of lost
earnings:
age of onset: at 3% at 7%
25-44 $8,774 $7,855
45-54 $12,932 $11,578
55-65 $74,746 $66,920
Direct medical expenses: An average of:
1. Wittels et al. (1990) ($102,658— no discounting)
2. Russell et al. (1998), 5-year period ($22,331 at 3% discount rate; $21,1 13
at 7% discount rate)
(continued)
12-15
-------
Final Regulatory Impact Analysis
Table 12.3-2. Unit Values Used for Economic Valuation of Health Endpoints (2000$)a (continued)
Health Endpoint
Central Estimate of Value Per Statistical
Incidence
1990 Income
Level
2020 Income
Level"
2030 Income
Level"
Derivation of Estimates
Hospital Admissions
Chronic Obstructive Pulmonary
Disease (COPD)
(ICD codes 490-492, 494-496)
Pneumonia
(ICD codes 480-487)
Asthma Admissions
All Cardiovascular
(ICD codes 390-429)
Emergency Room Visits for Asthma
$12,378
$14,693
$6,634
$18,387
$286
$12,378
$14,693
$6,634
$18,387
$286
$12,378
$14,693
$6,634
$18,387
$286
The COI estimates (lost earnings plus direct medical costs) are based on
ICD-9 code-level information (e.g., average hospital care costs, average
length of hospital stay, and weighted share of total COPD category illnesses)
reported in Agency for Healthcare Research and Quality (2000)33
(www.ahrq.gov).
The COI estimates (lost earnings plus direct medical costs) are based on
ICD-9 code-level information (e.g., average hospital care costs, average
length of hospital stay, and weighted share of total pneumonia category
illnesses) reported in Agency for Healthcare Research and Quality (2000)
(www.ahrq.gov).
The COI estimates (lost earnings plus direct medical costs) are based on
ICD-9 code-level information (e.g., average hospital care costs, average
length of hospital stay, and weighted share of total asthma category illnesses)
reported in Agency for Healthcare Research and Quality (2000)
(www.ahrq.gov).
The COI estimates (lost earnings plus direct medical costs) are based on
ICD-9 code-level information (e.g., average hospital care costs, average
length of hospital stay, and weighted share of total cardiovascular category
illnesses) reported in Agency for Healthcare Research and Quality (2000)
(www.ahrq.gov).
Simple average of two unit COI values:
(1) $31 1.55, from Smith et al. (1997)34 and
(2) $260.67, from Stanford et al. (1999). 35
(continued)
12-16
-------
Final Regulatory Impact Analysis
Table 12.3-2. Unit Values Used for Economic Valuation of Health Endpoints (2000$)a (continued)
Health Endpoint
Central Estimate of Value Per Statistical
Incidence
1990 Income
Level
2020 Income
Level"
2030 Income
Level"
Derivation of Estimates
Respiratory Ailments Not Requiring Hospitalization
Upper Respiratory Symptoms (URS)
Lower Respiratory Symptoms (LRS)
Asthma Exacerbations
Acute Bronchitis
$25
$16
$42
$360
$27
$17
$45
$380
$27
$17
$45
$390
Combinations of the three symptoms for which WTP estimates are available
that closely match those listed by Pope et al. result in seven different
"symptom clusters," each describing a "type" of URS. A dollar value was
derived for each type of URS, using mid-range estimates of WTP (lEc,
1994)36 to avoid each symptom in the cluster and assuming additivity of
WTPs. The dollar value for URS is the average of the dollar values for the
seven different types of URS.
Combinations of the four symptoms for which WTP estimates are available
that closely match those listed by Schwartz et al. result in 1 1 different
"symptom clusters," each describing a "type" of LRS. A dollar value was
derived for each type of LRS, using mid-range estimates of WTP (lEc, 1994)
to avoid each symptom in the cluster and assuming additivity of WTPs. The
dollar value for LRS is the average of the dollar values for the 1 1 different
types of LRS.
Asthma exacerbations are valued at $42 per incidence, based on the mean of
average WTP estimates for the four severity definitions of a "bad asthma
day," described in Rowe and Chestnut (1986). 37 This study surveyed
asthmatics to estimate WTP for avoidance of a "bad asthma day," as defined
by the subjects. For purposes of valuation, an asthma attack is assumed to be
equivalent to a day in which asthma is moderate or worse as reported in the
Rowe and Chestnut (1986) study.
Assumes a 6-day episode, with daily value equal to the average of low and
high values for related respiratory symptoms recommended in Neumann et
al. (1994).38
(continued)
12-17
-------
Final Regulatory Impact Analysis
Table 12.3-2. Unit Values Used for Economic Valuation of Health Endpoints (2000$)a (continued)
Health Endpoint
Central Estimate of Value Per Statistical Incidence
1990 Income
Level
2020 Income
Level"
2030 Income
Level"
Derivation of Estimates
Restricted Activity and Work/School Loss Days
Work Loss Days (WLDs)
Minor Restricted Activity Days
(MRADs)
Variable
(national
median = )
$51
$54
$55
County-specific median annual wages divided by 50 (assuming 2 weeks of
vacation) and then by 5 — to get median daily wage. U.S. Year 2000
Census, compiled by Geolytics, Inc.
Median WTP estimate to avoid one MRAD from Tolley et al. (1986).J9
a Although the unit values presented in this table are in year 2000 dollars, all monetized annual benefit estimates associated with the final standards have been inflated to reflect
values in year 2003 dollars. We use the Consumer Price Indexes to adjust both WTP- and COI-based benefits estimates to 2003 dollars from 2000 dollars.40 For WTP-based
estimates, we use an inflation factor of 1.07 based on the CPI-U for "all items." For COI-based estimates, we use an inflation factor of 1.14 based on the CPI-U for medical care.
b Our analysis accounts for expected growth in real income over time. Economic theory argues that WTP for most goods (such as environmental protection) will increase if real
incomes increase. Benefits are therefore adjusted by multiplying the unadjusted benefits by the appropriate adjustment factor to account for income growth over time. For a
complete discussion of how these adjustment factors were derived, we refer the reader to Chapter 9 of the CAND regulatory impact analysis (EPA, 2004). Note that similar
adjustments do not exist for cost-of-illness-based unit values. For these, we apply the same unit value regardless of the future year of analysis.
12-18
-------
EPA typically estimates the welfare impacts of effects such as changes in recreational
visibility (related to reductions in ambient PM) and agricultural productivity (related to
reductions in ambient ozone) in its RIAs of air quality policy. For the analysis of the final
standards, however, we are unable to quantitatively characterize these impacts because of limited
data availability; we are not quantifying ozone benefits related to the final standards and the PM
scaling approach does not provide the spatial detail necessary to attribute specific air quality
improvements to specific areas of visual interest (Class I areas). Instead, we discuss these
welfare effects qualitatively in Section 12.5 of this chapter. We also qualitatively describe the
impacts of other environmental and ecological effects for which we do not have an economic
value.
12.4 Benefits Analysis Results for the Final Cold Temperature Vehicle
Standards
Applying the impact and valuation functions described previously in this chapter to the
estimated changes in PM2 5 associated with the final cold temperature vehicle standards results in
estimates of the changes in physical damages (e.g., premature mortalities, cases, admissions) and
the associated monetary values for those changes. Estimates of physical health impacts are
presented in Table 12.4-1. Monetized values for those health endpoints are presented in Table
12.4-2, along with total aggregate monetized benefits. All of the monetary benefits are in
constant-year 2003 dollars.
12-19
-------
Table 12.4-1. Estimated Reduction in Incidence of Adverse Health Effects Related to the
Final Cold Temperature Standards"
2020 2030
Health Effect Incidence Reduction
PM-Related Endpoints
Premature Mortality1"'0
Adult, age 30+ and Infant, age <1 year 480 880
Chronic bronchitis (adult, age 26 and over) 330 570
Nonfatal myocardial infarction (adults, age 18 and older) 810 1,600
Hospital admissions—respiratory (all ages)d 260 530
Hospital admissions—cardiovascular (adults, age >18)e 210 390
Emergency room visits for asthma (age 18 years and younger) 350 610
Acute bronchitis (children, age 8-12) 780 1,400
Lower respiratory symptoms (children, age 7-14) 9,300 16,000
Upper respiratory symptoms (asthmatic children, age 9-18) 7,000 12,000
Asthma exacerbation (asthmatic children, age 6-18) 12,000 20,000
Work loss days (adults, age 18-65) 62,000 100,000
Minor restricted-activity days (adults, age 18-65) 370,000 600,000
a Incidences are rounded to two significant digits. PM estimates are nationwide.
b PM premature mortality impacts for adults are based on application of the effect estimate derived from the ACS
cohort study (Pope et al., 2002).41 Infant premature mortality based upon studies by Woodruff, et al 1997.42
0 PM-related mortality benefits estimated using an assumed PM threshold at background levels (3 ug/m3). There
is uncertainty about which threshold to use and this may impact the magnitude of the total benefits estimate. For
a more detailed discussion of this issue, please refer to Section 12.6.2.2 of the RIA.
d Respiratory hospital admissions for PM include admissions for COPD, pneumonia, and asthma.
e Cardiovascular hospital admissions for PM include total cardiovascular and subcategories for ischemic heart
disease, dysrhythmias, and heart failure.
12-20
-------
Table 12.4-2. Estimated Monetary Value in Reductions in Incidence of Health and Welfare
Effects (in millions of 2003$)a'b
2020 2030
PM-Related Health Effect Estimated Value of Reductions
Premature mortality0'46
Adult, age 30+ and Infant, < 1 year
3% discount rate $3,100 $5,800
7% discount rate $2,800 $5,200
Chronic bronchitis (adults, 26 and over) $150 $260
Non-fatal acute myocardial infarctions
3% discount rate $79 $150
7% discount rate $76 $140
Hospital admissions for respiratory causes $4.7 $10
Hospital admissions for cardiovascular causes $5.0 $9.1
Emergency room visits for asthma $0.11 $0.20
Acute bronchitis (children, age 8-12) $0.32 $0.56
Lower respiratory symptoms (children, 7-14) $0.16 $0.29
Upper respiratory symptoms (asthma, 9-11) $0.20 $0.35
Asthma exacerbations $0.56 $1.0
Work loss days $9.1 $14
Minor restricted-activity days (MRADs) $21 $35
Monetized Totalf
Base Estimate:
3% discount rate $3,300+B $6,300+B
7% discount rate $3,000+B $5,700+B
Monetary benefits are rounded to two significant digits for ease of presentation and computation. PM benefits are
nationwide.
b Monetary benefits adjusted to account for growth in real GDP per capita between 1990 and the analysis year (2020 or
2030)
0 PM-related mortality benefits estimated using an assumed PM threshold at background levels (3 ug/m3). There
is uncertainty about which threshold to use and this may impact the magnitude of the total benefits estimate. For
a more detailed discussion of this issue, please refer to Section 12.6.2.2 of the RIA.
d Valuation assumes discounting over the SAB recommended 20-year segmented lag structure described earlier. Results
reflect the use of 3 percent and 7 percent discount rates consistent with EPA and OMB guidelines for preparing
economic analyses (EPA, 2000; OMB, 2003).43'44
Adult premature mortality estimates based upon the ACS cohort study (Pope et al., 2002). Infant premature
mortality based upon Woodruff et al 1997.
f B represents the monetary value of health and welfare benefits and disbenefits not monetized. A detailed listing is
provided in Table 12.1-2.
In addition to omitted benefits categories such as air toxics, ozone, and various welfare
effects, not all known direct PM-related health and welfare effects could be quantified or
monetized. Furthermore, we did not quantify reductions in secondary PM2.5 and the associated
health and welfare effects. The monetized value of all of these unquantified effects is represented
by adding an unknown "B" to the aggregate total. The estimate of total monetized health
benefits of the final MS AT control package is thus equal to the subset of monetized PM-related
health benefits plus B, the sum of the nonmonetized health and welfare benefits.
12-21
-------
Total monetized benefits are dominated by benefits of mortality risk reductions. The
primary estimate projects that the final cold temperature vehicle standards will result in 480
avoided premature deaths annually in 2020 and 880 avoided premature deaths annually in 2030.
The increase in annual benefits from 2020 to 2030 reflects additional emission reductions from
the final cold temperature vehicle standards, as well as increases in total population and the
average age (and thus baseline mortality risk) of the population.
Our estimate of total monetized benefits in 2020 for the final cold temperature vehicle
standards is $3.3 billion using a three percent discount rate and $3.0 billion using a seven percent
discount rate. In 2030, the monetized benefits are estimated at $6.3 billion using a three percent
discount rate and $5.7 billion using a seven percent discount rate. The monetized benefit
associated with reductions in the risk of premature mortality, which accounts for $3.1 billion in
2020 and $5.8 billion in 2030 (assuming a three percent discount rate), is over 90 percent of total
monetized health benefits. The next largest benefit is for reductions in chronic illness (CB and
nonfatal heart attacks), although this value is more than an order of magnitude lower than for
premature mortality. Hospital admissions for respiratory and cardiovascular causes, minor
restricted activity days, and work loss days account for the majority of the remaining benefits.
The remaining categories each account for a small percentage of total benefit; however, they
represent a large number of avoided incidences affecting many individuals. A comparison of the
incidence table to the monetary benefits table reveals that there is not always a close
correspondence between the number of incidences avoided for a given endpoint and the
monetary value associated with that endpoint. For example, there are over 100 times more work
loss days than premature mortalities, yet work loss days account for only a very small fraction of
total monetized benefits. This reflects the fact that many of the less severe health effects, while
more common, are valued at a lower level than the more severe health effects. Also, some
effects, such as hospital admissions, are valued using a proxy measure of willingness-to-pay
(e.g., cost-of-illness).J As such, the true value of these effects may be higher than that reported
in Table 12.4-2.
12.5 Unquantified Health and Welfare Effects
In considering the monetized benefits estimates, the reader should remain aware of the
many limitations of conducting the analyses mentioned throughout this RIA. One significant
limitation of both the health and welfare benefits analyses is the inability to quantify many of the
effects listed in Table 12.1-2. For many health and welfare effects, such as changes in health
effects due to reductions in air toxics exposure, changes in ecosystem functions and PM-related
materials damage, reliable impact functions and/or valuation functions are not currently
available. In general, if it were possible to monetize these benefit categories, the benefits
estimates presented in this analysis would increase, although the magnitude of such an increase
is highly uncertain.
Other welfare effects that EPA has monetized in past RIAs, such as recreational
1 See Table 12.3-2 for a description of how each particular endpoint is valued.
12-22
-------
visibility, are omitted from the current analysis. Due to time and resource constraints, we did not
run the full-scale PM air quality modeling needed to estimate this benefit category. Instead, we
relied on the PM scaling benefits transfer approach that provides analytical efficiency but
sacrifices the full range of outputs typically generated when models such as the Community
Multiscale Air Quality (CMAQ) model or the Regional Modeling System for Aerosols and
Deposition (REMSAD) are run.
Unquantified benefits are qualitatively discussed in the following health and welfare
effects sections. In addition to unquantified benefits, there may also be environmental costs
(disbenefits) that we are unable to quantify, which we qualitatively discuss as well. The net
effect of excluding benefit and disbenefit categories from the estimate of total benefits depends
on the relative magnitude of the effects. Although we are not currently able to estimate the
magnitude of these unquantified and unmonetized benefits, specific categories merit further
discussion. EPA believes, however, the unquantified benefits associated with health and non-
health benefit categories are likely significant and that their omission lends a downward bias to
the monetized benefits presented in this analysis.
12.5.1 Human Health Impact Assessment
In addition to the PM2.s health effects discussed above, there is emerging evidence that
human exposure to PM may be associated a number of health effects not quantified in this
analysis (see Table 12.1-2). An improvement in ambient PM2 5 concentrations may reduce the
number of incidences within each of these unquantified effect categories that the U.S. population
would experience. Although these health effects are believed to be PM-induced, effect estimates
are not available for quantifying the benefits associated with reducing these effects.
Furthermore, the health effects associated with reductions in air toxics are not quantified in this
analysis. The health endpoints associated with individual air toxic reductions achieved by the
final standards are discussed in Chapter 1 of the RIA.
Other standards included in this final rulemaking, such as the PFC standards, will also
reduce the national emissions inventory of precursors to ozone, such as VOCs. Exposure to
ozone has been linked to a variety of respiratory effects including hospital admissions,
emergency room visits, minor restricted activity days, worker productivity and illnesses resulting
in school absences. Emerging evidence has also shown that human exposure to ozone may be
associated with a number of other health effects not quantified in this analysis (see Table 12.1-2).
Ozone can also adversely affect the agricultural and forestry sectors by decreasing yields of
crops and forests. Although ozone benefits are typically quantified in regulatory impact
analyses, we do not evaluate them for this analysis because of the magnitude of, and uncertainty
associated with, the ambient ozone modeling data. As discussed earlier in this chapter (and in
Chapter 3), the ozone modeling conducted for the PFC standards results in a net reduction, when
population-weighted, in the ozone design value metric measured within the modeled domain (37
Eastern states and the District of Columbia). The net improvement, however, is very small. For
the most part, quantifiable ozone benefits will not contribute significantly to the monetized
benefits; thus, their omission will not materially affect the conclusions of the benefits analysis.
12-23
-------
12.5.2 Welfare Impact Assessment
For many welfare effects, such as changes in ecosystem functions and PM-related
materials damage, reliable impact functions and/or valuation functions are not currently
available. In general, if it were possible to monetize these benefit categories, the benefits
estimates presented in this analysis would increase, although the magnitude of such an increase
is highly uncertain.
12.5.2.1 Visibility Benefits
Changes in the level of ambient PM2.5 caused by the final standards will change the level
of visibility in much of the United States. Visibility directly affects people's enjoyment of a
variety of daily activities. Individuals value visibility both in the places they live and work, in
the places they travel to for recreational purposes, and at sites of unique public value, such as the
Great Smoky Mountains National Park. Though not quantified in this analysis, the value of
improvements in visibility monetized for regulatory analyses such as the final CAIR are
significant. We refer the reader to that analysis for a complete description of the methods used to
value visibility.47
12.5.2.2 Agricultural and Forestry Benefits
The Ozone Criteria Document notes that "ozone affects vegetation throughout the United
States, impairing crops, native vegetation, and ecosystems more than any other air pollutant"
(EPA, 1996, page 5-11).48 Though we do not quantify the potential improvements in ambient
ozone concentrations associated with the final standards, it is possible that yields will improve in
areas of agricultural or forestry production impacted by the standards. The net ozone
improvement, however, is very small. We expect that the omission of agricultural impacts will
not materially affect the conclusions of the benefits analysis.
With that said, however, well-developed techniques exist to provide monetary estimates
of these benefits to agricultural producers and to consumers. These techniques use models of
planting decisions, yield response functions, and agricultural products' supply and demand. The
resulting welfare measures are based on predicted changes in market prices and production costs.
Models also exist to measure benefits to silvicultural producers and consumers. However, these
models have not been adapted for use in analyzing ozone-related forest impacts. Because of
resource limitations, we are unable to provide agricultural or forestry benefits estimates for the
final standards.
12.5.2.2.1 Agricultural Benefits
Laboratory and field experiments have shown reductions in yields for agronomic crops
exposed to ozone, including vegetables (e.g., lettuce) and field crops (e.g., cotton and wheat).
The most extensive field experiments, conducted under the National Crop Loss Assessment
Network (NCLAN), examined 15 species and numerous cultivars. The NCLAN results show
12-24
-------
that "several economically important crop species are sensitive to ozone levels typical of those
found in the United States."54 In addition, economic studies have shown a relationship between
observed ozone levels and crop yields.49
12.5.2.2.2 Forestry Benefits
Ozone also has been shown conclusively to cause discernible injury to forest trees (EPA,
1996; Fox and Mickler, 1996).54'50 In our previous analysis of the Heavy-Duty Engine/Diesel
Fuel rule, we were able to quantify the effects of changes in ozone concentrations on tree growth
for a limited set of species. Because the net change in measured ozone associated with the final
standards was so small, we were not able to quantify such impacts for this analysis.
12.5.2.3 Benefits from Reductions in Materials Damage
The final standards that we modeled are expected to produce economic benefits in the
form of reduced materials damage. There are two important categories of these benefits.
Household soiling refers to the accumulation of dirt, dust, and ash on exposed surfaces. PM also
has corrosive effects on commercial/industrial buildings and structures of cultural and historical
significance. The effects on historic buildings and outdoor works of art are of particular concern
because of the uniqueness and irreplaceability of many of these objects.
Previous EPA benefits analyses have been able to provide quantitative estimates of
household soiling damage. Consistent with SAB advice, we determined that the existing data
(based on consumer expenditures from the early 1970s) are too out of date to provide a reliable
estimate of current household soiling damages (EPA-SAB-COUNCIL-ADV-98-003, 1998).51
EPA is unable to estimate any benefits to commercial and industrial entities from reduced
materials damage. Nor is EPA able to estimate the benefits of reductions in PM-related damage
to historic buildings and outdoor works of art. Existing studies of damage to this latter category
in Sweden (Grosclaude and Soguel, 1994)52 indicate that these benefits could be an order of
magnitude larger than household soiling benefits.
12.5.3 UVb Exposure
In contrast to the unquantified benefits of the final standards discussed above, it is also
possible that this rule will result in disbenefits in some areas of the United States. The effects of
ozone and PM on radiative transfer in the atmosphere can lead to effects of uncertain magnitude
and direction on the penetration of ultraviolet light and climate. Ground level ozone makes up a
small percentage of total atmospheric ozone (including the stratospheric layer) that attenuates
penetration of ultraviolet - b (UVb) radiation to the ground. EPA's past evaluation of the
information indicates that potential disbenefits would be small, variable, and with too many
uncertainties to attempt quantification of relatively small changes in average ozone levels over
the course of a year.53 EPA's most recent provisional assessment of the currently available
information indicates that potential but unquantifiable benefits may also arise from ozone-related
attenuation of UVb radiation.54 EPA believes that we are unable to quantify any net climate-
12-25
-------
related disbenefit or benefit associated with the combined ozone and PM reductions in this rule.
12.6 Methods for Describing Uncertainty
In any complex analysis using estimated parameters and inputs from numerous models,
there are likely to be many sources of uncertainty. This analysis is no exception. As outlined
both in this and preceding chapters, many inputs were used to derive the benefits estimate,
including emission inventories, air quality models (with their associated parameters and inputs),
epidemiological health effect estimates, estimates of values (both from WTP and COI studies),
population estimates, income estimates, and estimates of the future state of the world (i.e.,
regulations, technology, and human behavior). Each of these inputs may be uncertain and,
depending on its role in the benefits analysis, may have a disproportionately large impact on
estimates of total benefits. For example, emissions estimates are used in the first stage of the
analysis. As such, any uncertainty in emissions estimates will be propagated through the entire
analysis. Some of the key uncertainties in the quantified benefits analysis are presented in Table
12.6-1.
Table 12.6-1. Primary Sources of Uncertainty in the Quantified Benefits Analysis
1. Uncertainties Associated with Impact Functions
• The value of the PM effect estimate in each impact function.
• Application of a single impact function to pollutant changes and populations in all locations.
• Similarity of future-year impact functions to current impact functions.
• Correct functional form of each impact function.
• Extrapolation of effect estimates beyond the range of PM concentrations observed in the source
epidemiological study.
• Application of some impact functions only to those subpopulations matching the original study
population.
2. Uncertainties Associated with PM Concentrations
• Responsiveness of the models to changes in precursor emissions resulting from the control policy.
• Projections of future levels of precursor emissions, especially organic carbonaceous particle emissions.
• Model chemistry for the formation of ambient nitrate concentrations.
• Lack of speciation monitors in some areas requires extrapolation of observed speciation data.
• CMAQ model performance in the Western U.S., especially California indicates significant
underprediction of PM2 5.
3. Uncertainties Associated with PM Mortality Risk
• Differential toxicity of specific component species within the complex mixture of PM has not been
determined.
• The extent to which adverse health effects are associated with low-level exposures that occur many times
in the year versus peak exposures.
• The extent to which effects reported in the long-term exposure studies are associated with historically
higher levels of PM rather than the levels occurring during the period of study.
• Reliability of the limited ambient PM2 5 monitoring data in reflecting actual PM2 5 exposures.
5. Uncertainties Associated with Possible Lagged Effects
• The portion of the PM-related long-term exposure mortality effects associated with changes in annual PM
levels that would occur in a single year is uncertain as well as the portion that might occur in subsequent
years.
12-26
-------
6. Uncertainties Associated with Baseline Incidence Rates
• Some baseline incidence rates are not location specific (e.g., those taken from studies) and therefore may
not accurately represent the actual location-specific rates.
• Current baseline incidence rates may not approximate well baseline incidence rates in 2020 and 2030.
• Projected population and demographics may not represent well future-year population and demographics.
7. Uncertainties Associated with Economic Valuation
• Unit dollar values associated with health and welfare endpoints are only estimates of mean WTP and
therefore have uncertainty surrounding them.
• Mean WTP (in constant dollars) for each type of risk reduction may differ from current estimates because
of differences in income or other factors.
8. Uncertainties Associated with Aggregation of Monetized Benefits
• Health and welfare benefits estimates are limited to the available impact functions. Thus, unqualified or
unmonetized benefits are not included.
As part of EPA's approach to characterizing uncertainties in the benefits assessment, we
generate a probabilistic estimate of statistical uncertainty based on standard errors reported in the
underlying studies used in the benefits modeling framework, with particular emphasis on the
health impact functions. Using a Monte Carlo procedure, the distribution of each health endpoint
and its unit dollar value is characterized by the reported mean and standard error derived from
the epidemiology and valuation literature. Details on the distributions used to value individual
health endpoints are provided in Section 12.6.1, as well as in the CAIR RIA (Appendix B; EPA,
2005).55 It should be noted that the Monte Carlo-generated distributions of benefits reflect only
some of the uncertainties in the input parameters (described in Table 12.6-1). Uncertainties
associated with emissions, air quality modeling, populations, and baseline health effect incidence
rates are not represented in the distributions of benefits of attaining alternative standards. Issues
such as correlation between input parameters and the identification of reasonable upper and
lower bounds for input distributions characterizing uncertainty in additional model elements will
be addressed in future versions of the uncertainty framework.
In benefit analyses of air pollution regulations conducted to date, the estimated impact of
reductions in premature mortality has accounted for 85% to 95% of total benefits. Therefore, in
characterizing the uncertainty related to the estimates of total benefits it is particularly important
to attempt to characterize the uncertainties associated with this endpoint. As such, we
specifically discuss the uncertainty related to PM-related premature mortality in Section 12.6.2.
12.6.1 Analysis of Statistical Uncertainty
For the final standards, we did not attempt to assign probabilities to all of the uncertain
parameters in the model because of a lack of resources and reliable methods. At this time, we
simply generate estimates of the distributions of dollar benefits for PM health effects and for
total dollar benefits. For all quantified PM endpoints, we scaled the likelihood distributions of
the benefit estimates from the CAND uncertainty analysis,K based on the same benefits transfer
K U.S. Environmental Protection Agency. May 2004. Final Regulatory Analysis: Control of Emissions from
Nonroad Diesel Engines. Prepared by: Office of Air and Radiation. Available at http://www.epa.gov/nonroad-
diesel/2004fr.htm#documents. Accessed December 15, 2005.
12-27
-------
approach we used to estimate the benefits of the standards presented in Section 12.2. The CAND
likelihood distributions were based solely on the statistical uncertainty surrounding the estimated
C-R functions and the assumed distributions around the unit values. We use the benefits transfer
approach to scale those distributions to reflect the predicted direct PM emission reductions of the
final cold temperature standards. Though the scaling approach adds another element of
uncertainty that we cannot characterize in the distributions, we believe the scaled uncertainty is a
reasonable approximation of the statistical uncertainty based on standard errors reported in the
underlying epidemiological and valuation studies.
Our scaled estimates of the likelihood distributions for health-related PM benefits should
be viewed as incomplete because of the wide range of sources of uncertainty that we have not
incorporated. The 5th and 95th percentile points of our scaled estimate are based on statistical
error, and cross-study variability provides some insight into how uncertain our estimate is with
regard to those sources of uncertainty. However, it does not capture other sources of uncertainty
regarding the benefits transfer scaling approach or the inputs to the CAND modeling upon which
the scaling is based, including emissions, air quality, baseline population incidence, and
projected exposures. It also does not account for aspects of the health science not captured in the
studies, such as the likelihood that PM is causally related to premature mortality and other
serious health effects. Thus, a likelihood description based on the standard error would provide a
misleading picture about the overall uncertainty in the estimates.
Both the uncertainty about incidence changesL and uncertainty about unit dollar values
can be characterized by distributions. Each "likelihood distribution" characterizes our beliefs
about what the true value of an unknown variable (e.g., the true change in incidence of a given
health effect in relation to PM exposure) is likely to be, based on the available information from
relevant studies.M Unlike a sampling distribution (which describes the possible values that an
estimator of an unknown variable might take on), this likelihood distribution describes our
beliefs about what values the unknown variable itself might be. Such likelihood distributions
can be constructed for each underlying unknown variable (such as a particular pollutant
coefficient for a particular location) or for a function of several underlying unknown variables
(such as the total dollar benefit of a regulation). In either case, a likelihood distribution is a
characterization of our beliefs about what the unknown variable (or the function of unknown
variables) is likely to be, based on all the available relevant information. A likelihood
description based on such distributions is typically expressed as the interval from the 5th
percentile point of the likelihood distribution to the 95th percentile point. If all uncertainty had
been included, this range would be the "credible range" within which we believe the true value is
likely to lie with 90 percent probability.
L Because this is a national analysis in which, for each endpoint, a single C-R function is applied everywhere, there
are two sources of uncertainty about incidence: statistical uncertainty (due to sampling error) about the true value of
the pollutant coefficient in the location where the C-R function was estimated and uncertainty about how well any
given pollutant coefficient approximates (3*.
M Although such a "likelihood distribution" is not formally a Bayesian posterior distribution, it is very similar in
concept and function (see, for example, the discussion of the Bayesian approach in Kennedy, 1990. A Guide to
Econometrics. 2nd ed. MIT Press: Cambridge, MA., pp. 168-172).
12-28
-------
12.6.1.1 Monte Carlo Approach
The uncertainty about the total dollar benefit associated with any single endpoint
combines the uncertainties from these two sources (the C-R relationship and the valuation) and is
estimated with a Monte Carlo method. In each iteration of the Monte Carlo procedure, a value is
randomly drawn from the incidence distribution, another value is randomly drawn from the unit
dollar value distribution; the total dollar benefit for that iteration is the product of the two.N
When this is repeated for many (e.g., thousands of) iterations, the distribution of total dollar
benefits associated with the endpoint is generated.
Using this Monte Carlo procedure, a distribution of dollar benefits can be generated for
each endpoint. As the number of Monte Carlo draws gets larger and larger, the Monte Carlo-
generated distribution becomes a better and better approximation of a joint likelihood
distribution (for the considered parameters) making up the total monetary benefits for the
endpoint.
After endpoint-specific distributions are generated, the same Monte Carlo procedure can
then be used to combine the dollar benefits from different (nonoverlapping) endpoints to
generate a distribution of total dollar benefits.
The estimate of total benefits may be thought of as the end result of a sequential process
in which, at each step, the estimate of benefits from an additional source is added. Each time an
estimate of dollar benefits from a new source (e.g., a new health endpoint) is added to the
previous estimate of total dollar benefits, the estimated total dollar benefits increases. However,
our bounding or likelihood description of where the true total value lies also increases as we add
more sources.
As an example, consider the benefits from reductions in PM-related hospital admissions
for cardiovascular disease. Because the actual dollar value is unknown, it may be described
using a variable, with a distribution describing the possible values it might have. If this variable
is denoted as XI, then the mean of the distribution, E(X1) and the variance of XI, denoted
Var(Xl), and the 5th and 95th percentile points of the distribution (related to Var(Xl)), are ways
to describe the likelihood for the true but unknown value for the benefits reduction.
Now suppose the benefits from reductions in PM-related hospital admissions for
respiratory diseases are added. Like the benefits from reductions in PM-related hospital
admissions for cardiovascular disease, the likelihood distribution for where we expect the true
value to be may be considered a variable, with a distribution. Denoting this variable as X2, the
benefits from reductions in the incidence of both types of hospital admissions is XI + X2. This
variable has a distribution with mean E(X1 + X2) = E(X1) + E(X2), and a variance of Var(Xl +
N This method assumes that the incidence change and the unit dollar value for an endpoint are stochastically
independent.
12-29
-------
X2) = Var(Xl) + Var(X2) + 2Cov(Xl,X2); if XI and X2 are stochastically independent, then it
has a variance of Var(Xl + X2) = Var(Xl) + Var(X2), and the covariance term is zero.
The benefits from reductions in all nonoverlapping PM-related health and welfare
endpoints are (Xm+1, ..., Xn) is X = XI + ... + Xn. The mean of the distribution of total
benefits, X, is
E(X) = E(X1) + E(X2) + ... + E(Xn)
and the variance of the distribution of total benefits—assuming that the components are
stochastically independent of each other (i.e., no covariance between variables), is
Var(X) = Var(Xl) + Var(X2) + ... + Var(Xn)
If all the means are positive, then each additional source of benefits increases the point estimate
(mean) of total benefits. However, with the addition of each new source of benefits, the variance
of the estimate of total benefits also increases. That is,
E(X1) < E(X1 + X2) < E(X1 + X2 + X3) < ... < E(X1 + ... + Xn) = E(X)
Var(Xl) < Var(Xl + X2) < Var(Xl + X2 + X3) < ... < Var(Xl + ... + Xn) = Var(X)
That is, the addition of each new source of benefits results in a larger mean estimate of total
benefits (as more and more sources of benefits are included in the total) about which there is less
certainty. This phenomenon occurs whenever estimates of benefits are added.
Calculated with a Monte Carlo procedure, the distribution of X is composed of random
draws from the components of X. In the first draw, a value is drawn from each of the
distributions, XI, X2, through Xn; these values are summed; and the procedure is repeated again,
with the number of repetitions set at a high enough value (e.g., 5,000) to reasonably trace out the
distribution of X. The 5th percentile point of the distribution of X will be composed of points
pulled from all points along the distributions of the individual components and not simply from
the 5th percentile. Although the sum of the 5th percentiles of the components would be
represented in the distribution of X generated by the Monte Carlo, it is likely that this value
would occur at a significantly lower percentile. For a similar reason, the 95th percentile of X
will be less than the sum of the 95th percentiles of the components, and instead the 95th
percentile of X will be composed of component values that are significantly lower than the 95th
percentiles.
The physical effects estimated in this analysis are assumed to occur independently. It is
possible that, for any given pollution level, there is some correlation between the occurrence of
physical effects, due to say avoidance behavior or common causal pathways and treatments (e.g.,
stroke, some kidney disease, and heart attack are related to treatable blood pressure). Estimating
accurately any such correlation, however, is beyond the scope of this analysis, and instead it is
simply assumed that the physical effects occur independently.
12-30
-------
12.6.1.2 Monte Carlo Results
Based on the Monte Carlo techniques and benefits transfer methods described above, we
scaled the CAND likelihood distributions for the dollar value of total PM health-related benefits
for the final standards. For this analysis, the likelihood descriptions for the true value of each of
the health endpoint incidence estimates, including premature mortality, were based on classical
statistical uncertainty measures. The measures include the mean and standard deviation of the C-
R relationships in the epidemiological literature, and assumptions of particular likelihood
distribution shapes for the valuation of each health endpoint value based on reported values in
the economic literature. The distributions for the value used to represent incidence of a health
effect in the total benefits valuation represent both the simple statistical uncertainty surrounding
individual effect estimates and, for those health endpoints with multiple effects from different
epidemiology studies, interstudy variability. Distributions for unit dollar values are summarized
in Chapter 12, Table 12.3-2.
Results of the scaled Monte Carlo simulations are presented in Table 12.6-2. The table
provides the scaled means of the distributions and the estimated 5th and 95th percentiles of the
distributions. The contribution of mortality to the mean benefits and to both the 5th and 95th
percentiles of total benefits is substantial, with mortality accounting for over 90 percent of the
mean estimate, and even the 5th percentile of mortality benefits dominating close to the 95th
percentile of all other benefit categories. Thus, the choice of value and the shape for likelihood
distribution for VSL should be examined closely and is key information to provide to decision
makers for any decision involving this variable. The 95th percentile of total benefits is
approximately twice the mean, while the 5th percentile is approximately one-fourth of the mean.
The overall range from 5th to 95th represents about one order of magnitude.
12-31
-------
Table 12.6-2. Distribution of Value of Annual PM-Related Human Health Benefits in 2030
for the Final Mobile Source Air Toxics Rule: Cold Temperature Controls a
Endpoint
Monetary Benefits'1' c (Millions 2003$, Adjusted for Income
Growth)
5th Percentile Mean 95th Percentile
Premature mortality0, Long-term exposure
Adults, 30+ yrs and Infants,
-------
As part of an overall program to improve the Agency's characterization of uncertainties in
health benefits analyses, we attempt to address uncertainties associated with the PM2.5 mortality
health impact function relationship and valuation. Use of the ACS cohort (Pope et al., 2002)
mortality function to support this analysis does not address uncertainty associated with: (a)
potential of the study to incompletely capture short-term exposure-related mortality effects, (b)
potential mis-match between study and analysis populations which introduces various forms of
bias into the results, (c) failure to identify all key confounders and effects modifiers, which could
result in incorrect effects estimates relating mortality to PM2.5 exposure, and (d) model
uncertainty. EPA is researching methods to characterize all elements of uncertainty in the dose-
response function for mortality.
As is discussed in detail in the final PM NAAQS RIA, EPA uses three methods to
quantify uncertainties in the mortality function, including: the statistical uncertainty derived from
the standard errors reported in the ACS cohort study, the presentation of additional estimates of
mortality based upon the peer-reviewed literature, and the use of results of an expert elicitation
conducted to explore a more thorough characterization of uncertainties in the mortality estimate.
Because this analysis utilizes the PM scaling benefits transfer approach to estimate mortality
incidence for the final cold temperature vehicle standard, we cannot quantify the PM mortality
uncertainty to the same extent as was done for the CAIR or PM NAAQS analyses. However, in
a similar fashion to the analysis conducted for the Clean Air Visibility Rule (CAVR),56 we can
scale the results of the CAND mortality uncertainty analysis to the PM precursor emission
changes modeled for the final cold temperature standard.
12.6.2.1 Uncertainty Associated with the Concentration-Response Function
In the benefit analysis of the CAND 2030 emission control standards, the statistical
uncertainty represented by the standard error of the American Cancer Society cohort study (Pope
et al, 2002) was one and one-half times the mean benefit estimate at the 95th percentile and less
than one-half of the mean at the 5th percentile. The CAND analysis also derived mortality from
the reanalysis of the Harvard Six-Cities study (Krewski et al., 2000).57 At the time of the CAND
analysis, EPA's Science Advisory Board provided guidance stating, "The Six-Cities estimates
may be used in a sensitivity analysis to demonstrate that with different but also plausible
selection criteria for C-R functions, benefits may be considerably larger than suggested by the
ACS study." (EPA-SAB-COUNCIL-ADV-04-002).58 In the CAND analysis, the Harvard Six-
Cities mean benefits estimate was over twice the size of the mean estimate of mortality benefits
derived from the ACS study.
Recently, a new peer-reviewed extension of the Six-Cities study has been published
(Laden et al., 2006).59 This follow-up to the Harvard Six-Cities study both confirmed the effect
size from the first analysis and provided additional evidence that reductions in PM2.5 are likely
associations with reductions in the risk of premature death. This additional evidence stems from
the observed reductions in PM2.5 in each city during the extended follow-up period. Laden et al.
(2006) found that mortality rates consistently went down at a rate proportionate to the observed
reductions in PM2.5. In the recently finalized PM NAAQS RIA, results from this study were
12-33
-------
presented as an additional estimate of premature mortality benefits along with the benefits
derived from the ACS study. The mean benefits estimate derived from the Six-Cities study was
more than twice the size of the mean estimate of mortality benefits derived from the ACS study.
Because this study was not available during the CAND analysis, from which the benefits of
today's final standards are scaled, we are unable to provide an estimate of mortality benefits
based on the Six-Cities study for this final analysis. However, based on the relationship between
the Six-Cities study and the ACS cohort study observed in the final PM NAAQS RIA, we can
surmise that the mean estimate of PM-related mortality associated with the final cold
temperature standards could be approximately twice as large. For a full discussion of the
epidemiological basis of EPA's premature mortality estimates, we refer the reader to Chapter 5.1
of the final PM NAAQS RIA.
EPA recently completed a full-scale expert elicitation that incorporated peer-review
comments on the pilot application used in CAND, and that provides a more robust
characterization of the uncertainty in the premature mortality function. This expert elicitation
was designed to evaluate uncertainty in the underlying causal relationship, the form of the
mortality impact function (e.g., threshold versus linear models) and the fit of a specific model to
the data (e.g., confidence bounds for specific percentiles of the mortality effect estimates).
Additional issues, such as the ability of long-term cohort studies to capture premature mortality
resulting from short-term peak PM exposures, were also addressed in the expert elicitation. The
recently published RIA supporting the Particulate Matter National Ambient Air Quality
Standards (PM NAAQS) used the results of this expert elicitation to quantitatively characterize
uncertainty.
Due to the analytical constraints associated with the PM benefits scaling approach, we are
unable to assess the premature mortality health impacts derived from the formally elicited expert
judgments. Compared to the final PM NAAQS estimate of mean premature mortality derived
from the ACS cohort study, however, expert-based mortality incidence ranged from
approximately 50 percent of the mean ACS estimate to approximately five times the size of the
mean ACS estimate. In total, PM-related premature mortality derived from eleven of the experts
was greater than the ACS estimate, while one expert-based estimate fell below the ACS result.
12.6.2.2 PM2.5-Mortality Cutpoint/Threshold Analysis
Another source of uncertainty that has received recent attention from several scientific
review panels is the shape of the concentration-response function for PM-related mortality, and
specifically whether there exists a threshold below which there would be no benefit to further
reductions in PM2.5. The consistent advice from EPA's SAB° has been to model premature
0 The advice from the 2004 SAB-HES (EPA-SAB-COUNCIL-ADV-04-002)69 is characterized by the following:
"For the studies of long-term exposure, the HES notes that Krewski et al. (2000) have conducted the most careful
work on this issue. They report that the associations between PM2 5 and both all-cause and cardiopulmonary
mortality were near linear within the relevant ranges, with no apparent threshold. Graphical analyses of these
studies (Dockery et al., 1993, Figure 3, and Krewski et al., 2000, page 162) also suggest a continuum of effects
12-34
-------
mortality associated with PM exposure as a nonthreshold effect, that is, with harmful effects to
exposed populations regardless of the absolute level of ambient PM concentrations. However,
EPA's most recent PM2.5 Criteria Document concludes that "the available evidence does not
either support or refute the existence of thresholds for the effects of PM on mortality across the
range of concentrations in the studies."60 Some researchers have hypothesized the presence of a
threshold relationship. That is, the hypothesized relationship includes the possibility that there
exists a PM concentration level below which further reductions no longer yield premature
mortality reduction benefits.
To consider the impact of a threshold in the response function for the chronic mortality
endpoint, the final PM NAAQS RIA61 constructed a sensitivity analysis by assigning different
cutpoints below which changes in PM2.5 are assumed to have no impact on premature mortality.
In applying the cutpoints, the PM NAAQS analysis adjusted the mortality function slopes
accordingly.1" Five cutpoints (including the base case assumption) were included in the
sensitivity analysis: (a) 14 |ig/m3 (assumes no impacts below a level being considered at the time
for the annual PM2.5 NAAQS), (b) 12 |ig/m3 (c) 10 |ig/m3 (reflects comments from CASAC,
2005), 62 (d) 7.5 |ig/m3 (reflects recommendations from SAB-HES to consider estimating
mortality benefits down to the lowest exposure levels considered in the ACS cohort study (Pope
et al., 2002) used as the basis for modeling chronic mortality) 63 and (e) background or 3 |ig/m3
(reflects NRC recommendation to consider effects all the way to background).64 The results of
the sensitivity analysis displayed the change in avoided mortality cases and associated monetary
benefits associated with the alternative cutpoints (see the final PM NAAQS RIA, Chapter 5.1
and Tables 5-28 to 5-31).
A sensitivity analysis such as this can be difficult to interpret, because when a threshold
above the lowest observed level of PM2.s in the underlying ACS cohort study (Pope et al., 2002)
is assumed, the slope of the concentration-response function above that level must be adjusted
upwards to account for the assumed threshold.*2 Depending on the amount of slope adjustment
and the proportion of the population exposed above the assumed threshold, the estimated
mortality impact can either be lower (if most of the exposures occur below the threshold) or
higher (if most of the exposures occur above the threshold). To demonstrate this, we present an
example from the proposed PM NAAQS RIA. In its examination of the benefits of attaining
alternative PM NAAQS in Chicago,R the analysis found that, because annual mean levels are
generally higher in Chicago, there was a two-part pattern to the relationship between assumed
threshold and mortality impacts. As the threshold increased from background to 7.5 ug/m3, the
mortality impact fell (because there is no slope adjustment). However, at an assumed threshold
of 10 ug/m3, estimated mortality impacts actually increased, because the populations exposed
down to lower levels. Therefore, it is reasonable for EPA to assume a no threshold model down to, at least, the low
end of the concentrations reported in the studies."
p Note that the PM NAAQS analysis only adjusted the mortality slopes for the 10 ug/m3, 12 ug/m3 and 14 ug/m3
cutpoints since the 7.5 ug/m3 and background cutpoints were at or below the lowest measured exposure levels
reported in the Pope et al. (2002) study for the combined exposure dataset.
Q See NAS (2002)71 and CASAC (2005)68 for discussions of this issue.
R See the proposed PM NAAQS RIA (2005),67 Appendix A, pp. A63-A64.
12-35
-------
above 10 ug/m were assumed to have a larger response to particulate matter reductions (due to
the increased slope above the assumed threshold). And finally, mortality impacts again fell to
zero if a 15 ug/m3 threshold was assumed, because these impacts were measured incremental to
attainment of the current standard.
We are unable to do this type of sensitivity analysis for the final MSAT rule because of
the analytical limitations of the PM benefits scaling procedure. When EPA conducted the
CAND analysis (from which the primary estimates of benefits for the final cold temperature
vehicle standards are based), there were no PM mortality concentration-response functions with
the slope adjusted upwards to account for an assumed threshold. Instead, our primary PM
benefits estimate for the final cold temperature vehicle standards reflects a background threshold
assumption of 3 ug/m3. We present in Table 12.6-3 the results of our scaled PM-related
mortality benefits in the context of its relationship to other cutpoints.
Table 12.6-3. PM-Related Mortality Benefits of the Final Cold Temperature Vehicle
Standards: Cutpoint Sensitivity Analysis3
Certainty that Benefits are
At Least Specified Value
More Certain that Benefits
Are at Least as Large
V
Less Certain that Benefits
Are at Least as Large
Level of Assumed
Threshold
14 ug/m3 c
12 ug/m3
10 ug/m3 d
7.5 ug/m3 e
3 ug/m3 f
Discount
Rate
3%
7%
3%
7%
3%
7%
3%
7%
3%
7%
PM Mortality Benefits (Billion 2003$)
2020 2030
N/Ab
N/A
N/A
N/A
$o o (t'/r i
5.5 3>O.J
$3.0 $5.7
a Note that this table only presents the effects of a outpoint on PM-related mortality incidence and valuation
estimates.
b Not Available. We are unable to provide cutpoint analysis results for the final MSAT rule because of the
analytical limitations of the PM benefits scaling procedure.
0 EPA intends to analyze a cutpoint between 12 ug/m3and 15 ug/m3 for the final RIA.
d CASAC (2005)68
e SAB-HES (2004)69
fNAS(2002)71
12.7 Health-Based Cost Effectiveness Analysis
Health-based cost-effectiveness analysis (CEA) and cost-utility analysis (CUA) have
been used to analyze numerous health interventions but have not been widely adopted as tools to
analyze environmental policies. The Office of Management and Budget (OMB) issued Circular
12-36
-------
A-4 guidance on regulatory analyses, requiring Federal agencies to "prepare a CEA for all major
rulemakings for which the primary benefits are improved public health and safety to the extent
that a valid effectiveness measure can be developed to represent expected health and safety
outcomes." Environmental quality improvements may have multiple health and ecological
benefits, making application of CEA more difficult and less straightforward. For the CAIR
analysis, the first to incorporate an analysis of this kind, CEA provided a useful framework for
evaluation: nonhealth benefits were substantial, but the majority of quantified benefits came
from health effects. EPA included in the CAIR RIA a preliminary and experimental application
of one type of CEA—a modified quality-adjusted life-years (QALYs) approach. For CAIR,
EPA concluded that the direct usefulness of cost-effectiveness analysis is mitigated by the lack
of rule alternatives to compare relative effectiveness, but that comparisons could still be made to
other benchmarks bearing in mind methodological differences.
QALYs were developed to evaluate the effectiveness of individual medical treatments,
and EPA is still evaluating the appropriate methods for CEA of environmental regulations.
Agency concerns with the standard QALY methodology include the treatment of people with
fewer years to live (the elderly); fairness to people with preexisting conditions that may lead to
reduced life expectancy and reduced quality of life; and how the analysis should best account for
nonhealth benefits, such as improved visibility.
The Institute of Medicine (a member institution of the National Academies of Science)
established the Committee to Evaluate Measures of Health Benefits for Environmental, Health,
and Safety Regulation to assess the scientific validity, ethical implications, and practical utility
of a wide range of effectiveness measures used or proposed in CEA. This committee prepared a
report titled "Valuing Health for Regulatory Cost-Effectiveness Analysis," which concluded that
CEA is a useful tool for assessing regulatory interventions to promote human health and safety,
although not sufficient for informed regulatory decisions (Miller, Robinson, and Lawrence,
2006).65 They emphasized the need for additional data and methodological improvements for
CEA analyses, and urged greater consistency in the reporting of assumptions, data elements, and
analytic methods. They also provided a number of recommendations for the conduct of
regulatory CEA analyses. EPA is evaluating these recommendations and will determine a
response for upcoming analyses.
In Appendix G of the RIA for the CAIR,63 EPA conducted an extensive cost-
effectiveness analysis using morbidity inclusive life years (MILY). That analysis concluded that
reductions in PM2.s associated with CAIR were expected to be cost-saving (because the value of
expenditures on illnesses and non-health benefits exceeded costs), and that costs of the CAIR
could have been significantly higher and still result in cost-effective improvements in public
health. Because the current analysis relies on a benefits transfer approach to estimate PM-related
benefits, scaling PM benefits from the CAND rule, we do not have the necessary inputs to
develop a valid cost-effectiveness measure for the final cold temperature standards.
Furthermore, the CAND analysis did not include a health-based CEA, the results of which might
have been scaled in a similar fashion to the benefits.
12-37
-------
For the CAVR rule, EPA was able to draw inferences from the CAIR CEA by scaling the
relative magnitude of the costs and health impacts between the two rules.66 While the CAVR
was not expected to be cost-saving like CAIR, EPA expected that CAVR was likely to have a
relatively low cost per MILY. For the final cold temperature standards, however, it is difficult to
draw similar inferences with CAIR because the geographic distribution of emission changes, the
distribution of those changes over time, and the age distribution of the mortality and chronic
disease reductions are all expected to differ between the two rules. For these reasons, we do not
scale the CAIR health-based cost-effectiveness analysis for the final cold temperature standards.
12.8 Comparison of Costs and Benefits
The final rule provides three separate provisions that reduce air toxics emissions: cold
temperature vehicle controls, an emissions control program for PFCs, and a control program
limiting benzene in gasoline. A full appreciation of the overall economic consequences of these
provisions requires consideration of the benefits and costs expected to result from each standard,
not just those that could be expressed here in dollar terms. As noted above, due to limitations in
data availability and analytical methods, our benefits analysis only monetizes the PM2.s-related
benefits from direct PM emission reductions associated with the cold temperature standards.
There are a number of health and environmental effects associated with the final standards that
we were unable to quantify or monetize (see Table 12.1-2).
Table 12.8-1 contains the estimates of monetized benefits of the final cold temperature
vehicle standards and estimated social welfare costs for each of the final control programs.8 The
annual social welfare costs of all provisions of this rule are described more fully in Chapter 13.
It should be noted that the estimated social welfare costs for the vehicle program contained in
this table are for 2019. The 2019 vehicle program costs are included for comparison purposes
only and are therefore not included in the total 2020 social costs. There are no compliance costs
associated with the vehicle program after 2019; as explained in Chapter 13, the vehicle
compliance costs are primarily R&D and facilities costs that are expected to be recovered by
manufacturers over the first ten years of the program.
The results in Table 12.8-1 suggest that the 2020 monetized benefits of the cold
temperature vehicle standards are greater than the expected social welfare costs of that program
in 2019. Specifically, the annual benefits of the program will be approximately $3,300 + B
million or $3,000 + B million annually in 2020 (using a three percent and seven percent discount
rate in the benefits analysis, respectively), compared to estimated social welfare costs of
approximately $10.6 million in the last year of the program (2019). These benefits are expected
to increase to $6,300 + B million or $5,700 + B million annually in 2030 (using a three percent
and seven percent discount rate in the benefits analysis, respectively), even as the social welfare
costs of that program fall to zero. Table 12.8-1 also presents the costs of the other rule
provisions: an emissions control program for PFCs and a control program limiting benzene in
s Social costs represent the welfare costs of the rule to society. These social costs do not consider transfer payments
(such as taxes) that are simply redistributions of wealth.
12-38
-------
gasoline. Though we are unable to present the benefits associated with these two programs, we
note for informational purposes that the benefits associated with the final cold temperature
vehicle standards alone exceed the costs of all three rule provisions combined.
Table 12.8-1. Summary of Annual Benefits of the Final Cold Temperature Vehicle
Standards and Costs of All Provisions of the Final Standards3
(Millions of 2003 dollars)
Description
Estimated Social Welfare Costsb
Cold Temperature Vehicle Standards
PFC Container Standards
Fuel Standards'1
Total
Fuel Savings
Net Social Welfare Costs
Total PM2.5-Related Health Benefits of the Cold
Temperature Vehicle Standards6
3 percent discount rate
7 percent discount rate
2020
(Millions of 2003
dollars)
$10.6C
$37.5
$402.6
$440.1
-$80.7
$359.4
$3,300 + Bf
$3,000 + Bf
2030
(Millions of 2003
dollars)
$0
$45.7
$445.8
$491.5
-$91.5
$400.0
$6,300 + Bf
$5,700 + Bf
a All estimates are rounded to two significant digits and represent annualized benefits and costs anticipated for the years 2020 and
2030, except where noted. Totals may not sum due to rounding.
bNote that costs are the annual costs of reducing all pollutants associated with each provision of the final MSAT control package
in 2020 and 2030 (unless otherwise noted). To estimate fixed costs associated with the vehicle standards, we use a 7 percent
average before-tax rate of return over 5 years to amortize the capital fixed costs. For the fuel standards, we use a 7 percent
before-tax rate of return over 15 years to amortize the capital costs. Note that by 2020, PFC container standard costs are only
variable and do not use a rate of return assumption. See Chapters 8 and 9 for discussion of the vehicle and fuel standard costs,
respectively. In Chapter 13, however, we do use both a 3 percent and 7 percent social discount rate to calculate the net present
value of total social costs consistent with EPA and OMB guidelines for preparing economic analyses (US EPA, 2000 and OMB,
2003).
0 These costs are for 2019; the vehicle program compliance costs terminate after 2019 and are included for illustrative purposes.
They are not included in the total social welfare cost sum for 2020.
d Our modeling for the total costs of the proposed gasoline benzene program included participation by California refineries
(achieving benzene reductions below the 0.62 proposed benzene standard - thus generating credits), since it was completed
before we decided that California gasoline would not be covered by the program. For the final rule, we exclude California
refineries from the analysis. By excluding California refineries, other higher cost refineries will have to comply in their place,
slightly increasing the costs for the program.
e Annual benefits reflect only direct PM reductions associated with the cold temperature vehicle standards. Annual benefits
analysis results reflect the use of a 3 percent and 7 percent discount rate in the valuation of premature mortality and nonfatal
myocardial infarctions, consistent with EPA and OMB guidelines for preparing economic analyses (US EPA, 2000 and OMB,
2003). 61'6S Valuation of premature mortality based on long-term PM exposure assumes discounting over the SAB recommended
20-year segmented lag structure described in the Regulatory Impact Analysis for the Final Clean Air Interstate Rule (March
2005). Valuation of nonfatal myocardial infarctions (MI) assumes discounting over a 5-year period, reflecting lost earnings and
direct medical costs following a nonfatal MI. Note that we do not calculate a net present value of benefits associated with the
cold temperature vehicle standards.
f Not all possible benefits or disbenefits are quantified and monetized in this analysis. B is the sum of all unquantified benefits
and disbenefits. Potential benefit categories that have not been quantified and monetized are listed in Table 12.1-2.
12-39
-------
References for Chapter 12
1 U.S. Environmental Protection Agency. March 2005. Regulatory Impact Analysis for the
Final Clean Air Interstate Rule. Prepared by: Office of Air and Radiation. Available at
http://www.epa.gov/cair. Accessed December 15, 2005.
2 U.S. Environmental Protection Agency. May 2004. Final Regulatory Analysis: Control of
Emissions from Nonroad Diesel Engines. Prepared by: Office of Air and Radiation. Available
at http://www.epa.gov/nonroad-diesel/2004fr.htmtfdocuments. Accessed December 15, 2005.
3 U.S. Environmental Protection Agency. 2004. Air Quality Criteria for Particulate Matter
Volume II of II. National Center for Environmental Assessment, Office of Research and
Development. Research Triangle Park, NC. EPA/600/P-99/002bF.
4 U.S. Environmental Protection Agency. September 2000. Guidelines for Preparing Economic
Analyses. EPA240-R-00-003.
5 U.S. Office of Management and Budget (OMB). 2003. Circular A-4 Guidance for Federal
Agencies Preparing Regulatory Analyses, Available at:
http://www/whitehouse.gov/omb/inforeg/iraguide.html. Accessed December 15, 2005.
6 Science Advisory Board. 2001. NATA - Evaluating the National-Scale Air Toxics
Assessment for 1996 - an SAB Advisory, http://www.epa.gov/ttn/atw/sab/sabrev.html.
7 Kunzli, N., S. Medina, R. Kaiser, P. Quenel, F. Horak Jr, and M. Studnicka. 2001.
"Assessment of Deaths Attributable to Air Pollution: Should We Use Risk Estimates Based on
Time Series or on Cohort Studies?" American Journal of Epidemiology 153(11): 1050-55.
8 U.S. Environmental Protection Agency. May 2004. Final Regulatory Analysis: Control of
Emissions from Nonroad Diesel Engines. Prepared by: Office of Air and Radiation. Available
at http://www.epa.gov/nonroad-diesel/2004fr.htmtfdocuments. Accessed December 15, 2005.
9 U.S. Environmental Protection Agency. May 2004. Final Regulatory Analysis: Control of
Emissions from Nonroad Diesel Engines. Prepared by: Office of Air and Radiation. Available
at http://www.epa.gov/nonroad-diesel/2004fr.htmtfdocuments. Accessed December 15, 2005.
10 Abt Associates. 2005. Methodology for County-Level Mortality Rate Projections.
Memorandum from Ellen Post and Don McCubbin (Abt Associates) to Bryan Hubbell and Zach
Pekar (EPA). Sent October 25, 2005.
11 Pope, C.A., III, R.T. Burnett, M.J. Thun, E.E. Calle, D. Krewski, K. Ito, and G.D. Thurston.
2002. "Lung Cancer, Cardiopulmonary Mortality, and Long-term Exposure to Fine Particulate
Air Pollution." Journal of the American Medical Association 287:1132-1141.
12 Woodruff, T.J., J. Grille, and K.C. Schoendorf. 1997. "The Relationship Between Selected
Causes of Postneonatal Infant Mortality and Particulate Air Pollution in the United States."
Environmental Health Perspectives 105(6):608-612.
13 Abbey, D.E., B.L. Hwang, R.J. Burchette, T. Vancuren, and P.K. Mills. 1995. "Estimated
Long-Term Ambient Concentrations of PM(10) and Development of Respiratory Symptoms in a
Nonsmoking Population." Archives of Environmental Health 50(2): 139-152.
14 Peters, A., D.W. Dockery, I.E. Muller, and M.A. Mittleman. 2001. "Increased Particulate Air
Pollution and the Triggering of Myocardial Infarction." Circulation 103:2810-2815.
15 Moolgavkar, S.H. 2003. "Air Pollution and Daily Deaths and Hospital Admissions in Los
Angeles and Cook Counties." In Revised Analyses of Time-Series Studies of Air Pollution and
Health. Special Report. Boston, MA: Health Effects Institute.
12-40
-------
16 Ito, K. 2003. "Associations of Parti culate Matter Components with Daily Mortality and
Morbidity in Detroit, Michigan." In Revised Analyses of Time-Series Studies of Air Pollution
and Health. Special Report. Health Effects Institute, Boston, MA.
17 Moolgavkar, S.H. 2000. "Air Pollution and Hospital Admissions for Diseases of the
Circulatory System in Three U.S. Metropolitan Areas." Journal of the Air and Waste
Management Association 50:1199-1206.
18 Sheppard, L. 2003. "Ambient Air Pollution and Nonelderly Asthma Hospital Admissions in
Seattle, Washington, 1987-1994." In Revised Analyses of Time-Series Studies of Air Pollution
and Health. Special Report. Boston, MA: Health Effects Institute.
19 Norris, G., S.N. YoungPong, J.Q. Koenig, T.V. Larson, L. Sheppard, and J.W. Stout. 1999.
"An Association between Fine Particles and Asthma Emergency Department Visits for Children
in Seattle." Environmental Health Perspectives 107(6):489-493.
20 Dockery, D.W., J. Cunningham, A.I. Damokosh, L.M. Neas, J.D. Spengler, P. Koutrakis, J.H.
Ware, M. Raizenne, and F.E. Speizer. 1996. "Health Effects of Acid Aerosols On North
American Children-Respiratory Symptoms." Environmental Health Perspectives 104(5): 500-
505.
21 Pope, C.A., III, D.W. Dockery, J.D. Spengler, and M.E. Raizenne. 1991. "Respiratory Health
and PMio Pollution: A Daily Time Series Analysis." American Review of Respiratory Diseases
144:668-674.
22 Schwartz, J., and L.M. Neas. 2000. "Fine Particles are More Strongly Associated than Coarse
Particles with Acute Respiratory Health Effects in Schoolchildren." Epidemiology 11:6-10.
23 Ostro, B., M. Lipsett, J. Mann, H. Braxton-Owens, and M. White. 2001. "Air Pollution and
Exacerbation of Asthma in African-American Children in Los Angeles." Epidemiology
12(2):200-208.
24 Vedal, S., J. Petkau, R. White, and J. Blair. 1998. "Acute Effects of Ambient Inhalable
Particles in Asthmatic and Nonasthmatic Children." American Journal of Respiratory and
Critical Care Medicine 15 7(4): 103 4-1043.
25 Ostro, B.D. 1987. "Air Pollution and Morbidity Revisited: A Specification Test." Journal of
Environmental Economics Management 14:87-98.
26 Ostro, B.D. and S. Rothschild. 1989. "Air Pollution and Acute Respiratory Morbidity: An
Observational Study of Multiple Pollutants." Environmental Research 50:238-247.
27 Mrozek, J.R., and L.O. Taylor. 2002. "What Determines the Value of Life? AMeta-
Analysis." Journal of Policy Analysis and Management 21(2):253-270.
28 Viscusi, V.K., and I.E. Aldy. 2003. "The Value of a Statistical Life: A Critical Review of
Market Estimates Throughout the World." Journal of Risk and Uncertainty 27(l):5-76.
29 Viscusi, W.K., W. A. Magat, and J. Huber. 1991. "Pricing Environmental Health Risks:
Survey Assessments of Risk-Risk and Risk-Dollar Trade-Offs for Chronic Bronchitis." Journal
of 'EnvironmentalEconomics andManagement 21:32-51.
30 Cropper, M.L., and AJ. Krupnick. 1990. "The Social Costs of Chronic Heart and Lung
Disease." Resources for the Future. Washington, DC. Discussion Paper QE 89-16-REV.
31 Russell, M.W., D.M. Huse, S. Drowns, E.C. Hamel, and S.C. Hartz. 1998. "Direct Medical
Costs of Coronary Artery Disease in the United States." American Journal of Cardiology
110-1115.
12-41
-------
32 Wittels, E.H., J.W. Hay, and A.M. Gotto, Jr. 1990. "Medical Costs of Coronary Artery
Disease in the United States." American Journal of Cardiology 65(7):432-440.
33 Agency for Healthcare Research and Quality (AHRQ). 2000. HCUPnet, Healthcare Cost and
Utilization Project. Rockville, MD. http://www.ahrq.gov/HCUPnet/.
34 Smith, D.H., D.C. Malone, K.A. Lawson, LJ. Okamoto, C. Battista, and W.B. Saunders.
1997. "A National Estimate of the Economic Costs of Asthma." American Journal of
Respiratory and Critical Care Medicine 156(3 Pt l):787-793.
35 Stanford, R., T. McLaughlin, and LJ. Okamoto. 1999. "The Cost of Asthma in the
Emergency Department and Hospital." American Journal of Respiratory and Critical Care
Medicine 160(1):211-215.
36 Industrial Economics, Incorporated (lEc). March 31, 1994. Memorandum to Jim DeMocker,
Office of Air and Radiation, Office of Policy Analysis and Review, U.S. Environmental
Protection Agency.
37 Rowe, R.D., and L.G. Chestnut. 1986. "Oxidants and Asthmatics in Los Angeles: A Benefits
Analysis—Executive Summary." Prepared by Energy and Resource Consultants, Inc. Report to
the U.S. Environmental Protection Agency, Office of Policy Analysis. EPA-230-09-86-018.
Washington, DC.
38 Neumann, J.E., M.T. Dickie, and R.E. Unsworth. March 31, 1994. "Linkage Between Health
Effects Estimation and Morbidity Valuation in the Section 812 Analysis—Draft Valuation
Document." Industrial Economics Incorporated (lEc) Memorandum to Jim DeMocker, U.S.
Environmental Protection Agency, Office of Air and Radiation, Office of Policy Analysis and
Review.
39 Tolley, G.S. et al. January 1986. Valuation of Reductions in Human Health Symptoms and
Risks. University of Chicago. Final Report for the U.S. Environmental Protection Agency.
40 Council of Economic Advisors. 2005. The Annual Report of the Council of Economic
Advisors. In: Economic Report of the President. Table B-60. U.S. Government Printing Office:
Washington, DC.
41 Pope, C.A., III, R.T. Burnett, M.J. Thun, E.E. Calle, D. Krewski, K. Ito, and G.D. Thurston.
2002. "Lung Cancer, Cardiopulmonary Mortality, and Long-term Exposure to Fine Particulate
Air Pollution." Journal of the American Medical Association 287:1132-1141.
42 Woodruff, T.J., J. Grille, and K.C. Schoendorf. 1997. "The Relationship Between Selected
Causes of Postneonatal Infant Mortality and Particulate Air Pollution in the United States."
Environmental Health Perspectives 105(6):608-612.
43 Pope, C.A., III, R.T. Burnett, M.J. Thun, E.E. Calle, D. Krewski, K. Ito, and G.D. Thurston.
2002. "Lung Cancer, Cardiopulmonary Mortality, and Long-term Exposure to Fine Particulate
Air Pollution." Journal of the American Medical Association 287:1132-1141.
44 Woodruff, T.J., J. Grille, and K.C. Schoendorf. 1997. "The Relationship Between Selected
Causes of Postneonatal Infant Mortality and Particulate Air Pollution in the United States."
Environmental Health Perspectives 105(6):608-612.
45 U.S. Office of Management and Budget (OMB). 2003. Circular A-4 Guidance for Federal
Agencies Preparing Regulatory Analyses, Available at:
http://www/whitehouse.gov/omb/inforeg/iraguide.html. Accessed December 15, 2005.
12-42
-------
46 U.S. Office of Management and Budget (OMB). 2003. Circular A-4 Guidance for Federal
Agencies Preparing Regulatory Analyses, Available at:
http://www/whitehouse.gov/omb/inforeg/iraguide.html. Accessed December 15, 2005.
47 U.S. Environmental Protection Agency. March 2005. Regulatory Impact Analysis for the
Final Clean Air Interstate Rule. Prepared by: Office of Air and Radiation. Available at
http://www.epa.gov/cair. Accessed December 15, 2005.
48 U.S. Environmental Protection Agency (EPA). 1996. Review of the National Ambient Air
Quality Standards for Ozone: Assessment of Scientific and Technical Information. Office of Air
Quality Planning and Standards, Research Triangle Park, NC, EPA report no. EPA/4521R-96-
007.
49 Garcia, P., B. Dixon, and J. Mjelde. 1986. "Measuring the Benefits of Environmental Change
Using a Duality Approach: The Case of Ozone and Illinois Cash Grain Farms." Journal of
Environmental Economics and Management 13:69-80.
50 Fox, S., and R.A. Mickler. 1996. "Impact of Air Pollutants on Southern Pine Forests."
Ecological Studies 118. New York: Springer Verlag.
51 EPA-SAB-COUNCIL-ADV-98-003. 1998. "Advisory Council on Clean Air Compliance
Analysis Advisory on the Clean Air Act Amendments (CAAA) of 1990 Section 812 Prospective
Study: Overview of Air Quality and Emissions Estimates: Modeling, Health and Ecological
Valuation Issues Initial Studies."
52 Grosclaude, P., and N.C. Soguel. 1994. "Valuing Damage to Historic Buildings Using a
Contingent Market: A Case Study of Road Traffic Externalities." Journal of Environmental
Planning and Management 37: 279-287.
53 U.S. Environmental Protection Agency. 2005. Air Quality Criteria for Ozone and Related
Photochemical Oxidants (First External Review Draft). January.
http ://cfpub .epa. gov/ncea/cfm/recor display. cfm?deid= 114523
54 EPA, 2005. Air Quality Criteria for Ozone and Related Photochemical Oxidants (Second
External Review Draft). August. http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=137307
55 U.S. Environmental Protection Agency. March 2005. Regulatory Impact Analysis for the
Final Clean Air Interstate Rule. Prepared by: Office of Air and Radiation. Available at
http://www.epa.gov/cair. Accessed December 15, 2005.
56 U.S. Environmental Protection Agency. June 2005. Regulatory Impact Analysis for the Final
Clean Air Visibility Rule or the Guidelines for Best Available Retrofit Technology (BART)
Determinations Under the Regional Haze Regulations. Prepared by: Office of Air and
Radiation. Available at http://www.epa.gov/visibility/pdfs/bart ria 2005 6 15.pdf. Accessed
December 15,2005.
57 Krewski D., R.T. Burnett, M.S. Goldbert, K. Hoover, J. Siemiatycki, M. Jerrett, M.
Abrahamowicz, and W.H. White. July 2000. Reanalysis of the Harvard Six Cities Study and the
American Cancer Society Study of Particulate Air Pollution and Mortality. Special Report to the
Health Effects Institute, Cambridge MA.
58 EPA-SAB-COUNCIL_ADV_04-002. March 2004. Advisory on Plans for Health Effects
Analysis in the Analytical Plan for EPA 's Second Prospective Analysis - Benefits and Costs of
the Clean Air Act, 1990-2020: Advisory by the Health Effects Subcommittee of the Advisory
Council on Clean Air Compliance Analysis.
12-43
-------
59 Laden, F., J. Schwartz, F.E. Speizer, and D.W. Dockery. 2006. Reduction in Fine Particulate
Air Pollution and Mortality. American Journal of Respiratory and Critical Care Medicine. 173:
667-672.
60 U.S. EPA. 2004. Air Quality Criteria for Particulate Matter, Volume II. Office of Research
and Development. EPA/600/P-99/002bF, October.
61 U.S. Environmental Protection Agency. October 2006. Final Regulatory Impact Analysis
(RIA)for the Proposed National Ambient Air Quality Standards for Particulate Matter.
Prepared by: Office of Air and Radiation. Available at http://www.epa.gov/ttn/ecas/ria.html
Accessed October 18, 2006.
62 Clean Air Science Advisory Committee. June 2005. EPA 's Review of the National Ambient
Air Quality Standards for Particulate Matter (Second Draft PM Staff Paper, January 2005). A
Review by the PM Review Panel of the EPA Clean Air Science Advisory Committee. EPAS AB-
CASAC-05-007.
63 EPA-SAB-COUNCIL_ADV_04-002. March 2004. Advisory on Plans for Health Effects
Analysis in the Analytical Plan for EPA 's Second Prospective Analysis - Benefits and Costs of
the Clean Air Act, 1990-2020: Advisory by the Health Effects Subcommittee of the Advisory
Council on Clean Air Compliance Analysis.
64 National Research Council (NRC). 2002. Estimating the Public Health Benefits of Proposed
Air Pollution Regulations. Washington, DC: The National Academies Press.
65 Miller W, Robinson LA, Lawrence RS, eds. Valuing Health: Cost Effectiveness Analysis for
Regulation. Committee to Evaluate Measures of Health Benefits for Environmental, Health, and
Safety Regulation (Lawrence RS, chair), Board on Health Care Services, Institute of Medicine,
National Academy Press, Washington D.C., 2006.
66 U.S. Environmental Protection Agency. June 2005. Regulatory Impact Analysis for the Final
Clean Air Visibility Rule or the Guidelines for Best Available Retrofit Technology (BART)
Determinations Under the Regional Haze Regulations. Prepared by: Office of Air and
Radiation. Available at http://www.epa.gov/visibility/pdfs/bart_ria_2005_6_l 5.pdf. Accessed
December 15, 2005.
67 U.S. Environmental Protection Agency, 2000. Guidelines for Preparing Economic Analyses.
www.yosemitel.epa.gov/ee/epa/eed/hsf/pages/Guideline.html.
68 U.S. Office of Management and Budget (OMB). 2003. Circular A-4 Guidance for Federal
Agencies Preparing Regulatory Analyses, Available at:
http://www/whitehouse.gov/omb/inforeg/iraguide.html. Accessed December 15, 2005.
12-44
-------
Final Regulatory Impact Analysis
Chapter 13: Table of Contents
CHAPTER 13: Economic Impact Analysis 3
13.1 Overview and Results 3
13.1.1 What is an Economic Impact Analysis? 3
13.1.2 What is the Economic Impact Model? 4
13.1.3 What Economic Sectors are Included in the Economic Impact Model? 4
13.1.4.1 Market Analysis Results 7
13.1.4.2 Economic Welfare Results 9
13.2 Economic Methodology 14
13.2.1 What Is A Behavioral Economic Model? 14
13.2.2 What Is the Economic Theory Underlying the EIM? 15
13.2.2.1 Partial Market Equilibrium Model 15
13.2.2.2 Perfect Competition Model 17
13.2.3.3 Intermediate-Run Model 18
13.2.3 How is the EIM Used to Estimate Economic Impacts? 22
13.2.3.1 Estimation of Market Impacts 22
13.2.3.2 Estimation of Social Costs 24
13.2.4. How Are Special Market Characteristics Addressed? 26
13.2.4.1 Fixed and Variable Costs 27
13.2.4.2 Gasoline Fuel Savings and Fuel Taxes 29
13.2.4.3 Flexibility Provisions 30
13.2.4.4 Substitution 30
13.2.4.5 Market-Level Analysis 31
13.3 EIM Data Inputs and Model Solution 32
13.3.1 Description of Product Markets 32
13.3.1.1 Portable Fuel Container Market 32
13.3.1.2 Gasoline Fuel Market 33
13.3.2 Initial Market Conditions 35
13.3.2.1 Portable Fuel Container Market Quantities and Prices 35
13.3.2.2 Gasoline Fuel Market Quantities and Prices 36
13.3.3 Compliance Costs 39
13.3.3.1 Portable Fuel Container Compliance Costs 39
13.3.3.2 Gasoline Fuel Compliance Costs 40
13.3.3.3 Vehicle Compliance Costs 42
13.3.4 Gasoline Fuel Savings 43
13.3.5 Supply and Demand Elasticity Estimates 45
13.3.6 Economic Impact Model Structure 46
Appendix 13A: Impacts on Portable Fuel Container Markets 48
Appendix 13B: Impacts on Gasoline Fuel Markets 51
Appendix 13C: Time Series of Social Costs 56
Appendix 13D: Overview of Economic Model Equations 60
13D.1 Discussion and Specification of Model Equations 60
13D.2 Consumer and Producer Welfare Calculations 61
Appendix 13E: Elasticity Parameters 63
13-1
-------
13E.1 Gasoline Market Parameters 63
13E.2 Portable Fuel Container Market Parameters 64
13E.3 Portable Fuel Container Demand Elasticity Estimation Procedure 65
13E.3.1 Numerical Example: Base Case 66
13E.3.2 Numerical Example: Sensitivity 67
13E.4 Portable Fuel Container Supply Elasticity Estimation 68
13E.4.1 Data Sets 69
13E.4.2 Results of Supply Elasticity Estimation 70
Appendix 13F: Initial Market Equilibrium - Price Forecasts 72
Appendix 13G: Sensitivity Analyses 74
13G.1 Scenario 1: Model Elasticity Parameters 74
13G.1.1 Alternative Demand and Supply Elasticities 75
13G.1.2 Results 76
13G.2 Scenario 2: Fuel Market Compliance Costs 78
13G.2.1 Scenarios Modeled 78
13G.2.2 Compliance Costs 80
13G.2.3 Results 81
13G.3 Scenario 3: Alternative Gasoline Price 83
13G.4 Scenario 4: Alternative Social Discount Rates 90
13-2
-------
Final Regulatory Impact Analysis
CHAPTER 13: Economic Impact Analysis
We prepared an Economic Impact Analysis (EIA) to estimate the economic impacts of
this rule on the portable fuel container (PFC), gasoline fuel, and light-duty vehicle markets. In
this chapter we describe the Economic Impact Model (EIM) we developed to estimate both the
market-level changes in prices and outputs for affected markets and the social costs of the
program and their distribution across affected stakeholders. We also present the result of our
analysis.
We estimate the net social costs of the rule to be about $359.4 million in 2020. This
estimate reflects the estimated costs associated with compliance with the gasoline, PFC, and
vehicle controls and the expected gasoline fuel savings from better evaporative controls on PFCs.
The results of the economic impact modeling performed for the gasoline fuel and PFC control
programs suggest that the social costs of those two programs are expected to be about $440.1
million in 2020, with consumers of these products expected to bear about 58.4 percent of these
costs. We estimate gasoline fuel savings of about $80.7 million in 2020, which will accrue to
consumers. There are no social costs associated with the vehicle program in 2020 (these accrue
only in the 10-year period from 2010 through 2019). These estimates, and all costs presented in
this chapter, are in year 2003 dollars.
With regard to market-level impacts in 2020, the maximum price increase for gasoline
fuel is expected to be about 0.3 percent (0.5 cents per gallon), for PADD 5.A The price of PFCs
is expected to increase by about 1.9 percent ($0.20 per can) in areas that already have PFC
requirements and 32.5 percent ($1.52 per can) in areas that do not.
13.1 Overview and Results
13.1.1 What is an Economic Impact Analysis?
An Economic Impact Analysis (EIA) is prepared to inform decision makers about the
potential economic consequences of a regulatory action. The analysis consists of estimating the
social costs of a regulatory program and the distribution of these costs across stakeholders.
These estimated social costs can then be compared with estimated social benefits (as presented in
Chapter 12). As defined in EPA's Guidelines for Preparing Economic Analyses, social costs are
the value of the goods and services lost by society resulting from a) the use of resources to
comply with and implement a regulation and b) reductions in output. * In this analysis, social
costs are explored in two steps. In the market analysis, we estimate how prices and quantities of
goods affected by the rule can be expected to change once the program goes into effect. In the
economic welfare analysis, we look at the total social costs associated with the program and their
distribution across stakeholders.
A PADD: Petroleum Administration for Defense District.
13-3
-------
13.1.2 What is the Economic Impact Model?
The Economic Impact Model (EIM) is a behavioral model developed to estimate price
and quantity changes and total social costs associated with the emission controls set out in this
rule. The model relies on basic microeconomic theory to simulate how producers and consumers
of affected products can be expected to respond to an increase in production costs associated
with compliance with the emission control program. The economic theory that underlies the
model is described in detail in Section 13.2, below.
The ELM is designed to estimate the economic impacts of the rule by simulating
economic behavior. At current, pre-control market equilibrium conditions consumers are willing
to purchase the same amount of that product that producers are willing to produce at that price.
This is represented by pre-control market prices and quantities. Compliance with the standards
ould increase the production costs of affected goods by the amount of the compliance costs. This
represents a "shock" to equilibrium market conditions. Producers of affected products will try to
pass some or all of the increased costs on to the consumers of these goods through price
increases. In response to the price increases, consumers will adjust their consumption of affected
goods. Producers will react to the change in quantity demanded by adjusting their prices and the
quantity they produce. These interactions continue until a new market equilibrium price and
quantity combination is achieved. The amount of the compliance costs that can be passed on to
consumers is ultimately limited by the price sensitivity of purchasers and producers in the
relevant market (price elasticity of demand and supply). The EIM explicitly models these
behavioral responses and estimates new equilibrium prices and output and the resulting
distribution of social costs across these stakeholders (producers and consumers).
13.1.3 What Economic Sectors are Included in the Economic Impact Model?
There are three economic sectors affected by the control programs described in this rule:
PFCs, gasoline fuel, and light-duty vehicles.
In this Economic Impact Analysis we do not model the market impacts on the vehicle
program; we model only the impacts on the PFC and gasoline fuel markets. This approach is
appropriate for several reasons. As described in Chapter 8, above, the compliance costs for the
light-duty vehicle controls are expected to be very small, less than $1 per vehicle. These costs
are R&D and facilities costs that are expected to be recovered by the manufacturers over 10
years (completely recovered by 2019) and are not expected to be passed on in the form of higher
prices. Such small compliance costs are well within the normal variation of input prices
experienced by most vehicle manufacturers at any given time. In addition, a price change this
small, even if it is passed on entirely, is unlikely to affect producer or consumer behavior given
the price of a new vehicle. On a more practical level, a cost increase of this magnitude is not
large enough to disturb an economic impact model like the one used in this analysis. At the
same time, however, the light-duty vehicle compliance costs are a cost to society and should be
included in the economic welfare analysis. We do this by using the engineering cost estimates as
a proxy for the social costs of the light-duty vehicle controls and adding them to the estimated
social costs of the gasoline fuel and PFC programs.
13-4
-------
Final Regulatory Impact Analysis
With regard to the gasoline fuel and PFC market analyses, we model the impacts on
residential users of these products. This means that we focus the analysis on the use of these
products for personal transportation (gasoline fuel) or residential lawn and garden care or
recreational uses (PFCs) and do not separately model how the costs of complying with the
standards may affect the production of goods and services that use gasoline fuel or PFCs as
production inputs. The result is that we group residential and commercial users in a single
market and assume the behavioral responses to increased costs for commercial users are similar
to residential users. This is reasonable because the vast majority of users of these products are
residential users. While there are commercial users of PFCs and gasoline fuel, their share of the
end-user markets is relatively small. The U.S Department of Energy estimates that about 92
percent of gasoline used in the United States for transportation is used in light-duty vehicles.2
According to DoE, only about six percent of gasoline fuel is used for commercial or industrial
transportation, and the remaining two percent is used in recreational marine vessels. Similarly,
although there is little publicly available national data on the users of PFCs, a 1999 study by
CARB found that 94 percent of portable fuel containers in California were used by residential
households.3 In addition, for most commercial users the share of these products to total
production costs is small (e.g., the cost of a PFC is only a very small part of the total production
costs for an agricultural or construction firm). Therefore, a price increase of the magnitude
anticipated for this control program is not expected to have a noticeable impact on prices or
quantities of goods produced using these inputs (e.g., agricultural produce or buildings).
Consistent with the cost analysis, the economic impact analysis for the gasoline fuel
market does not distinguish between reformulated and conventional gasoline fuels.8 For more
information, see Chapter 9 on how gasoline compliance costs were estimated. Also consistent
with the cost analysis, this EIA also does not consider impacts of the fuel program on the
benzene market (i.e., the market for recovered benzene). This is because, as explained elsewhere
in this RIA, any impacts on that market are expected to be insignificant. Finally, as explained in
Section 13.3.2.2, the gasoline fuel analysis is based on post-tax gasoline prices since state and
federal taxes are included in the prices consumers pay at the pump.
The EIM relies on the estimated compliance costs for the PFC and gasoline fuel programs
described elsewhere in this RIA. Thus, the EIM reflects cost savings associated with ABT or
other flexibility programs to the extent they are included in the estimated compliance costs.
As summarized in Table 13.1-1, this EIA considers the economic impacts of the rule on
four gasoline fuel markets and two PFC markets, for a total of six markets. More detailed
information on the markets and model inputs is provided in Section 13.3.3, below, and in the
industry profiles prepared for this rule (see also Chapter 4 of this document).4'5
B The cost analysis does not differentiate between conventional and reformulated gasoline because their benzene
levels are expected to be similar as a result of the standards and because the cost modeling technique does not allow
for estimating how the blending of gasoline blendstocks will occur.
13-5
-------
Table 13.1-1. Summary of Markets in Economic Impact Model
Model Dimension
Number of Markets
Geographic scope
Market structure
Baseline population
Growth projections
Supply elasticity
Demand elasticity
Regulatory shock
Light-Duty
Vehicles
Not included in
market analysis;
engineering
costs used to
estimate total
social costs
Gasoline (4)
Four regions
• PADDs 1 & 3
• PADD2
• PADD4
• PADD 5 (includes Alaska
and Hawaii; California not
included)
No distinction between
conventional and reformulated
gasoline
49-state; California not included in
the program because they already
control fuel benzene to low levels
Perfectly competitive
Energy Information Administration
Energy Information Administration
Literature estimate: 0.2 (inelastic)
Literature estimate: -0.2 (inelastic)
Direct compliance costs (fixed +
variable) cause shift in supply
function
Portable Fuel Containers (2)
Two markets
• States with current controls
(12 plus DC)
• States without current
controls (38)
50-State
Perfectly competitive
Provided by manufacturers
2%
Econometric estimate (production
function cost minimization method):
1.5 (elastic)
EPA estimate (Hicks-Allen derived
demand method): -0.01 (inelastic)
Direct compliance costs (fixed +
variable) cause shift in supply
function
In the EIM, behavioral responses to price changes are incorporated through the price
elasticity of supply and demand (reflected in the slope of the supply and demand curves). The
price elasticities used in this analysis are described in Section 13.3, below. The gasoline fuel
price elasticity parameters were obtained from the literature; we estimated those for the PFCs.
For gasoline fuel, both the demand and supply elasticities are inelastic, meaning that both the
quantity supplied and demanded are expected to be fairly insensitive to price changes. For PFCs,
however, the demand elasticity is inelastic but the supply elasticity is elastic. This means that
producers are expected to be sensitive to price changes but consumers are not. This will allow
producers to pass more of the compliance costs on to consumers.
13-6
-------
Final Regulatory Impact Analysis
13.1.4 Summary of Results
The EIA consists of two parts: a market analysis and welfare analysis. The market
analysis looks at expected changes in prices and quantities for affected products. The welfare
analysis looks at economic impacts in terms of annual and present value changes in social costs.
For this rule, the social costs are estimated as the sum of market surplus (the aggregate change in
consumer and producer surplus based on the estimated market impacts associated with the rule)
offset by operating cost savings (the gasoline fuel savings associated with better evaporative
controls for PFCs).
Economic impact results of our modeling for selected years are summarized in this
section. The year 2009 is presented because that is the first year in which both the PFC and the
gasoline programs are in effect (the PFC program begins in 2009; the gasoline fuel program go
into effect January 1, 2011 but the compliance cost analysis includes a phase-in starting in 2007
that ends May 2015). The year 2012 is presented because it is a high cost year due to the way
the fuel program compliance costs were estimated.0 The year 2015 is presented because
beginning with that year compliance costs are stabilized for future years for both the gasoline
and PFC programs (the vehicle program compliance costs continue for five more years). More
detailed results for all years are included in the appendices to this chapter.
13.1.4.1 Market Analysis Results
In the market analysis, we estimate how prices and quantities of goods affected by the
emission control program can be expected to change once the program goes into effect. As
explained above, we estimated market impacts for only the gasoline fuel and PFC markets. The
analysis relies on the baseline equilibrium prices and quantities for each market and the price
elasticity of supply and demand. It predicts market reactions to the increase in production costs
due to the new compliance costs. It should be noted that this analysis does not allow any other
factors to vary. In other words, it does not consider that manufacturers may adjust their
production processes or marketing strategies in response to the control program.
The market analysis results for 2009, 2012, 2015, and 2020 are presented in Table 13.1-2.
With regard to the gasoline fuel program, the market impacts are expected to be small, on
average. The price of gasoline fuel is expected to increase by less than 0.5 percent, depending on
PADD, with smaller increases during the program phase-in. The expected reduction in quantity
of fuel produced is expected to be less than 0.1 percent.
The market impacts for the PFC program are expected to be more significant. In 2009,
the first year of PFC program, the model predicts a price increase of about seven percent for
PFCs in states that currently have regulations for PFCs and about 57 percent for those that do
not. Even with these large price increases, however, the quantity produced is not expected to
decrease by very much: less than 0.6 percent. These percent price increases and quantity
decreases are much smaller after the first five years. In 2015, the estimated PFC price increase is
c Actual fuel program compliance costs are expected to be spread more smoothly across years.
13-7
-------
expected to be less than two percent for states that currently regulate PFCs and about 32.5
percent for states without such regulations. The quantity produced is expected to decrease by
less than 0.4 percent. The results for 2020 are substantially the same as 2015, with larger
decreases in the number of PFCs produced.
Table 13.1-2. Summary of Market Impacts (2009, 2012, 2015 and 2020; 2003$)
Market
Engineering
Cost Per Unit
Change in Price
Absolute
Percent
Change in Quantity
Absolute
Percent
2009
Gasoline Fuel
PADD 1 & 3
PADD2
PADD 4
PADD 5 (w/out CA)
Portable Fuel
Containers
States with existing
Programs
States without
existing
programs
0/gallon
0.0160
0.0910
0.0330
0.0070
0/gallon
0.0090
0.0500
0.0180
0.0040
0.006%
0.033%
0.011%
0.002%
$/can
$0.77
$2.70
$0.76
$2.68
6.9%
57.5%
Million Gallons
-0.9
-2.7
-0.1
-0.0
0.001%
-0.007%
-0.002%
0.000%
Thousand Cans
-8.0
-104.7
-0.07%
-0.57%
2012
Gasoline Fuel
PADD 1 & 3
PADD 2
PADD 4
PADD 5 (w/out CA)
Portable Fuel
Containers
States with existing
Programs
States without
existing
programs
0/gallon
0.0580
0.3080
0.2130
0.1400
0.0320
0.1680
0.1160
0.0760
0.021%
0.111%
0.074%
0.046%
$/can
$0.77
$2.70
$0.76
$2.68
6.9%
57.5%
Million Gallons
-3.3
-9.7
-0.8
-0.8
-0.004%
-0.022%
-0.015%
-0.009%
Thousand Cans
-8.5
-111.1
-0.07%
-0.57%
13-8
-------
Final Regulatory Impact Analysis
2015
Gasoline Fuel
PADD 1 & 3
PADD2
PADD 4
PADD 5 (w/out CA)
Portable Fuel
Containers
States with existing
Programs
States without
existing
programs
0/gallon
0.1490
0.3070
0.5010
0.9970
0.0810
0.1670
0.2730
0.5440
0.055%
0.111%
0.174%
0.327%
$/can
$0.21
$1.53
$0.20
$1.52
1.9%
32.5%
Million Gallons
-8.9
-10.1
-1.8
-6.1
-0.011%
-0.022%
-0.035%
-0.065%
Thousand Cans
-2.4
-66.7
-0.02%
-0.32%
2020
Gasoline Fuel
PADD 1 & 3
PADD 2
PADD 4
PADD 5 (w/out CA)
Portable Fuel
Containers
States with existing
Programs
States without
existing
programs
0/gallon
0.1490
0.3070
0.5010
0.9970
0.0810
0.1670
0.2730
0.5440
0.055%
0.111%
0.174%
0.327%
$/can
$0.21
$1.53
$0.20
$1.52
1.9%
32.5%
Million Gallons
-9.5
-10.7
-2.0
-6.4
-0.011%
-0.022%
-0.035%
-0.065%
Thousand Cans
-2.7
-73.6
-0.02%
-0.32%
13.1.4.2 Economic Welfare Results
In the economic welfare analysis we look at the costs to society of the rule in terms of
losses to key stakeholder groups that are the producers and consumers in the gasoline and PFC
markets. These surplus losses are combined with estimated vehicle compliance costs, gasoline
fuel savings, and government revenue losses to estimate the net economic welfare impacts of the
program. Detailed economic welfare results for the rule are presented in Appendix C and are
summarized below.
The estimated annual net social costs (total social costs less gasoline fuel savings) for all
years are presented in Table 13.1-3 and Figure 13.1-1. These social costs follow the trend of the
fuel program compliance costs. Initially, the estimated social costs of the program are relatively
small as the gasoline program begins to phase in. The net social costs increase to 2012, fall
somewhat for 2013 and 2014 due to changes in the fuel program compliance costs, and then
increase again in 2015, after which time the per-gallon costs are expected to be stable. Some of
the decrease in social costs in 2014 is also due a decrease in costs associated with the PFC
program, since fixed costs are fully amortized by 2014. The slight decrease in 2020 is due to the
13-9
-------
end of the vehicle compliance costs, which are incurred in the 10-year period from 2010 through
2019.
Table 13.1-3. Estimated Engineering Compliance and Social Costs Through 2035
(Smillion; 2003$)
Year
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
PFC
0.0
0.0
58.1
58.7
59.3
59.9
60.6
33.3
34.0
34.7
35.4
36.1
36.8
37.5
38.3
39.1
39.8
40.6
41.5
42.3
43.1
44.0
44.9
45.8
46.7
47.6
48.6
49.5
50.5
Vehicles
0.0
0.0
0.0
11.1
11.8
12.5
13.3
13.4
12.9
12.2
11.4
10.7
10.6
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
Gasoline
29.5
51.3
52.3
114.1
115.9
203.0
176.3
178.5
379.5
384.1
388.7
393.7
398.4
402.7
407.0
410.9
414.8
419.1
423.4
428.0
432.7
436.9
441.6
445.9
450.5
455.2
459.9
464.7
469.5
Fuel Savings
0.0
0.0
11.3
22.6
35.6
48.5
61.5
74.5
75.5
76.5
77.6
78.6
79.7
80.7
81.8
82.9
83.9
85.0
86.1
87.2
88.3
89.3
90.4
91.5
92.5
93.6
94.6
95.7
96.7
3% NPV (2006-35)
7%NPV (2006-35)
Total
Engineering
$29.5
$51.3
$99.0
$161.9
$152.6
$228.7
$190.9
$150.8
$350.8
$354.5
$358.0
$361.9
$366.1
$359.5
$363.5
$367.1
$370.7
$374.7
$378.7
$383.1
$387.5
$391.6
$396.0
$400.1
$404.6
$409.2
$413.9
$418.6
$423.4
$5,356.8
$2,901.0
Total Social
Costs
$29.5
$51.3
$98.9
$161.7
$152.4
$228.5
$190.8
$150.7
$350.7
$354.4
$357.9
$361.8
$366.0
$359.4
$363.4
$367.0
$370.6
$374.6
$378.6
$383.0
$387.4
$391.4
$395.9
$400.0
$404.5
$409.1
$413.7
$418.4
$423.2
$5,354.6
$2,899.7
13-10
-------
Final Regulatory Impact Analysis
Figure 13.1-1 Estimated Engineering Costs (Smillion, 2003$)
$600
$500
$100
$0
* m m m m
^m-m-m
2007 2009 2011 2013 2015 2017 2019 2021 2023 2025 2027 2029 2031 2033 2035
Year
• Gas cans A Vehicles )( Gasoline • Total ^— — Total Net Fuel Savings
Table 13.1-4 shows how the social costs are expected to be shared across stakeholders,
for selected years. Information for all years can be found in Appendix C. According to these
results, consumers are expected to bear approximately 99 percent of the cost of the PFC
program. This reflects the inelastic price elasticity on the demand side of the market and the
elastic price elasticity on the supply side. The burden of the gasoline fuel program is expected to
be shared more evenly, with about 54.5 percent expected to be borne by consumers and about
45.5 percent expected to be borne by producers. In all years, the estimated loss to consumer
welfare will be offset somewhat by the gasoline fuel savings associated with PFCs. Beginning at
about $11 million per year, these savings increase to about $76 million by 2015 as compliant
PFCs are phased in. These savings continue for the life of the PFCs; total annual savings
increase as the number of cans increases (see Table 13.3-9).
13-11
-------
Table 13.1-4. Summary of Estimated Social Costs, 2009, 2012, 2015 and 2020
(Smillion; 2003$)
Market
Change in
Consumer Surplus
Change in
Producer Surplus
Total
2009
Gasoline US
PADD 1 & 3
PADD2
PADD 4
PADD 5 (w/out CA)
Portable Fuel Containers US
States with existing programs
States without existing programs
Subtotal
Fuel Savings
Vehicle Program
Total
Gasoline US
PADD 1 & 3
PADD 2
PADD 4
PADD 5 (w/out CA)
Portable Fuel Containers US
States with existing programs
States without existing programs
Subtotal
Fuel Savings
Vehicle Program
Total
Gasoline US
PADD 1 & 3
PADD 2
PADD 4
PADD 5 (w/out CA)
Portable Fuel Containers US
States with existing programs
States without existing programs
Subtotal
Fuel Savings
Vehicle Program
-$28.5
(54.6%)
-$6.7
-$20.6
-$0.9
-$0.3
-$57.5
(99.3%)
-$8.9
-$48.7
-$86.1
(78.1%)
-$23.8
(45.4%)
-$5.6
-$17.2
-$0.7
-$0.3
-$0.4
(0. 7%)
-$0.1
-$0.3
-$24.1
(22%)
-$52.3
-$12.2
-$37.8
-$1.6
-$0.6
-$57.9
-$8.9
-$49.0
-$110.2
$11.3
$0
-$98.9
2012
-$110.7
(54.5%)
-$24.8
-$73.2
-$5.9
-$6.8
-$61.1
(99.3%)
-$9.4
-$51.7
-$171.8
(65.0%)
-$92.3
(45.5%)
-$20.7
-$61.0
-$4.9
-$5.7
-$0.4
(0. 7%)
-$0.1
-$0.3
-$92.7
(35.0%)
-$203.0
-$45.5
-$134.2
-$10.9
-$12.4
-$61.5
-$9.5
-$52.0
-$264.5
$48.5
-$12.5
-$228.5
2015
-$207.0
(54.5%)
-$66.3
-$75.9
-$14.5
-$50.3
-$33.7
(99.3%)
-$2.7
-$31.0
-$240.7
(58.2%)
-$172.5
(45.5%)
-$55.3
-$63.2
-$12.1
-$41.9
-$0.2
(0. 7%)
$0.0
-$0.2
-$172.7
(41.8%)
-$379.4
-$121.6
-$139.1
-$26.6
-$92.2
-$34.0
-$2.7
-$31.3
-$413.4
$75.5
-$12.9
13-12
-------
Final Regulatory Impact Analysis
Market
Total
Change in
Consumer Surplus
Change in
Producer Surplus
Total
-$350.7
2020
Gasoline US
PADD 1 & 3
PADD2
PADD 4
PADD 5 (w/out CA)
Portable Fuel Containers US
States with existing programs
States without existing programs
Subtotal
Fuel Savings
Vehicle Program
Total
-$219.6
(54.5%)
-$70.4
-$80.5
-$15.4
-$53.4
-$37.2
(99.3%)
-$3.0
-$34.3
-$256.8
(58.4%)
-$183.0
(45.5%)
-$58.6
-$67.1
-$12.8
-$44.5
-$0.3
(0. 7%)
$0.0
-$0.2
-$183.3
(41.6%)
-$402.6
-$129.0
-$147.6
-$28.2
-$97.8
-$37.5
-$3.0
-$34.5
-$440.1
$80.7
-$0
-$359.4
The present value of net social costs (discounted back to 2006) of the standards through
2035, contained in Table 13.1-3, is estimated to be about $5.4 billion (2003$). This present
value is calculated using a social discount rate of three percent and the stream of economic
welfare costs through 2035. We also performed an analysis using a seven percent social discount
rate.0 Using that discount rate, the present value of the net social costs through 2035 is
estimated to be about $2.9 billion (2003$).
D EPA presents the present value of cost and benefits estimates using both a three percent and a seven percent social
discount rate. According to OMB Circular A-4, "the 3 percent discount rate represents the 'social rate of time
preference'... [which] means the rate at which 'society' discounts future consumption flows to their present value";
"the seven percent rate is an estimate of the average before-tax rate of return to private capital in the U. S. economy
... [that] approximates the opportunity cost of capital."
13-13
-------
Table 13.1-5. Net Present of Estimated Social Costs 2007 through 2035, Discounted to 2006
(Smillion; 2003$)
Market
Gasoline, U.S.
PADD 1 & 3
PADD2
PADD 4
PADD 5 (w/out CA)
Portable Fuel Containers US
States with existing programs
States without existing programs
Subtotal
Fuel Savings
Vehicle Program
Total
Change in
Consumer Surplus
-$3,115.4
(54.6%)
-$959.7
-$1,260.4
-$210.8
-$684.5
-$754.9
(99.3%)
-$78.7
-$676.2
-$3870.3
59.8%
$1,208.0
-$2,662.3
Change in
Producer Surplus
-$2,596.2
(45.4%)
-$799.8
-$1,050.4
-$175.6
-$570.4
-$5.0
(0. 7%)
-$0.5
-$4.5
-$2,601.2
40.2%
-$91.1
-$2,692.3
Total
-$5,711.6
-$1,759.5
-$2,310.8
-$386.4
-$1,254.8
-$759.9
-$79.3
-$680.7
-$6,471.6
$1,208.0
-$91.1
-$5,354.6
Table 13.3-5 shows the distribution of total surplus losses for the cumulative net social
costs of the rule. This analysis includes the estimated social costs through 2035, discounted to
2006 at a 3 percent discount rate. These results suggest that consumers will bear about 60
percent of the total social costs associated with the PFC and gasoline fuel programs for that
period. The consumer share of the NPV social costs is about $3,870 million, or about 60 percent
of the total. Of that loss of consumer surplus, about $3,115 million (80 percent) is from the
gasoline fuel program. When the total costs of the program are taken into account, including the
fuel savings and the vehicle program costs, the loss of consumer surplus decreases to about
$2,662.3 million (about 50 percent of the social costs of the program).
13.2 Economic Methodology
Economic impact analysis uses a combination of theory and econometric modeling to
evaluate potential behavior changes associated with a new regulatory program. As noted above,
the goal is to estimate the impact of the regulatory program on producers and consumers. This is
done by creating a mathematical model based on economic theory and populating the model
using publicly available price and quantity data. A key factor in this type of analysis is
estimating the responsiveness of the quantity of PFCs and gasoline fuel demanded by consumers
or supplied by producers to a change in the price of that product. This relationship is called the
elasticity of demand or supply.
The EIM's methodology is rooted in applied microeconomic theory and was developed
following the OAQPS Economic Analysis Resource Document6 This section discusses the
economic theory underlying the modeling for this EIA and several key issues that affect the way
the model was developed.
13.2.1 What Is A Behavioral Economic Model?
13-14
-------
Final Regulatory Impact Analysis
Models incorporating different levels of economic decision making can be categorized as
with-bohavior responses or without-bohavior responses. The EIM is a behavioral model.
Engineering cost analysis is an example of a without-behavior response model. These
models estimate the cost of a regulation based on the projected number of affected units and
engineering estimates of the annualized costs. The result is an estimate of the total compliance
costs for a program. However, these models do not attempt to estimate how a regulatory
program will change the prices or output of an affected industry. Therefore, the results may
over-estimate the total costs of a program because they do not take decreases in quantity
produced into account.
The w/Y/z-behavior response approach builds on the engineering cost analysis and
incorporates economic theory related to producer and consumer behavior to estimate changes in
market conditions. As Bingham and Fox note, this framework provides "a richer story" of the
expected distribution of economic welfare changes across producers and consumers.7 In
behavioral models, manufacturers of goods affected by a regulation are economic agents that can
make adjustments, such as changing production rates or altering input mixes that will generally
affect the market environment in which they operate. As producers change their production
levels in response to a new regulation, consumers of the affected goods are typically faced with
changes in prices that cause them to alter the quantity that they are willing to purchase. These
changes in price and output from the market-level impacts are used to estimate the distribution of
social costs between consumers and producers.
If markets are competitive and per-unit regulatory costs are small, the behavioral
approach will yield approximately the same total cost impact as the engineering cost approach.
However, the advantage of the w/Y/z-behavior response approach is that it illustrates how the
costs flow through the economic system and it identifies which stakeholders, producers, and
consumers are likely to be most affected.
13.2.2 What Is the Economic Theory Underlying the EIM?
The EIM is a partial-equilibrium, single market numerical simulation model that
estimates price and quantity changes in the intermediate run under competitive market
conditions. Each of these model features is described in this section.
13.2.2.1 Partial Market Equilibrium Model
In the broadest sense, all markets are directly or indirectly linked in the economy, and a
new regulatory program will theoretically affect all commodities and markets to some extent.
However, not all regulatory programs have noticeable impacts on all markets. For example, a
regulation that imposes significant per unit compliance costs on an important manufacturing
input, such as steel, will have a larger impact on the national economy than a regulation that
imposes very small per unit compliance costs on an input used by only a small number of
producers.
13-15
-------
The appropriate level of market interactions to be included in an economic impact
analysis is determined by the number of industries directly affected by the requirements and the
ability of affected firms to pass along the regulatory costs in the form of higher prices. There are
at least three alternative approaches for modeling interactions between economic sectors that
reflect three different levels of analysis.
In ^partial equilibrium model, individual markets are modeled in isolation. The only
factor affecting the market is the cost of the regulation on facilities in the industry being
modeled; there are no interaction effects with other markets. Conditions in other markets are
assumed either to be unaffected by a policy or unimportant for cost estimation.
In a multimarket model, a subset of related markets is modeled together, with sector
linkages, and hence selected interaction effects, explicitly specified. This approach represents an
intermediate step between a simple, single-market partial equilibrium approach and a full general
equilibrium approach. This technique has most recently been referred to in the literature as
"partial equilibrium analysis of multiple markets".8
In a general equilibrium model, all sectors of the economy are modeled together,
incorporating interaction effects between all sectors included in the model. General equilibrium
models operationalize neoclassical microeconomic theory by modeling not only the direct effects
of control costs but also potential input substitution effects, changes in production levels
associated with changes in market prices across all sectors, and the associated changes in welfare
economy-wide. A disadvantage of general equilibrium modeling is that substantial time and
resources are required to develop a new model or tailor an existing model for analyzing
regulatory alternatives.
This EIM uses a partial equilibrium, single-market approach to model the economic
impacts of the rule. The model examines impacts that affect the two markets that are affected
(PFCs and gasoline) and does not look at potential impacts on other sectors of the economy.E
This approach is reasonable because, as described above, most of the users of these products are
households. For those commercial sectors that use these products, the impacts would be
expected to be negligible and not affect output in those sectors. With regard to the gasoline fuel
market, the estimated compliance costs on a per gallon basis are well within the normal price
variations of gasoline. With regard to PFCs, the share of these products to total production costs
is very small and therefore an increase in their price is not expected to change output. For these
reasons, the additional costs of using a general equilibrium or multimarket approach far outweigh
the additional precision in the results.
The two separate sub-models in the EIM, for gasoline and PFCs, are not linked (there is
no feedback mechanism between them). This approach is appropriate because these sectors
represent different aspects of fuel consumption (fuel storage and fuel production), and
production and consumption of one is not affected by the other. In other words, an increase in
the price of PFCs is not expected to have an impact on the production and supply of gasoline,
E Market impacts were not modeled for the vehicle market; see Section 13.1.3, above.
13-16
-------
Final Regulatory Impact Analysis
and vice versa. Production and consumption of each of these products are the result of other
factors that have little cross-over impacts (the need for fuel storage; the need for personal
transportation).
13.2.2.2 Perfect Competition Model
For all markets that are modeled, the analyst must characterize the degree of competition
within each market. The discussion generally focuses on perfect competition (price-taking
behavior) versus imperfect competition (the lack of price-taking behavior). It should be noted
that the perfect competition assumption is not primarily about the number of firms in a market.
It is about how the market operates: whether or not individual firms have sufficient market
power to influence the market price. Indicators that allow us to assume perfect competition
include absence of barriers to entry, absence of strategic behavior among firms in the market,
and product differentiation.
This EIM relies on an assumption of perfect competition. This means that consumers and
firms are price takers and do not have the ability to influence market prices.
In a perfectly competitive market at equilibrium the market price equals the value society
(consumers) places on the marginal product, as well as the marginal cost to society (producers).
Producers are price takers, in that they respond to the value that consumers put on the product. It
should be noted that the perfect competition assumption relies not only on the number of firms in
a market but also on other market characteristics such as absence of barriers to entry and
strategic behavior among firms in the market, and the lack of product differentiation.
In contrast, imperfect competition implies firms have some ability to influence the market
price of output they produce. One of the classic reasons firms may be able to do this is their
ability to produce commodities with unique attributes that differentiate them from competitors'
products. This allows them to limit supply, which in turn increases the market price, given the
traditional downward-sloping demand curve. Decreasing the quantity produced increases the
monopolist's profits but decreases total social surplus because a less than optimal amount of the
product is being consumed. In the monopolistic equilibrium, the value society (consumers)
places on the marginal product, the market price, exceeds the marginal cost to society
(producers) of producing the last unit. Thus, social welfare would be increased by inducing the
monopolist to increase production. Social cost estimates associated with a regulation are larger
with monopolistic market structures and other forms of imperfect competition because the
regulation exacerbates the existing social inefficiency of too little output from a social
perspective. The Office of Management and Budget (OMB) explicitly mentions the need to
consider these market power-related welfare costs in evaluating regulations under Executive
Order 12866.9
Perfect competition is a widely accepted economic practice for this type of analysis and
only in rare cases are other approaches used.10 For the markets affected by this rule, the perfect
competition assumption is appropriate.
13-17
-------
With regard to the fuel market, the Federal Trade Commission (FTC) has developed an
approach to ensure competitiveness in this sector. The FTC reviews oil company mergers and
frequently requires divestiture of refineries, terminals, and gas stations to maintain a minimum
level of competition. This is discussed in more detail in the industry profile prepared for this
rule.11 Therefore, it is reasonable to assume a competitive market structure in this analysis.
With regard to the PFC market, the small number of firms in the market is offset by
several features of this market. Because PFCs are compact and lightweight, they are easy to
transport far from their place of manufacture. This means that production is not limited to local
producers. Although they vary by size and material, consumers are likely to view all PFCs
designed for storing a particular fuel (gasoline, diesel fuel, kerosene) as good substitutes for the
storage of that specific fuel. Because the products are similar enough to be considered
homogeneous (e.g., perfectly substitutable), consumers can shift their purchases from one
manufacturer to another. There are only minimal technical barriers to entry that would prevent
new firms from freely entering the market, since manufacturing is based on well-known plastic
processing methods. In addition, there is significant excess capacity, enabling competitors to
respond quickly to changes in price. Excess production capacity in the general container
manufacturing market also means that manufacturers could potentially switch their product lines
to compete in this segment of the market, often without a significant investment. In addition,
there is no evidence of high levels of strategic behavior in the price and quantity decisions of the
firms. Finally, it should be noted that contestable market theory asserts that oligopolies and even
monopolies will behave very much like firms in a competitive market if manufacturers have
extra production capacity and this capacity could allow them to enter the market costlessly (i.e.,
there are no sunk costs associated with this kind of market entry or exit).F'12'13 As a result of all
of these conditions, producers and consumers in the PFC market are expected to take the market
price as given when making their production and consumption choices and the market can be
modeled as a competitive market even though the number of producers is small. More
information about the structure of the PFC industry organization can be found in Section 3 of the
industry characterization prepared for this rule.14
13.2.3.3 Intermediate-Run Model
In developing partial equilibrium models, the choices available to producers must be
considered. For example, are producers able to increase their factors of production (e.g.,
increase production capacity) or alter their production mix (e.g., substitution between materials,
labor, and capital)? These modeling issues are largely dependent on the time horizon for which
the analysis is performed. Three benchmark time horizons are discussed below: the very short
run, the long run, and the intermediate run. This discussion relies in large part on the material
contained in the OAQPS Economic Analysis Resource Guide.15
F A monopoly or firms in oligopoly may not behave as neoclassical economic theories of the firm predict because
such firms may be concerned about new entrants to the market. If super-normal profits are earned, potential
competitors may enter the market. To respond to this threat, existing firm(s) in the market may keep prices and
output at a level where only normal profits are made, setting price and output levels at or close to the competitive
price and output.
13-18
-------
Final Regulatory Impact Analysis
The EIM models market impacts in the intermediate run. The use of the intermediate run
means that some factors of production are fixed and some are variable. This modeling period
allows analysis of the economic effects of the rule's compliance costs on current producers. As
described below, a short-run analysis imposes all compliance costs on producers, while a long-
run analysis imposes all costs on consumers. The use of the intermediate time frame is
consistent with economic practices for this type of analysis.
In the very short run, all factors of production are assumed to be fixed, leaving the
directly affected entity with no means to respond to increased costs associated with the
regulation (e.g., they cannot adjust labor or capital inputs). Within a very short time horizon,
regulated producers are constrained in their ability to adjust inputs or outputs due to contractual,
institutional, or other factors and can be represented by a vertical supply curve, as shown in
Figure 13.2-1. In essence, this is equivalent to the nonbehavioral model described earlier.
Neither the price nor quantity changes and the manufacturer's compliance costs become fixed or
sunk costs. Under this time horizon, the impacts of the regulation fall entirely on the regulated
entity. Producers incur the entire regulatory burden as a one-to-one reduction in their profit.
This is referred to as the "full-cost absorption" scenario and is equivalent to the engineering cost
estimates. Although there is no hard and fast rule for determining what length of time constitutes
the very short run, it is inappropriate to use that time horizon for this analysis because the very
short run assumes economic entities have no flexibility to adjust factors of production.
Price
Q Output
Figure 13.2-1. Short Run: All Costs Borne by Producers
13-19
-------
In the long run, all factors of production are variable, and producers can be expected to
adjust production plans in response to cost changes imposed by a regulation (e.g., using a
different labor/capital mix). Figure 13.2-2 illustrates a typical, if somewhat simplified, long-run
industry supply function. The function is horizontal, indicating that the marginal and average
costs of production are constant with respect to output.0 This horizontal slope reflects the fact
that, under long-run constant returns to scale, technology and input prices ultimately determine
the market price, not the level of output in the market.
Market demand is represented by the standard downward-sloping curve. The market is
assumed here to be perfectly competitive; equilibrium is determined by the intersection of the
supply and demand curves. In this case, the upward shift in the market supply curve represents
the regulation's effect on production costs. The shift causes the market price to increase by the
full amount of the per-unit control cost (i.e., from P to P'). With the quantity demanded sensitive
to price, the increase in market price leads to a reduction in output in the new with-regulation
equilibrium (i.e., Q to Q'). As a result, consumers incur the entire regulatory burden as
represented by the loss in consumer surplus (i.e., the area P ac P'). In the nomenclature of EIAs,
this long-run scenario is typically referred to as "full-cost pass-through" and is illustrated in
Figure 13.2-2.
Price /
Increase \
With Regulation
; Without Regulation
Output
Figure 13.2-2. Long Run: Full Cost Pass Through
G The constancy of marginal costs reflects an underlying assumption of constant returns to scale of production,
which may or may not apply in all cases.
13-20
-------
Final Regulatory Impact Analysis
Taken together, impacts modeled under the long-run/full-cost-pass-through scenario
reveal an important point: under fairly general economic conditions, a regulation's impact on
producers is transitory. Ultimately, the costs are passed on to consumers in the form of higher
prices. However, this does not mean that the impacts of a regulation will have no impact on
producers of goods and services affected by a regulation. For example, the long run may cover
the time taken to retire all of a facility's existing capital, which could take decades. Therefore,
transitory impacts could be protracted and could dominate long-run impacts in terms of present
value. In addition, to evaluate impacts on current producers, the long-run approach is not
appropriate. Consequently a time horizon that falls between the very short-run/full-cost-
absorption case and the long-run/full-cost-pass-through case is most appropriate for this EIA.
The intermediate run time frame allows examination of impacts of a regulatory program
during the transition between the short run and the long run. In the intermediate run, some
factors are fixed; some are variable.11 In other words, producers can adjust some, but not all,
factors of production, meaning they will bear some portion of the costs of the regulatory
program. The existence of fixed production factors generally leads to diminishing returns to
those fixed factors. This typically manifests itself in the form of a marginal cost (supply)
function that rises with the output rate, as shown in Figure 13.2-3.
H As a semantical matter, the situation where some factors are variable and some are fixed is often referred to as the
"short run" in economics, but the term "intermediate run" is used here to avoid any confusion with the term "very
short run."
13-21
-------
Price
Increase
. With Regulation
Unit Cost Increase
Without Regulation
Output
Figure 13.2-3. Intermediate Run: Partial Cost Pass Through
Again, the regulation causes an upward shift in the supply function. The lack of resource
mobility may cause producers to suffer profit (producer surplus) losses in the face of regulation;
however, producers are able to pass through some of the associated costs to consumers, to the
extent the market will allow. As shown, in this case, the market-clearing process generates an
increase in price (from P to P') that is less than the per-unit increase in costs, so that the
regulatory burden is shared by producers (net reduction in profits) and consumers (rise in price).
In other words, there is a loss of both producer and consumer surplus.
Consistent with other economic impact analyses performed by EPA, this EIM uses an
intermediate run approach. This approach allows us to examine the market and social welfare
impacts of the program as producers adjust their output and consumers adjust their consumption
of affected products in response to the increased production costs. During this period, the
distribution of the welfare losses between producer and consumer depends in large part on the
relative supply and demand elasticity parameters used in the model. For example, if demand for
PFCs is relatively inelastic (i.e., demand does not decrease much as price increases), then most
of the direct compliance cost on refiners will be passed along to PFC consumers in the form of
higher prices.
13.2.3 How is the EIM Used to Estimate Economic Impacts?
13.2.3.1 Estimation of Market Impacts
A graphical representation of a general economic competitive model of price formation,
as shown in Figure 13.2-4, posits that market prices and quantities are determined by the
intersection of the market supply and market demand curves. Under the baseline scenario, a
market price and quantity (p,Q) combination is determined by the intersection of the downward-
13-22
-------
Final Regulatory Impact Analysis
sloping market demand curve (DM) and the upward-sloping market supply curve (SM). The
market supply curve reflects the sum of the domestic (S
-------
market demand curve) declines from Q to Q'. This reduction in market output is the net result of
reductions in domestic and import supply
As indicated in Figure 13.2-3, when the standards are applied the supply curve will shift
upward by the amount of the estimated compliance costs. The demand curve, however, does not
shift. This is because a shift in the demand curve is determined by changes in factors such as
income, tastes, prices of substitute and complementary goods, expectations, and population. The
standards do not affect these factors and so it is appropriate to assume all these factors remain
constant.
13.2.3.2 Estimation of Social Costs
The economic welfare implications of the market price and output changes with the
regulation can be examined by calculating consumer and producer net "surplus" changes
associated with these adjustments. This is a measure of the negative impact of an environmental
policy change and is commonly referred to as the "social cost" of a regulation. It is important to
emphasize that this measure does not include the benefits that occur outside of the market, that
is, the value of the reduced levels of air pollution with the regulation. Including this benefit will
reduce the net cost of the regulation and even make it positive.
13-24
-------
Final Regulatory Impact Analysis
$/Q
I I
Q2 Q1
(a) Change in Consumer Surplus with
Regulation
Q/t
$/Q
Q2 Q,
(b) Change in Producer Surplus with
Regulation
Q/t
$/Q
Q2 Q,
(c) Net Change in Economic Welfare with
Regulation
Q/t
Figure 13.2-5. Market Surplus Changes with Regulations
Consumer and Producer Surplus
13-25
-------
The demand and supply curves that are used to project market price and quantity impacts
can be used to estimate the change in consumer, producer, and total surplus or social cost of the
regulation (see Figure 13.2-5a).
The difference between the maximum price consumers are willing to pay for a good and
the price they actually pay is referred to as "consumer surplus." Consumer surplus is measured
as the area under the demand curve and above the price of the product. Similarly, the difference
between the minimum price producers are willing to accept for a good and the price they actually
receive is referred to as "producer surplus." Producer surplus is measured as the area above the
supply curve below the price of the product. These areas can be thought of as consumers' net
benefits of consumption and producers' net benefits of production, respectively.
In Figure 13.2-5, baseline equilibrium occurs at the intersection of the demand curve, D,
and supply curve, S. Price is PI with quantity Qi. The increased cost of production with the
regulation will cause the market supply curve to shift upward to S'. The new equilibrium price
of the product is P2. With a higher price for the product there is less consumer welfare, all else
being unchanged. In Figure 13.2-5a, area A represents the dollar value of the annual net loss in
consumers' welfare associated with the increased price. The rectangular portion represents the
loss in consumer surplus on the quantity still consumed due to the price increase, Q2, while the
triangular area represents the foregone surplus resulting from the reduced quantity consumed, Qi
-Q2.
In addition to the changes in consumers' welfare, there are also changes in producers'
welfare with the regulatory action. With the increase in market price, producers receive higher
revenues on the quantity still purchased, Q2. In Figure 13.2-5b, area B represents the increase in
revenues due to this increase in price. The difference in the area under the supply curve up to the
original market price, area C, measures the loss in producer surplus, which includes the loss
associated with the quantity no longer produced. The net change in producers' welfare is
represented by area B - C.
The change in economic welfare attributable to the compliance costs of the regulations is
the sum of consumer and producer surplus changes, that is, -(A) + (B-C). Figure 13.2-5c shows
the net (negative) change in economic welfare associated with the regulation as area D.
As explained in Section 13.1.3, the vehicle market is not included in the EIM. Instead,
compliance costs are used as a proxy for the social welfare costs associated with that part of the
regulatory program. Vehicle compliance costs are likely to be absorbed by the manufacturers,
thus increasing their surplus loss.
13.2.4. How Are Special Market Characteristics Addressed?
In addition to the general model features described in Section 13.2.2, there are several
specific characteristics of the PFC and gasoline fuel markets that need to be addressed in the
EIM. These are the treatment of gasoline fuel savings, fixed and variable costs, flexibility
provisions, and substitution.
13-26
-------
Final Regulatory Impact Analysis
13.2.4.1 Fixed and Variable Costs
Related to short-run versus long-run modeling issues is the question of how fixed and
variable costs are defined or treated by a specific industry or in the market analysis. The
engineering estimates of fixed R&D and capital costs and variable material and operating and
maintenance (O&M) costs provide an initial measure of total annual compliance costs without
accounting for behavioral responses. The starting point for assessing the market impacts of a
regulatory action is to incorporate the regulatory compliance costs into the production decision
of the firm.
In general, shifting the supply curve by the total cost per unit implies that both capital and
operating costs vary with output levels. At least in the case of capital, this raises some questions.
In the long run, all inputs (and their costs) can be expected to vary with output. But a short(er)-
run analysis typically holds some capital factors fixed. For instance, to the extent that a market
supply function is tied to existing facilities, there is an element of fixed capital (or one-time
R&D). As indicated above, the current market supply function might reflect these fixed factors
with an upward slope. As shown in Figure 13.2-6, the marginal cost (MC) curve will only be
affected, or shift upwards, by the per-unit variable compliance costs (ci=TVCC/q), while the
average total cost (ATAC) curve will shift up by the per-unit total compliance costs (c2=TCC/q).
Thus, the variable costs will directly affect the production decision (optimal output rate), and the
fixed costs will affect the closure decision by establishing a new higher reservation price for the
firm (i.e., Pm). In other words, the fixed costs are important in determining whether the firm will
stay in this line of business (i.e., produce anything at all), and the variable costs determine the
level (quantity) of production.
13-27
-------
$/q
MC'
prri
pm
(a) Upward-sloping supply function
Figure 13.2-6. Modeling Fixed Costs
Depending on the industry type, fixed costs associated with complying with a new
regulation are generally treated differently in an analysis of market impacts. In a competitive
market, the industry supply curve is generally based on the market's marginal cost curve; fixed
costs do not influence production decisions at the margin. Therefore, the market analysis is
based on variable costs only. This is the case with the vehicle controls in this analysis. The
compliance costs for that program are fixed costs (R&D, test facilities) and do not affect
marginal costs. As a result, this economic impact analysis does not include market impacts for
the vehicle market. They are included in the social welfare analysis, however, since these
compliance costs are a cost to society. By adding the vehicle program compliance costs to the
social welfare costs we attribute all of the costs to the producers and assume that these costs do
not change the quantities of affected vehicles produced or their prices.
The market analysis of the PFC market, however, is different and is based on total
compliance costs (fixed + variable). The approach is appropriate even though this is a
competitive market due to the nature of production practices in this market. Specifically, PFC
manufacturers produce a product that changes very little over time. Portable fuel containers are a
fairly standard product and these manufacturers do not engage in research and development to
improve their products on a continuous basis as is the case with highway vehicles or nonroad
engines or equipment. A design change of nature that would be required by the standards will
require PFC manufacturers to devote new funds and resources to product redesign and facilities
13-28
-------
Final Regulatory Impact Analysis
changes. Portable fuel container manufacturers are expected to increase their prices by the full
amount of the compliance costs to recover those costs.
Fixed costs required to comply with the rule on the refiner side are also treated
differently, to reflect the refinery industry cost structure. Most of the petroleum refinery fixed
costs used are for production hardware. The decision to invest to increase, maintain, or decrease
production capacity may be made in response to anticipated or actual changes in price. To
reflect the different ways in which refiners can pass costs through to consumers, three scenarios
were run for the following supply curve shifts in the gasoline fuel markets:
shift by average total (variable + fixed cost)
shift by max total (variable + fixed cost)
• shift by max variable cost.
While it may seem reasonable to estimate costs based on maximum variable or maximum
total costs, it should be noted that both of those scenarios assume that refiners with the highest
benzene compliance costs are also the highest-cost gasoline producers absent benzene control.
We do not have information on the highest gasoline cost producers to be able to examine
whether these refineries are also expected to have the highest benzene control costs. However,
we believe this is an extreme assumption.
We estimate the market and social welfare impacts of each of these scenarios.
The first, shift by average total cost (variable + fixed), is the primary scenario and is included in
the primary analysis. The other two are investigated in the sensitivity analyses in Appendix G.
13.2.4.2 Gasoline Fuel Savings and Fuel Taxes
If all the costs of the regulation are not reflected in the supply shift, then the producer and
consumer surplus changes reflected in Figure 13.2-5a will not capture the total social costs of the
regulation. This will be the case, for example, if there are cost savings attributable to a program
that are not readily apparent to consumers. In this case, the PFC controls are expected to reduce
evaporative emissions from gasoline fuel storage, resulting in gasoline fuel savings for users of
these containers. These fuel savings are not included in the market analysis for this EIA because
these savings are not expected to affect consumer decisions with respect to the purchase of new
containers. In other words, we assume people base their decision on whether to buy a new
container on other needs (e.g., purchase of new equipment, replacement of a damaged container)
and not on expected fuel savings that would accrue to them from using a compliant container.
Gasoline fuel savings will be included in the social cost analysis, however, because they are a
savings that accrues to society. They will be added into the estimated social costs as a separate
line item.
The estimated gasoline fuel savings are estimated using the quantity of gasoline fuel
saved through better evaporative controls and the post-tax price of gasoline (see Section
13.3.2.2). The post-tax price is used because this is the price consumers see at the fuel pump and
is the price on which they base their purchasing decisions. In other words, consumers save the
13-29
-------
entire amount of the pump price. Also, in contrast to distillate diesel fuel used in nonroad
equipment, gasoline fuel taxes are not typically rebated. This is because most gasoline fuel used
in nonroad equipment is used by residential consumers and even those who could file for a tax
rebate probably don't given the small amounts of fuel involved. As a result, the consumer would
realize a savings equal to the pump price of gasoline for the gasoline fuel they save from
evaporative controls (i.e., the full cost of the fuel and not just the pre-tax cost). At the same time,
the tax savings realized on the fuel savings by consumers are reduced taxes revenues for local
and federal governments. These revenue losses are estimated separately in the social welfare
analysis, based on the gallons of gasoline fuel saved and the average national fuel tax (combined
state and Federal government).
13.2.4.3 Flexibility Provisions
Consistent with the engineering cost estimates, the EIM does not include cost savings
associated with compliance flexibility provisions or averaging, banking, and trading provisions.
As a result, the results of this EIA can be viewed as somewhat conservative.
13.2.4.4 Substitution
This analysis assumes that there will be no substitution away from gasoline fuel. As
explained in Section 13.2.3.3, the time horizon for this analysis is the intermediate run. In the
intermediate run, economic actors can adjust some of their costs but others are fixed. So, for
example, consumers can adjust the amount of gasoline they purchase but the type of vehicle or
equipment they own (i.e., gasoline or diesel) is fixed. This analysis assumes that the relative
proportions of gasoline to diesel vehicles and equipment are constant for the period of analysis.
This assumption seems reasonable because the average cost increase for gasoline is estimated to
be less than $0.01 per gallon. Gasoline prices vary considerably over time without provoking
dramatic shifts in consumer behavior. Therefore, our assumption that consumers will not
substitute away from gasoline vehicles and equipment in favor of diesels, or otherwise modify
their behavior, is reasonable.
The analysis also assumes there will be no substitution away from PFCs. Consumers
seeking to store a particular kind of fuel (gasoline, diesel, or kerosene) have only limited
alternatives for safely storing that fuel: metal or plastic fuel containers approved for storage of
that particular kind of fuel. Plastic containers account for the vast majority of PFCs sold due to
their safety characteristics and ease of use. They are light-weight, are very durable, and do not
rust. Plastic containers are also cheaper to manufacturer than their metal counterparts.
Consequently, about 95 percent of the PFCs sold in the United States are plastic. While it may
be the case that some consumers opt to use unapproved containers (e.g., milk jugs, glass jars),
the extent to which they do this is not known. This rule will make approved plastic PFCs more
expensive compared to unapproved containers, but we do not expect this rule to lead to more use
of inappropriate containers by consumers than is already the case. Unapproved containers have
serious defects. For example, it is difficult to pour fuel from containers such as plastic milk jugs,
glass jars, and similar containers, especially into the small mouths of some lawn and garden
equipment. In addition, these also are not long-term storage options as they may be damaged by
the fuel. Consumers are generally aware that fuel must be transported and stored safely and are
13-30
-------
Final Regulatory Impact Analysis
not likely to view these alternatives as safe relative to an approved fuel storage container.
Finally, it is illegal in most if not all states to dispense fuel into unapproved containers, with this
prohibition clearly marked on fuel pumps.
The elasticity of demand for PFCs estimated for this EIM reflects this no-substitution
assumption. As noted in Section 13.1.3 and explained in more detail in Section 13.3.5 and in
Appendix E, this estimated elasticity is inelastic at -0.01. This means that a 100 percent increase
in price is expected to result in a 1 percent decrease in demand. In acknowledgement of the
concern about use of inappropriate containers, we also performed a sensitivity analysis for the
elasticity of demand estimate relaxing the no-substitution assumption and using a rate of
substitution of 10 percent. This is a fairly high rate of substitution and means that 10 percent of
people who would otherwise buy a PFC find some other way to store fuel (e.g., inappropriate
containers) or opt not to purchase a PFC (for example, those with multiple containers will choose
not to replace a container, giving up having multiple cans in multiple locations or the capability
of filling multiple cans with a single trip to the gas station). Using a 10 percent rate of
substitution we estimate a demand elasticity that is less inelastic, at -0.25. This means that a 100
percent increase in price results in a 25 percent decrease in demand. As described in Appendix
G, this alternative demand elasticity has only a small impact on the results of the modeling. For
2015, the price impact is reduced by about 20 cents (decreasing from $1.52 to $1.31 in states that
do not already have PFC requirements). In addition, producers are expected to bear more of the
costs of the program (increasing from 0.7 percent to 15.1 percent). The emissions impacts of a
10 percent rate of substitution are small. If these purchasers exit the PFC market permanently
(i.e., this is not a short-term adjustment with consumers only postponing their purchases), we
would expect about 10 percent less emissions reductions from the PFC standards. Table 13.2-1
below provides an example of potential losses in VOC emission reductions from a ten percent
substitution rate. It is important to note that the costs of the overall program would also be
reduced by roughly the same 10 percent and so the overall cost per ton of emissions reduced
would not significantly change. Also, in cases where the substitution occurs from consumers
keeping their current PFCs for a longer period of time or by only leaving the market temporarily,
the emissions reductions are only postponed to a future date. Therefore, the lost emissions
reductions shown in the table below would represent a worst case for the 10 percent substitution
scenario.
Table 13.2-1 - VOC Emissions Reductions from Portable Fuel Containers (tons)
Base Case
w/ 10 Percent Substitution
Difference
2015
181,000
163,000
18,000
2020
193,000
174,000
19,000
2030
218,000
196,000
22,000
13.2.4.5 Market-Level Analysis
The EIM estimates the economic impacts of the rule at the market level. It is not a firm-
level analysis. The demand elasticity facing any particular manufacturer may be different from
the demand elasticity of the market as a whole, and therefore the share of the compliance costs a
13-31
-------
particular firm may pass on to consumers may be smaller or larger than estimated by this model.
This difference can be important, particularly where the rule affects different firms' costs over
different volumes of production. However, to the extent that there are differential effects, EPA
believes that the flexibilities provided in this rule will be adequate to address any cost inequities
that are likely to arise.
13.3 EIM Data Inputs and Model Solution
The EIM is a computer model comprised of a series of spreadsheet modules that simulate
the supply and demand characteristics of the affected markets. The model equations, presented
in Appendix D to this chapter, are based on the economic relationships described in Section 13.2.
The EIM analysis consists of four basic steps:
• Define the initial market equilibrium conditions of the markets affected by this rule
(equilibrium prices and quantities and behavioral parameters; these yield equilibrium
supply and demand curves).
• Introduce a policy "shock" into the model based on estimated compliance costs that shift
the supply functions.
• Use a solution algorithm to estimate a new, with-regulation equilibrium price and
quantity for all markets.
• Estimate the change in producer and consumer surplus in all markets included in the
model.
Supply responses and market adjustments can be conceptualized as an interactive
process. Producers facing increased production costs due to compliance are willing to supply
smaller quantities at the baseline price. This reduction in market supply leads to an increase in
the market price that all producers and consumers face, which leads to further responses by
producers and consumers and thus new market prices, and so on. The new with-regulation
equilibrium reflects the new market prices where total market supply equals market demand.
The remainder of this section describes the data used to construct the EIM: initial
equilibrium market conditions (equilibrium prices and quantities), compliance cost inputs, model
elasticity parameters. Also included is a brief discussion of the analytical expression used to
estimate with-regulation market conditions.
13.3.1 Description of Product Markets
There are six product markets included in this EIM: two PFC markets and four gasoline
fuel markets. While the vehicle market will also be affected by the standards, that market was
not included in the EIM (see Section 13.1.3). Each of these markets is described below. More
information can be found in the industry characterizations prepared for this rule.16'17
13.3.1.1 Portable Fuel Container Market
Portable fuel containers allow people to refuel equipment in circumstances where
refueling at a retail fuel establishment or central fueling location is not convenient. Gasoline
13-32
-------
Final Regulatory Impact Analysis
storage containers support the use of a wide variety of gasoline-powered equipment ranging from
lawnmowers, chainsaws, string trimmers, and garden tractors to all-terrain vehicles, off-road
motorcycles, and gasoline-powered golf carts. They are also used for emergency gasoline
supplies for highway vehicles. Diesel storage containers support equipment used on construction
sites, manufacturing facilities, and agricultural establishments. Kerosene storage containers also
support a range of construction, manufacturing, and agricultural equipment.
There is little additional publicly available national data on the users of PFCs. However,
a recent study by CARB found that 94 percent of portable fuel containers in California were used
by residential households.18 Commercial businesses account for a remaining PFC use. Industry
representatives have indicated that sales of PFCs are influenced by trends in sales of power
equipment (i.e., lawn and garden) and recreational vehicles. As a result, factors that influence
decisions to purchase these commodities (e.g., changes in the price of equipment, changes in
personal income, population growth rates, home sales) will indirectly influence the decision to
purchase PFCs. Economic theory for derived demand suggests that under some reasonable
assumptions we can predict that an increase in the price of PFCs will have little impact on sales
of PFCs both because PFCs represent a very small fraction of total expenditures and they are an
essential input into household and business production functions.19'20'21 In addition, there are
only limited alternatives for storing gasoline.
The vast majority of PFCs sold in the United States are plastic (about 98 percent).
Portable fuel container manufacturing is currently dominated by four firms (Blitz USA, Midwest
Can, Scepter Manufacturing, Ltd., and Wedco Molded Products) and one firm accounts for about
70 percent of U.S. sales and 50 percent of North American sales. Other PFC manufacturers have
very limited market share, are more geared for industrial use, and/or fill a niche specialty market.
Manufacturing PFCs is not constrained geographically in that these containers are lightweight
and fairly inexpensive to transport to distant markets.
Plastic PFCs are manufactured using well-known plastic processing methods to form
plastic material into gas containers and spouts. The production process combines capital
equipment, labor, and materials to produce portable fuel containers of desired size and technical
standards. Therefore, only minimal technical barriers prevent new firms from freely entering the
market, and there are many manufacturers of plastics and plastic containers who could join the
market if it were profitable to do so.
California established an emissions control program for PFCs that began in 2001.22
Twelve other states (Delaware, Maine, Maryland, Pennsylvania, New York, Connecticut,
Massachusetts, New Jersey, Rhode Island, Vermont, Virginia, and Texas) and the District of
Columbia have adopted the California program in recent years. Because of these existing control
measures, the costs of complying with the standards is expected to be reduced for these states
(fewer changes will be necessary for these PFCs). Consequently, the economic impact analysis
differentiates between two markets: those states that have controls and those that do not.
13.3.1.2 Gasoline Fuel Market
13-33
-------
Gasoline plays an important role in the American economy. The Federal Highway
Administration reported that the United States consumed over 130 billion gallons of gasoline in
2002.23 The overwhelming majority of gasoline is consumed for highway uses. About 92% of
gasoline consumption on a BTU basis was consumed by light-duty vehicles. Most people rely
on gasoline for personal transportation, unlike the commercial transportation that relies mostly
on diesel fuel. The remaining share of gasoline consumption is for non-highway use (i.e., lawn
and garden equipment and marine uses).
Consumers are not expected to be very sensitive to changes in the price of gasoline.
Consumers can respond to price changes in gasoline in two ways. In the short term, they may
simply consider reducing the number of vehicle miles traveled or their use of nonroad
equipment. However, their ability to reduce gasoline consumption in this way depends on their
ability to do without the service provided by the gasoline-consuming vehicle or equipment
(forego lawnmowing or personal transportation). If the relative price of gas remains higher for
longer periods, consumers might also consider long-term adjustments to their capital stock to
mitigate the effects of higher prices. For example, they may purchase vehicles with better fuel
economy, buy a home closer to work or shopping, or purchase nonroad equipment that relies on
electricity. In either case, the price of gasoline may have to rise considerably to trigger such a
change in consumption patterns.
Producers of gasoline are also expected to be insensitive to price changes, for two major
reasons. First, refineries produce finished motor gasoline through a complex process that
converts crude oil into three principal types of hydrocarbon products: gasoline, distillate (i.e., jet
fuel, diesel fuel, and heating oil), and heavy oils (i.e., residual fuel oil, asphalt). A refiner's
ability to alter the proportions of the three products generated by refining crude oil is somewhat
limited. Refiners have more, but not unlimited, flexibility in adjusting production among
different formulations of gasoline. Once a refiner has decided what formulations of gasoline it
will produce in an upcoming production campaign, it becomes increasingly difficult to alter the
planned output of the refinery as the production campaign approaches. Second, refining is a
capital-intensive, high fixed-cost operation. Consequently, refiners attempt to operate at high
capacity utilization rates. Industry statistics illustrate that refining capacity is generally tight, and
capacity utilization has been increasing over the past decade. Industry-wide crude oil refining
capacity utilization in the United States in the month of May was 85 percent in 1990, 89 percent
in 1992, 93 percent in 1994 and 1996, 94 percent in 1998, and 96 percent in 2000. The average
monthly capacity utilization rate in 2000 was 94 percent. These characteristics of the refining
industry limit further the ability of refiners to change refinery production significantly in the
short run.24
There are more than 100 refineries in the United States. Additional gasoline is obtained
through imports, especially on the East Coast. However, production tends to be regional in
nature. The Federal Trade Commission (FTC) has developed an approach to ensure
competitiveness in gasoline fuel markets. It reviews oil company mergers and frequently
requires divestiture of refineries, terminals, and gas stations to maintain a minimum level of
competitiveness.
Finished gasoline product leaves the refinery and reaches consumers through one or more
bulk transport services. Pipelines, tankers, or barges typically transport gasoline from refineries
13-34
-------
Final Regulatory Impact Analysis
or ports to terminals that provide storage and dispensing facilities. A variety of downstream
gasoline marketing arrangements (i.e. wholesale and retail) ultimately deliver gasoline to the
consumer.
Given the existing region-specific gasoline performance standards and other
transportation and economic barriers, this analysis uses the five regional markets (PADDs)
defined by the Department of Energy. For the purpose of this analysis, two PADDs are
combined, giving four regional district fuel markets. These are:
• PADD 1 & 3
• PADD2
• PADD4
• PADD 5 (includes Alaska and Hawaii; California fuel not included).
PADD 1 and 3 are combined because of the high level of regional trade between these
areas. Other regional trading is generally constrained due to inefficiencies in transporting
gasoline between regions and so is not included in this analysis. Also not included in the
analysis is inter-region trading on a consumer basis (drivers who cross state lines to purchase
fuel). PADD 5 does not include California fuel in the market analysis since California already
has fuel benzene controls. Finally, consistent with the cost analysis, the EIM does not
distinguish between conventional gasoline and reformulated gasoline (RFG).
13.3.2 Initial Market Conditions
The starting point for the economic impact analysis is initial market equilibrium
conditions that exist prior to the implementation of new standards. At pre-control market
equilibrium conditions, consumers are willing to purchase the same amount of a product that
producers are willing to produce at the market price. This section describes the initial market
equilibrium conditions (prices and quantities) for the PFC and gasoline markets.
13.3.2.1 Portable Fuel Container Market Quantities and Prices
The PFC market equilibrium sales and price data used in the EIM are contained in Tables
13.3-1 and 13.3-2. The data are based on information provided by industry.25 Industry sales
data from 2002 were grown for future years using a two percent growth rate. This growth rate is
consistent with information obtained from industry representatives, who indicated that sales are
expected to increase at the same pace as the retail market in general. The PFC prices for 2003
were obtained from industry. The prices in Table 3.3-2 are weighted averages of the observed
prices of 3 sizes of PFCs (1 gallon, 2 gallon, and 5 gallon; 33 percent weight for each). PFC
prices are held fixed for all years included in the analysis reflecting an assumption of constant
(real) price of goods and services over time (see Appendix F for an explanation of this
assumption).
Table 13.3-1. Portable Fuel Container Sales Data (2009 to 2035)
13-35
-------
Year
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
States without
Controls
18,218,155
18,582,518
18,954,168
19,333,252
19,719,917
20,114,315
20,516,601
20,926,933
21,345,472
21,772,381
22,207,829
22,651,985
23,105,025
23,567,126
24,038,468
24,519,238
25,009,622
25,509,815
26,020,011
26,540,411
27,071,220
27,612,644
28,164,897
28,728,195
29,302,759
29,888,814
30,486,590
States With
Controls
11,647,673
11,880,626
12,118,239
12,360,603
12,607,815
12,859,972
13,117,171
13,379,515
13,647,105
13,920,047
14,198,448
14,482,417
14,772,065
15,067,507
15,368,857
15,676,234
15,989,759
16,309,554
16,635,745
16,968,460
17,307,829
17,653,985
18,007,065
18,367,206
18,734,551
19,109,242
19,491,426
Total
29,865,827
30,463,144
31,072,407
31,693,855
32,327,732
32,974,287
33,633,772
34,306,448
34,992,577
35,692,428
36,406,277
37,134,402
37,877,090
38,634,632
39,407,325
40,195,471
40,999,381
41,819,368
42,655,756
43,508,871
44,379,048
45,266,629
46,171,962
47,095,401
48,037,309
48,998,055
49,978,017
Table 13.3-2. Portable Fuel Container Price Data (2003$)
States Without Controls
$4.66
States With Controls
$11.05
13.3.2.2 Gasoline Fuel Market Quantities and Prices
The gasoline fuel market equilibrium sales and price data used in the EIM are contained
in Tables 13.3-3 and 13.3-4. It should be noted that the sales data is for all gasoline and that this
analysis does not differentiate between reformulated and conventional gasoline. This is
13-36
-------
Final Regulatory Impact Analysis
consistent with the cost analysis performed for this rule.1 California gasoline is not included in
this program as that state has its own benzene control program.
The sales data is Energy Information Administration data, based on the Energy
Information Administration's Petroleum Market Annual fuel consumption data (Table 48) for
2004.26 This data was adjusted using the growth rates from the Energy Information
Administration's Annual Energy Outlook 2006 (with 2030 to 2035 growth based on 2025 to
2030 growth estimated by EIA).27 The gasoline volumes used in this economic impact analysis
are consumption volumes, which include imported gasoline as well as gasoline produced in the
United States for domestic purposes. Consumption volumes are used because the market
equilibrium price is determined by all the gasoline supplied and purchased in the market and not
just the gasoline produced in the U.S. for that market.
Gasoline retail prices were estimated using the following approach.28 First, the average
price of motor gasoline by PADD (all grades, sales to end users, excluding taxes) was obtained
from the Energy Information Administrations 2003 Petroleum Marketing Annual.29 Next, state
and federal motor gasoline taxes data were obtained from the Department of Transportation's
2003 Highway Statistics to create an average state tax per model region.30 State and federal
taxes were added to the price data obtained from the Energy Information Administration. Since
ELM model combines PADDs 1 and 3, the retail price for this market is an average price for the
region. Each PADD's price is weighted by the gasoline consumption data used in the market
model.
1 See Note B, above.
13-37
-------
Table 13.3-3. Gasoline Fuel Sales Data, by Region (2007 to 2035; MM gallons)
Year
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
PADD 1 & 3
72,263
73,414
74,794
76,252
77,479
78,553
79,551
80,548
81,545
82,542
83,540
84,614
85,611
86,531
87,452
88,296
89,140
90,060
90,981
91,978
92,975
93,896
94,893
95,814
96,810
97,818
98,836
99,864
100,903
PADD 2
40,154
40,793
41,560
42,370
43,052
43,649
44,203
44,757
45,311
45,866
46,420
47,016
47,571
48,082
48,594
49,063
49,531
50,043
50,554
51,109
51,663
52,174
52,728
53,240
53,794
54,354
54,919
55,491
56,068
PADD 4
4,702
4,777
4,867
4,962
5,042
5,111
5,176
5,241
5,306
5,371
5,436
5,506
5,571
5,631
5,691
5,745
5,800
5,860
5,920
5,985
6,050
6,110
6,175
6,235
6,299
6,365
6,431
6,498
6,566
PADD 5
w/out CA
8,197
8,327
8,484
8,649
8,788
8,910
9,023
9,137
9,250
9,363
9,476
9,598
9,711
9,815
9,920
10,015
10,111
10,215
10,320
10,433
10,546
10,651
10,764
10,868
10,981
11,095
11,211
11,328
11,445
Total
125,315
127,311
129,705
132,233
134,362
136,224
137,953
139,683
141,412
143,142
144,871
146,733
148,463
150,059
151,656
153,119
154,582
156,179
157,775
159,504
161,234
162,830
164,560
166,156
167,885
169,632
171,397
173,180
174,982
Table 13.3-4. Gasoline Fuel Prices, by Region (2003$; includes fuel taxes)
PADD 1 & 3
$1.48
PADD 2
$1.51
PADD 4
$1.57
PADD 5
w/out CA
$1.66
13-38
-------
Final Regulatory Impact Analysis
Gasoline fuel prices are held fixed for all years included in the analysis reflecting an
assumption of constant (real) price of goods and services over time (see Appendix F for an
explanation of this assumption). We also performed a sensitivity analysis using gasoline fuel
prices projected by the Energy Information Agency. The results of that sensitivity analysis can
be found in Appendix G.
13.3.3 Compliance Costs
The social costs of the standards are estimated by shocking the initial market equilibrium
conditions by the amount of the compliance costs. The compliance costs used in this analysis are
the engineering compliance costs described in Chapters 9 and 10 of this RIA and are summarized
in this section.
13.3.3.1 Portable Fuel Container Compliance Costs
The economic impacts of the PFC controls are estimated based on the estimated
engineering compliance costs described in Chapter 10. The compliance costs used in the EIA are
summarized in Table 13.3-5. The compliance costs begin to apply in 2009, when the program
goes into effect.
Even though this is a competitive market, the PFC market is shocked by the sum of the
fixed and variable compliance costs in the initial years of the program. The fixed costs are
included for the first five years of the program, which represents the capital recovery period for
the initial R&D and tooling costs. As explained in Section 13.2.4.1, in a competitive market the
industry supply curve is based on its marginal cost curve and therefore the market shock should
reflect only variable costs. However, as explained in that section, PFC manufacturing sector is
structured such that these manufacturers are expected to pass along the full amount of the
compliance costs, fixed and variable costs, to consumers in the form of higher prices.
In the engineering cost analysis, fixed costs are applied equally over the five-year
recovery period. For the purpose of the EIA, a simplified constant fixed cost approach was used
to allocate the fixed costs to a per-unit basis. Because the number of units produced is expected
to increase every year, this approach means that the model anticipates that engine manufacturers
would recover slightly more than the estimated fixed costs, and the supply curve shift would be
slightly more than of another method of allocating fixed costs were used. While the resulting
estimated social welfare costs of the program are slightly higher, this difference is not expected
to change the overall results of the analysis.
As reflected in Table 13.3-5, variable and fixed costs are different for PFCs in states with
or without existing controls. The estimated costs are expected to be less in states with existing
programs because manufacturers will incur fewer costs to bring their PFCs into compliance with
the standards.
13-39
-------
Table 13.3-5. Portable Fuel Container Compliance Costs (Per Unit; 2003$)
Year
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
States without State Program
Fixed
Costs
$1.17
$1.17
$1.17
$1.17
$1.17
Variable
Costs
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
Total Costs
$2.70
$2.70
$2.70
$2.70
$2.70
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
States with State Program
Fixed
Costs
$0.56
$0.56
$0.56
$0.56
$0.56
Variable
Costs
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
Total
Costs
$0.77
$0.77
$0.77
$0.77
$0.77
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
13.3.3.2 Gasoline Fuel Compliance Costs
The EIM uses the estimated gasoline fuel compliance costs described in Chapter 9. The
compliance costs for the primary scenario, average total (fixed + variable) costs, are summarized
in Table 13.3-6. The gasoline compliance costs are different across regions, reflecting different
13-40
-------
Final Regulatory Impact Analysis
refinery production practices. The compliance costs for PADD 1&3 is a weighted average of the
compliance costs for each of those two PADDs. Compliance costs are treated the same for
domestically produced fuel and imports for each PADD. This approach is reasonable because
many areas (e.g., Europe, Japan, and Australia) already have benzene standards. In addition,
although foreign refiners may face a compliance situation different from domestic producers in a
particular PADD, they can select fuel streams for export that require less benzene removal,
thereby keeping their costs low.
The compliance costs contained in Table 13.3-6 reflect a phase-in of the program starting
in 2007 and ending in May 2015. After the phase-in, gasoline fuel compliance costs are constant
for all years and each regional supply curve is shifted by the average total (variable + fixed)
regional cost of the regulation. This approach is used for the fuel market because most of the
petroleum refinery fixed costs are used for production hardware which is required by the
standards. This new capital investment (fixed costs) will be amortized each year and will be
replaced after a certain period. Therefore, the fixed costs required by this rule are expected to be
constant for all years included in the analysis.
As explained in Section 13.2.4.1, above, we investigate two other gasoline fuel
compliance cost scenarios. In the primary analysis, fuel compliance costs are based on the total
average compliance costs for the industry. However, if refiners' investment in benzene control
capacity is very close to that needed to satisfy the fuel demand for the benzene control program,
then economic theory suggests that the last or highest increment of control in that market would
determine the gasoline price. The compliance costs for each of the two alternative scenarios are
described and the results presented in Appendix G: one in which the high-cost refinery's total
(variable + fixed) compliance costs determine price, and a second in which only the high-cost
refinery's variable compliance costs determine price. It should be noted, however, that both of
these maximum cost scenarios assume that refiners with the highest benzene compliance costs
are also the highest-cost gasoline producers absent benzene control. This is an extreme
assumption.
13-41
-------
Table 13.3-6. Gasoline Fuel Compliance Costs - Total Average (Fixed + Variable) Cost
by Region (I/gallon, 2003$)
Year
2007
2008
2009
2010
2011
2012
2013
2014
2015+
PADD
1 &3
0.0100
0.0160
0.0160
0.0310
0.0310
0.0580
0.0530
0.0530
0.1490
PADD 2
0.0530
0.0910
0.0910
0.1940
0.1940
0.3080
0.2270
0.2270
0.3070
PADD 4
0.0190
0.0330
0.0330
0.0990
0.0990
0.2130
0.2270
0.2270
0.5010
PADD 5
(w/out
California)
0.0040
0.0070
0.0070
0.0350
0.0350
0.1400
0.2440
0.2440
0.9970
13.3.3.3 Vehicle Compliance Costs
The market impacts of the vehicle control program are not modeled because they are
fixed costs (primarily R&D and facility costs) and are therefore not included in the market
analysis (see Section 13.2.4.1, above). However, these compliance costs are costs to society and
should be included in the social cost analysis. We use the vehicle compliance costs as a proxy
for the social welfare costs associated with those controls. These are added to the social costs for
the gasoline fuel and PFC controls to obtain the total social costs of the program.
For this analysis, we used the vehicle compliance costs described in Chapter 8. These are
summarized in Table 13.3-7. These costs are primarily for R&D, tooling, certification, and
facilities. Because these costs are so small on a per vehicle basis, this analysis assumes that they
are expected to be absorbed by the manufacturers.
13-42
-------
Final Regulatory Impact Analysis
Table 13.3-7. Vehicle Compliance Costs (2003$)
Year
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020 and subsequent years
Compliance Costs (SMillion)
$11.1
$11.8
$12.5
$13.3
$13.4
$12.9
$12.2
$11.4
$10.7
$10.6
$0
13.3.4 Gasoline Fuel Savings
As noted in section 13.2.4.1, there are gasoline fuel savings attributable to the PFC
program, reflecting the reduction in evaporative emissions. As explained in that section, these
savings are included in the economic welfare analysis as a separate line item. Consumers of
PFCs will realize an increase in their welfare equivalent to the amount of gallons of gasoline
saved multiplied by the retail price of the gasoline (post-tax price). In the engineering cost
analysis the gasoline fuel savings are estimated in this manner. However, in the context of the
social welfare analysis, some of this increase in consumer welfare is offset by lost tax revenues
to local, state, and federal governments. These welfare losses must be accounted for as well.
Therefore, the net change in social welfare is the difference between the increase in consumer
welfare and the lost tax revenues. This is equivalent to using the pre-tax price of gasoline to
estimate the fuel savings for the social welfare analysis.
The amount of gallons of gasoline fuel saved is estimated based on the VOC inventory
reductions attributable to PFC controls. California fuel is not included in this estimate because
there are no emission reductions attributable to the federal program for that state. Tons of annual
VOC reductions are translated to gallons of gasoline saved using a fuel density of 6 Ibs per
gallon (for lighter hydrocarbons which evaporate first).
Because the gallons of gasoline saved are based on national VOC reductions and were
not estimated by PADD, we estimated a national average retail gasoline price. This estimate is
the sum of the weighted average of pre-tax gasoline prices by PADD and the weighted average
gasoline tax by PADD, using data from the 2003 Petroleum Marketing Annual31 The results of
this analysis are shown in Tables 13.3-8 and 13.3-9.
13-43
-------
Table 13.3-8. Estimated National Average Gasoline Fuel Prices (2003$)
PADD
PADD 1 & 3
PADD 2
PADD 4
PADD 5
Total
Weight
0.58
0.32
0.04
0.06
Pre-tax
Price/Gallon
$1.099
$1.117
$1.165
$1.272
$1.118
Average State
Taxes
$0.201
$0.208
$0.225
$0.200
Federal Tax
$0.184
$0.184
$0.184
$0.184
Post-Tax
Price/Gallon
$1.484
$1.509
$1.574
$1.663
$1.506
Source: 2003 Petroleum Marketing Annual (Table 31). U.S. Department of Energy, Energy Information
Administration, Annual Energy Outlook 2004 with projections to 2025. DOE/EIA-0383 (2004)
From 2009 until 2014 the estimated consumer savings associated with reduced gasoline
consumption from the PFC controls increases sharply, from $15.2 million to $100.3 million.
After 2014 the savings continue to accrue, but at a reduced rate as the PFC population turns over
and fuel savings are due to the continuing benefits of using compliant PFCs. Similarly, the tax
revenue losses are expected to increase from $3.9 million in 2009 to $25.8 million in 2014, but
only $8 million more, to $33.5 million, by 2035.
Table 13.3-9. Estimated Gasoline Fuel Savings From PFC Controls
and Tax Revenue Impacts (2003$)
Year
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
Gallons
10,096,667
20,193,333
31,775,000
43,356,333
54,938,000
66,519,333
67,449,000
68,378,880
69,308,677
70,238,474
71,168,271
72,098,068
73,063,422
74,028,775
74,994,128
75,959,482
76,924,835
77,890,188
Consumer Fuel
Savings
(SMillion)
$15.2
$30.4
$47.9
$65.4
$82.8
$100.3
$101.7
$103.1
$104.5
$105.9
$107.3
$108.7
$110.1
$111.6
$113.1
$114.5
$116.0
$117.4
Tax revenue Impacts
(SMillion)
-$3.9
-$7.8
-$12.3
-$16.8
-$21.3
-$25.8
-$26.2
-$26.5
-$26.9
-$27.3
-$27.6
-$28.0
-$28.4
-$28.7
-$29.1
-$29.5
-$29.9
-$30.2
Net Fuel Savings
(SMillion)
$ 11.3
$ 22.6
$ 35.6
$ 48.5
$ 61.5
$ 74.5
$ 75.5
$ 76.5
$ 77.6
$ 78.6
$ 79.7
$ 80.7
$ 81.8
$ 82.9
$ 83.9
$ 85.0
$ 86.1
$ 87.2
13-44
-------
Final Regulatory Impact Analysis
Year
2027
2028
2029
2030
2031
2032
2033
2034
2035
Gallons
78,855,542
79,820,895
80,786,248
81,751,602
82,681,399
83,611,196
84,540,993
85,470,790
86,400,587
Consumer Fuel
Savings
(SMillion)
$118.9
$120.3
$121.8
$123.2
$124.6
$126.0
$127.4
$128.8
$130.2
Tax revenue Impacts
(SMillion)
-$30.6
-$31.0
-$31.4
-$31.7
-$32.1
-$32.5
-$32.8
-$33.2
-$33.5
Net Fuel Savings
(SMillion)
$ 88.3
$ 89.3
$ 90.4
$ 91.5
$ 92.5
$ 93.6
$ 94.6
$ 95.7
$ 96.7
13.3.5 Supply and Demand Elasticity Estimates
The estimated market impacts and economic welfare costs of this emission control
program are a function of the ways in which producers and consumers of the PFC and gasoline
fuel affected by the standards change their behavior in response to the costs incurred in
complying with the standards. These behavioral responses are incorporated in the EIM through
the price elasticity of supply and demand (reflected in the slope of the supply and demand
curves), which measure the price sensitivity of consumers and producers.
Table 13.3-10 provides a summary of the demand and supply elasticities used to estimate
the economic impact of the rule. More detailed information is provided in Appendix E. The
gasoline elasticities were obtained from the literature. Because we were unable to find published
supply and demand elasticities for the PFC market, we estimated these parameters using the
procedures described in Appendix E. These methods are well-documented and are consistent
with generally accepted econometric practice. It should be noted that these elasticities reflect
intermediate run behavioral changes. In the long run, both supply and demand are expected to be
more elastic.
The price elasticity parameters for gasoline fuel used in this analysis are -0.2 for demand
and 0.2 for supply. This means that both the quantity supplied and demanded are expected to be
fairly insensitive to price changes and that increases in prices are not expected to cause sales to
fall or production to increase by very much. The inelastic supply elasticity for the gasoline fuel
market reflects the fact that most refineries operate near capacity and are therefore less
responsive to fluctuations in market prices. Note that these elasticities reflect intermediate run
behavioral changes. In the long run, both supply and demand are expected to be more elastic
since more substitutes may become available.
13-45
-------
The price elasticity parameters for PFCs used in this analysis are -0.01 for demand) and
1.5 for supply. The estimated demand elasticity is nearly perfectly inelastic (equal to zero). This
means that a change in price is expected to have very little effect on the quantity of PFCs
demanded. This makes intuitive sense since households needing to store gasoline for convenient
use do not have many alternatives. However, supply is fairly elastic, meaning producers are
expected to be fairly responsive to a change in price. This also makes intuitive sense since PFC
producers can take steps in both the short term and long term to adjust production in response to
price changes. In the short run, if prices decrease, they can easily store finished PFCs, holding
them out of the market until prices increase again. If prices increase, it is relatively inexpensive
for producers to increase output since the production processes are not complex or require
expensive equipment. Therefore, consumers are expected to bear more of the burden of PFC
regulatory control costs.
Because the elasticity estimates are a key input to the model, a sensitivity analysis for
supply and demand elasticity parameters was performed as part of this analysis. The results are
presented in Appendix E.
Table 13.3-10. Summary of Elasticities Used in the EIM
Market
Estimate
Source
Method
Input Data
Summary
Supply Elasticities
Gasoline Fuel
Portable Fuel
Containers
0.24
1.50
Considine (2002)32
EPA econometric
estimate (see
Appendix C)
Literature estimate
Cobb-Douglas
production function
NA
Bartlesman33; 1980-
1996; SIC 3089
Demand Elasticities
Gasoline Fuel
Portable Fuel
Containers
-0.20
-0.01
Federal Trade
Commission
(200 1)34
EPA numerical
simulation (see
Appendix D)
Literature estimate
Hicks-Allen derived
demand
NA
Described in
Appendix D
13.3.6 Economic Impact Model Structure
The EIM developed for this analysis is a spreadsheet model that estimates changes in
price and quantity in a market that are expected to occur as a result of an increase in producer
costs in the amount of the compliance costs associated with the standards. The impacts on the
gasoline and PFC markets are modeled separately, and there is no feedback between the two
models. The model for each of these two markets consists of one demand curve and one supply
curve, reflecting the fact that the standards affect only one group of producers (PFC
manufacturers, gasoline fuel refiners) and one group of consumers (residential PFC users,
13-46
-------
Final Regulatory Impact Analysis
residential gasoline fuel users). There are no intermediate levels in the market since there are no
intermediate producers and consumers affected by the standards.
This structure makes the model relatively simple to construct and solve. Specifically, the
EEVI's partial equilibrium models use a commonly used analytical expression used in the
analysis of supply and demand in a single market.35'36 Appendix D explains in detail how this
expression is derived using the following steps:
1. Specify a set of supply and demand relationships for each market.
2. Simplify the equations by transforming them into a set of linear equations.
3. Solve the equilibrium system of equations.
Using this expression, we can estimate the market price change in terms of the market's supply
and elasticity parameters and the regulatory program's per unit cost (Equation D.5 in Appendix
D)
, . Supply Elasticity
Apnce = ^^ x per - unit cost
(Supply Elasticity - Demand Elasticity)
Given the market price change due to increased cost required by the rule and the demand
elasticity for each market, we can also estimate the market quantity change.
Aquantity = Aprice x Demand Elasticity
13-47
-------
Appendix 13A: Impacts on Portable Fuel Container Markets
This appendix provides the time series of impacts from 2009 through 2035 for the PFC
markets. Two separate markets were modeled and segmented by existence of a state regulatory
program.
Table 13A-1 provides the time series of impacts for each market and includes the
following:
• average engineering costs (variable and fixed) per can
• absolute change in the market price ($)
• relative change in market price (%)
• absolute change in market quantity (%)
• relative change in market quantity (%)
• consumer, producer, and total surplus losses
All prices and costs are presented in 2003$ and real PFC prices are assumed to be
constant during the period of analysis.
13-48
-------
Final Regulatory Impact Analysis
Table 13A-1. Regional Impacts: Portable Fuel Container Markets
Without State Program
(Average price $4.66)
Average
Total Cost
Year (S/can)
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
NPV 3%
NPV 7%
$0.00
$0.00
$2.70
$2.70
$2.70
$2.70
$2.70
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
$1.53
Change in
Price
(S/can)
$0.00
$0.00
$2.68
$2.68
$2.68
$2.68
$2.68
$1.52
$1.52
$1.52
$1.52
$1.52
$1.52
$1.52
$1.52
$1.52
$1.52
$1.52
$1.52
$1.52
$1.52
$1.52
$1.52
$1.52
$1.52
$1.52
$1.52
$1.52
$1.52
Change
in Price
(%)
0.00%
0.00%
57.45%
57.45%
57.45%
57.45%
57.45%
32.49%
32.49%
32.49%
32.49%
32.49%
32.49%
32.49%
32.49%
32.49%
32.49%
32.49%
32.49%
32.49%
32.49%
32.49%
32.49%
32.49%
32.49%
32.49%
32.49%
32.49%
32.49%
Change in
Quantity
(thousand
cans)
0
0
-104.7
-106.8
-108.9
-111.1
-113.3
-65.35
-66.66
-67.99
-69.35
-70.74
-72.15
-73.60
-75.07
-76.57
-78.10
-79.66
-81.26
-82.88
-84.54
-86.23
-87.96
-89.72
-91.51
-93.34
-95.21
-97.11
-99.05
Change in
Quantity
(%)
0.00%
0.00%
-0.57%
-0.57%
-0.57%
-0.57%
-0.57%
-0.32%
-0.32%
-0.32%
-0.32%
-0.32%
-0.32%
-0.32%
-0.32%
-0.32%
-0.32%
-0.32%
-0.32%
-0.32%
-0.32%
-0.32%
-0.32%
-0.32%
-0.32%
-0.32%
-0.32%
-0.32%
-0.32%
CS Loss
(million S)
$0
$0
-$48.7
-$49.7
-$50.7
-$51.7
-$52.7
-$30.4
-$31.0
-$31.7
-$32.3
-$32.9
-$33.6
-$34.3
-$35.0
-$35.7
-$36.4
-$37.1
-$37.8
-$38.6
-$39.4
-$40.2
-$41.0
-$41.8
-$42.6
-$43.5
-$44.3
-$45.2
-$46.1
-$676.2
-$399.8
PS Loss
(million S)
$0
$0
-$0.3
-$0.3
-$0.3
-$0.4
-$0.4
-$0.2
-$0.2
-$0.2
-$0.2
-$0.2
-$0.2
-$0.2
-$0.2
-$0.2
-$0.2
-$0.3
-$0.3
-$0.3
-$0.3
-$0.3
-$0.3
-$0.3
-$0.3
-$0.3
-$0.3
-$0.3
-$0.3
-$4.5
-$2.7
Total
Social Cost
(million S)
$0
$0
-$49.0
-$50. .0
-$51.0
-$52.0
-$53.1
-$30.6
-$31.3
-$31.9
-$32.5
-$33.2
-$33.3
-$34.5
-$35.2
-$35.9
-$36.6
-$37.4
-$38.1
-$38.9
-$39.6
-$40.4
-$41.2
-$42.1
-$42.9
-$43.8
-$44.6
-$45.5
-$46.4
-$680.7
-$402.5
(continued)
13-49
-------
Table 13A-1. Regional Impacts: Portable Fuel Container Markets (continued)
With State Program
(Average price $11.05)
Average
Total Cost
Year (S/can)
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
NPV 3%
NPV 7%
$0.00
$0.00
$0.77
$0.77
$0.77
$0.77
$0.77
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
$0.21
Change in
Price
(S/can)
$0.00
$0.00
$0.76
$0.76
$0.76
$0.76
$0.76
$0.20
$0.20
$0.20
$0.20
$0.20
$0.20
$0.20
$0.20
$0.20
$0.20
$0.20
$0.20
$0.20
$0.20
$0.20
$0.20
$0.20
$0.20
$0.20
$0.20
$0.20
$0.20
Change
in Price
0.00%
0.00%
6.89%
6.89%
6.89%
6.89%
6.89%
1.85%
1.85%
1.85%
1.85%
1.85%
1.85%
1.85%
1.85%
1.85%
1.85%
1.85%
1.85%
1.85%
1.85%
1.85%
1.85%
1.85%
1.85%
1.85%
1.85%
1.85%
1.85%
Change in
Quantity
(thousand
cans)
0.00
0.00
-8.02
-8.19
-8.35
-8.52
-8.69
-2.38
-2.43
-2.48
-2.53
-2.58
-2.63
-2.68
-2.74
-2.79
-2.85
-2.91
-2.96
-3.02
-3.08
-3.14
-3.21
-3.27
-3.34
-3.40
-3.47
-3.54
-3.61
Change in
Quantity
0.00%
0.00%
-0.07%
-0.07%
-0.07%
-0.07%
-0.07%
-0.02%
-0.02%
-0.02%
-0.02%
-0.02%
-0.02%
-0.02%
-0.02%
-0.02%
-0.02%
-0.02%
-0.02%
-0.02%
-0.02%
-0.02%
-0.02%
-0.02%
-0.02%
-0.02%
-0.02%
-0.02%
-0.02%
CS Loss
(million S)
$0.00
$0.00
-$8.86
-$9.04
-$9.22
-$9.40
-$9.59
-$2.63
-$2.68
-$2.74
-$2.79
-$2.85
-$2.91
-$2.96
-$3.02
-$3.08
-$3.15
-$3.21
-$3.27
-$3.34
-$3.41
-$3.47
-$3.54
-$3.61
-$3.69
-$3.76
-$3.83
-$3.91
-$3.99
-$78.7
-$50.7
PS Loss
(million S)
$0.00
$0.00
-$0.06
-$0.06
-$0.06
-$0.06
-$0.06
-$0.02
-$0.02
-$0.02
-$0.02
-$0.02
-$0.02
-$0.02
-$0.02
-$0.02
-$0.02
-$0.02
-$0.02
-$0.02
-$0.02
-$0.02
-$0.02
-$0.02
-$0.02
-$0.03
-$0.03
-$0.03
-$0.03
-$0.5
-$0.3
Total
Social Cost
(million S)
$0.00
$0.00
-$8.92
-$9.10
-$9.28
-$9.47
-$9.65
-$2.65
-$2.70
-$2.76
-$2.81
-$2.87
-$2.93
-$2.98
-$3.04
-$3.10
-$3.17
-$3.23
-$3.29
-$3.36
-$3.43
-$3.50
-$3.57
-$3.64
-$3.71
-$3.78
-$3.86
-$3.94
-$4.02
-$79.3
-$51.1
13-50
-------
Final Regulatory Impact Analysis
Appendix 13B: Impacts on Gasoline Fuel Markets
This appendix provides the time series of impacts from 2009 through 2035 for the
gasoline markets. Four gasoline markets were modeled: Four PADDs (PADDs 1 & 3, PADD 2,
PADD 4, and PADD 5). Note that PADD 5 includes Alaska and Hawaii but excludes California
fuel volumes because they are covered by separate California standards.
Table 13B-1 provides the time series of impacts for each market and includes the
following:
• average engineering costs (variable and fixed) per gallon
• absolute change in the market price ($)
• relative change in market price (%)
• absolute change in market quantity (%)
• relative change in market quantity (%)
• consumer, producer, and total surplus losses
All prices and costs are presented in 2003$ and real gasoline prices are assumed to be
constant during the period of analysis. A sensitivity analysis of the constant price assumption is
provided in Appendix G.
13-51
-------
Table 13B-1. Regional Impacts: Gasoline Markets
PADD I&HI
(Average price $1.48)
Year
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
NPV 3%
NPV 7%
Average
Total Cost
(cents/
gallon)
0.010
0.016
0.016
0.031
0.031
0.058
0.053
0.053
0.149
0.149
0.149
0.149
0.149
0.149
0.149
0.149
0.149
0.149
0.149
0.149
0.149
0.149
0.149
0.149
0.149
0.149
0.149
0.149
0.149
Change in
Price
(cents/
gallon)
0.005
0.009
0.009
0.017
0.017
0.032
0.029
0.029
0.081
0.081
0.081
0.081
0.081
0.081
0.081
0.081
0.081
0.081
0.081
0.081
0.081
0.081
0.081
0.081
0.081
0.081
0.081
0.081
0.081
Change
in Price
0.004%
0.006%
0.006%
0.012%
0.012%
0.021%
0.019%
0.019%
0.055%
0.055%
0.055%
0.055%
0.055%
0.055%
0.055%
0.055%
0.055%
0.055%
0.055%
0.055%
0.055%
0.055%
0.055%
0.055%
0.055%
0.055%
0.055%
0.055%
0.055%
Change in
Quantity
(million
gallons)
-0.507
-0.883
-0.900
-1.762
-1.790
-3.345
-3.100
-3.139
-8.935
-9.044
-9.153
-9.271
-9.380
-9.481
-9.582
-9.674
-9.767
-9.868
-9.969
-10.078
-10.187
-10.288
-10.397
-10.498
-10.607
-10.718
-10.829
-10.942
-11.056
Change in
Quantity
-0.001%
-0.001%
-0.001%
-0.002%
-0.002%
-0.004%
-0.004%
-0.004%
-0.011%
-0.011%
-0.011%
-0.011%
-0.011%
-0.011%
-0.011%
-0.011%
-0.011%
-0.011%
-0.011%
-0.011%
-0.011%
-0.011%
-0.011%
-0.011%
-0.011%
-0.011%
-0.011%
-0.011%
-0.011%
CS Loss
(million S)
-$3.760
-$6.550
-$6.680
-$13.070
-$13.280
-$24.830
-$23.010
-$23.290
-$66.300
-$67.110
-$67.920
-$68.800
-$69.610
-$70.350
-$71.100
-$71.790
-$72.470
-$73.220
-$73.970
-$74.780
-$75.590
-$76.340
-$77.150
-$77.900
-$78.710
-$79.530
-$80.360
-$81.190
-$82.040
-$959.735
-$499.236
PS Loss
(million S)
-$3.140
-$5.460
-$5.560
-$10.890
-$11.070
-$20.690
-$19.170
-$19.410
-$55.250
-$55.930
-$56.600
-$57.330
-$58.000
-$58.630
-$59.250
-$59.820
-$60.400
-$61.020
-$61.640
-$62.320
-$62.990
-$63.620
-$64.290
-$64.920
-$65.590
-$66.280
-$66.970
-$67.660
-$68.370
-$799.789
-$416.034
Total
Social Cost
(million S)
-$6.900
-$12.010
-$12.240
-$23.970
-$24.350
-$45.510
-$42.180
-$42.710
-$121.550
-$123.040
-$124.520
-$126.120
-$127.610
-$128.980
-$130.350
-$131.610
-$132.870
-$134.240
-$135.610
-$137.100
-$138.590
-$139.960
-$141.450
-$142.820
-$144.300
-$145.810
-$147.320
-$148.860
-$150.410
-$1,759.536
-$915.276
13-52
-------
Final Regulatory Impact Analysis
Table 13B-1. Regional Impacts: Gasoline Markets (continued)
PADDII
(Average price $1.51)
Year
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
NPV 3%
NPV 7%
Average
Total Cost
(cents/
gallon)
0.053
0.091
0.091
0.194
0.194
0.308
0.227
0.227
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
0.307
Change in
Price
(cents/
gallon)
0.029
0.050
0.050
0.106
0.106
0.168
0.124
0.124
0.167
0.167
0.167
0.167
0.167
0.167
0.167
0.167
0.167
0.167
0.167
0.167
0.167
0.167
0.167
0.167
0.167
0.167
0.167
0.167
0.167
Change
in Price
0.019%
0.033%
0.033%
0.070%
0.070%
0.111%
0.082%
0.082%
0.111%
0.111%
0.111%
0.111%
0.111%
0.111%
0.111%
0.111%
0.111%
0.111%
0.111%
0.111%
0.111%
0.111%
0.111%
0.111%
0.111%
0.111%
0.111%
0.111%
0.111%
Change in
Quantity
(million
gallons)
-1.541
-2.683
-2.734
-5.942
-6.038
-9.702
-7.253
-7.344
-10.056
-10.179
-10.302
-10.434
-10.557
-10.670
-10.784
-10.888
-10.992
-11.106
-11.219
-11.342
-11.465
-11.579
-11.702
-11.815
-11.938
-12.062
-12.188
-12.315
-12.443
Change
in
Quantity
-0.004%
-0.007%
-0.007%
-0.014%
-0.014%
-0.022%
-0.016%
-0.016%
-0.022%
-0.022%
-0.022%
-0.022%
-0.022%
-0.022%
-0.022%
-0.022%
-0.022%
-0.022%
-0.022%
-0.022%
-0.022%
-0.022%
-0.022%
-0.022%
-0.022%
-0.022%
-0.022%
-0.022%
-0.022%
CS Loss
(million S)
-$11.630
-$20.250
-$20.630
-$44.830
-$45.550
-$73.200
-$54.730
-$55.410
-$75.870
-$76.800
-$77.720
-$78.720
-$79.650
-$80.510
-$81.360
-$82.150
-$82.930
-$83.790
-$84.650
-$85.570
-$86.500
-$87.360
-$88.290
-$89.140
-$90.070
-$91.010
-$91.950
-$92.910
-$93.880
-$1,260.43
-$699.59
PS Loss
(million S)
-$9.690
-$16.870
-$17.190
-$37.360
-$37.960
-$61.000
-$45.610
-$46.180
-$63.220
-$64.000
-$64.770
-$65.600
-$66.380
-$67.090
-$67.800
-$68.460
-$69.110
-$69.820
-$70.540
-$71.310
-$72.080
-$72.800
-$73.570
-$74.290
-$75.060
-$75.840
-$76.630
-$77.430
-$78.230
-$1,050.36
-$582.99
Total
Social Cost
(million S)
-$21.310
-$37.120
-$37.820
-$82.190
-$83.520
-$134.210
-$100.330
-$101.590
-$139.090
-$140.790
-$142.490
-$144.320
-$146.030
-$147.600
-$149.170
-$150.610
-$152.040
-$153.610
-$155.180
-$156.890
-$158.590
-$160.160
-$161.860
-$163.430
-$165.130
-$166.850
-$168.580
-$170.340
-$172.110
-$2,310.79
-$1,282.59
13-53
-------
Table 13B-1. Regional Impacts: Gasoline Markets (continued)
PADDIV
(Average price $1.57)
Year
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
Average
Total Cost
(cents/
gallon)
0.019
0.033
0.033
0.099
0.099
0.213
0.227
0.227
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
0.501
NPV 3%
NPV 7%
Change in
Price
(cents/
gallon)
0.011
0.018
0.018
0.054
0.054
0.116
0.124
0.124
0.273
0.273
0.273
0.273
0.273
0.273
0.273
0.273
0.273
0.273
0.273
0.273
0.273
0.273
0.273
0.273
0.273
0.273
0.273
0.273
0.273
Change
in Price
(%)
0.007%
0.011%
0.011%
0.034%
0.034%
0.074%
0.079%
0.079%
0.174%
0.174%
0.174%
0.174%
0.174%
0.174%
0.174%
0.174%
0.174%
0.174%
0.174%
0.174%
0.174%
0.174%
0.174%
0.174%
0.174%
0.174%
0.174%
0.174%
0.174%
Change in
Quantity
(million
gallons)
-0.063
-0.109
-0.111
-0.340
-0.346
-0.753
-0.814
-0.825
-1.842
-1.865
-1.888
-1.912
-1.934
-1.955
-1.976
-1.995
-2.014
-2.035
-2.056
-2.078
-2.101
-2.122
-2.144
-2.165
-2.187
-2.210
-2.233
-2.256
-2.280
Change in
Quantity
(%)
-0.001%
-0.002%
-0.002%
-0.007%
-0.007%
-0.015%
-0.016%
-0.016%
-0.035%
-0.035%
-0.035%
-0.035%
-0.035%
-0.035%
-0.035%
-0.035%
-0.035%
-0.035%
-0.035%
-0.035%
-0.035%
-0.035%
-0.035%
-0.035%
-0.035%
-0.035%
-0.035%
-0.035%
-0.035%
CS Loss
(million S)
-$0.490
-$0.860
-$0.880
-$2.680
-$2.720
-$5.920
-$6.410
-$6.490
-$14.500
-$14.680
-$14.850
-$15.040
-$15.220
-$15.380
-$15.550
-$15.700
-$15.850
-$16.010
-$16.180
-$16.350
-$16.530
-$16.690
-$16.870
-$17.030
-$17.210
-$17.390
-$17.570
-$17.750
-$17.940
-$210.758
-$109.585
PS Loss
(million S)
-$0.410
-$0.720
-$0.730
-$2.230
-$2.270
-$4.940
-$5.340
-$5.410
-$12.080
-$12.230
-$12.380
-$12.540
-$12.680
-$12.820
-$12.960
-$13.080
-$13.210
-$13.340
-$13.480
-$13.630
-$13.770
-$13.910
-$14.060
-$14.200
-$14.340
-$14.490
-$14.640
-$14.800
-$14.950
-$175.646
-$91.329
Total
Social Cost
(million S)
-$0.910
-$1.580
-$1.610
-$4.910
-$4.990
-$10.860
-$11.750
-$11.900
-$26.580
-$26.900
-$27.230
-$27.580
-$27.900
-$28.200
-$28.500
-$28.780
-$29.050
-$29.350
-$29.650
-$29.980
-$30.300
-$30.600
-$30.930
-$31.230
-$31.550
-$31.880
-$32.220
-$32.550
-$32.890
-$386.393
-$200.910
13-54
-------
Final Regulatory Impact Analysis
Table 13B-1. Regional Impacts: Gasoline Markets (continued)
PADD V (excluding California)
(Average price $1.66)
Year
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
Average
Total Cost
(cents/
gallon)
0.004
0.007
0.007
0.035
0.035
0.140
0.244
0.244
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
0.997
NPV 3%
NPV 7%
Change in
Price
(cents/
gallon)
0.002
0.004
0.004
0.019
0.019
0.076
0.133
0.133
0.544
0.544
0.544
0.544
0.544
0.544
0.544
0.544
0.544
0.544
0.544
0.544
0.544
0.544
0.544
0.544
0.544
0.544
0.544
0.544
0.544
Change
in Price
(%)
0.001%
0.002%
0.002%
0.011%
0.011%
0.046%
0.080%
0.080%
0.327%
0.327%
0.327%
0.327%
0.327%
0.327%
0.327%
0.327%
0.327%
0.327%
0.327%
0.327%
0.327%
0.327%
0.327%
0.327%
0.327%
0.327%
0.327%
0.327%
0.327%
Change in
Quantity
(million
gallons)
-0.022
-0.038
-0.039
-0.199
-0.202
-0.816
-1.445
-1.463
-6.051
-6.125
-6.199
-6.279
-6.353
-6.421
-6.489
-6.552
-6.614
-6.683
-6.751
-6.825
-6.899
-6.967
-7.041
-7.110
-7.184
-7.258
-7.334
-7.410
-7.487
Change in
Quantity
(%)
0.000%
0.000%
0.000%
-0.002%
-0.002%
-0.009%
-0.016%
-0.016%
-0.065%
-0.065%
-0.065%
-0.065%
-0.065%
-0.065%
-0.065%
-0.065%
-0.065%
-0.065%
-0.065%
-0.065%
-0.065%
-0.065%
-0.065%
-0.065%
-0.065%
-0.065%
-0.065%
-0.065%
-0.065%
CS Loss
(million S)
-$0.180
-$0.320
-$0.320
-$1.650
-$1.680
-$6.780
-$12.010
-$12.160
-$50.280
-$50.900
-$51.510
-$52.180
-$52.790
-$53.360
-$53.930
-$54.450
-$54.970
-$55.540
-$56.100
-$56.720
-$57.330
-$57.900
-$58.520
-$59.080
-$59.700
-$60.320
-$60.950
-$61.580
-$62.220
-$684.454
-$343.746
PS Loss
(million S)
-$0.150
-$0.270
-$0.270
-$1.380
-$1.400
-$5.650
-$10.010
-$10.130
-$41.900
-$42.420
-$42.930
-$43.480
-$43.990
-$44.470
-$44.940
-$45.370
-$45.810
-$46.280
-$46.750
-$47.270
-$47.780
-$48.250
-$48.760
-$49.240
-$49.750
-$50.270
-$50.790
-$51.320
-$51.850
-$570.394
-$286.466
Total
Social Cost
(million S)
-$0.330
-$0.580
-$0.590
-$3.030
-$3.080
-$12.430
-$22.020
-$22.290
-$92.190
-$93.320
-$94.440
-$95.660
-$96.790
-$97.830
-$98.870
-$99.820
-$100.770
-$101.820
-$102.860
-$103.980
-$105.110
-$106.150
-$107.280
-$108.320
-$109.450
-$110.590
-$111.740
-$112.900
-$114.070
-$1,254.848
-$630.211
13-55
-------
Appendix 13C: Time Series of Social Costs
This appendix provides a time series of the rule's estimated social costs from 2009
through 2035. Costs are presented in 2003 dollars.
13-56
-------
Table 13C-1. Time Series of Social Costs
Consumer Surplus Change, Total
Gasoline, U.S.
PADD I & III
PADDII
PADD IV
PADD V (excludes California)
Gas Cans, U.S.
States With State Regulatory
Programs
States Without State Regulatory
Programs
Producer Surplus Change, Total
Gasoline, U.S.
PADD I & III
PADDII
PADD IV
PADD V (excludes California)
PADD V (California)
Gas Cans, U.S.
States With State Regulatory
Programs
States Without State Regulatory
Programs
Fuel Savings
Consumer Savings
Fuel
Tax
Government Revenue
Vehicle Program
Total Surplus Change
2007
-$16.1
-$16.1
-$3.8
-$11.6
-$0.5
-$0.2
$0.0
$0.0
$0.0
-$13.4
-$13.4
-$3.1
-$9.7
-$0.4
-$0.2
$0.0
$0.0
$0.0
$0.0
$0.0
$0.0
$0.0
$0.0
$0.0
$0.0
-$29.5
2008
-$28.0
-$28.0
-$6.6
-$20.3
-$0.9
-$0.3
$0.0
$0.0
$0.0
-$23.3
-$23.3
-$5.5
-$16.9
-$0.7
-$0.3
$0.0
$0.0
$0.0
$0.0
$0.0
$0.0
$0.0
$0.0
$0.0
$0.0
-$51.3
2009
-$86.1
-$28.5
-$6.7
-$20.6
-$0.9
-$0.3
-$57.5
-$8.9
-$48.7
-$24.1
-$23.8
-$5.6
-$17.2
-$0.7
-$0.3
$0.0
-$0.4
-$0.1
-$0.3
$11.3
$15.2
$11.3
$3.9
-$3.9
$0.0
-$98.9
2010
-$120.9
-$62.2
-$13.1
-$44.8
-$2.7
-$1.7
-$58.7
-$9.0
-$49.7
-$52.3
-$51.9
-$10.9
-$37.4
-$2.2
-$1.4
$0.0
-$0.4
-$0.1
-$0.3
$22.6
$30.4
$22.6
$7.8
-$7.8
-$11.1
-$161.7
2011
-$123.1
-$63.2
-$13.3
-$45.6
-$2.7
-$1.7
-$59.9
-$9.2
-$50.7
-$53.1
-$52.7
-$11.1
-$38.0
-$2.3
-$1.4
$0.0
-$0.4
-$0.1
-$0.3
$35.6
$47.9
$35.6
$12.3
-$12.3
-$11.8
-$152.4
2012
-$171.8
-$110.7
-$24.8
-$73.2
-$5.9
-$6.8
-$61.1
-$9.4
-$51.7
-$92.7
-$92.3
-$20.7
-$61.0
-$4.9
-$5.7
$0.0
-$0.4
-$0.1
-$0.3
$48.5
$65.4
$48.5
$16.8
-$16.8
-$12.5
-$228.5
2013
-$158.5
-$96.2
-$23.0
-$54.7
-$6.4
-$12.0
-$62.3
-$9.6
-$52.7
-$80.5
-$80.1
-$19.2
-$45.6
-$5.3
-$10.0
$0.0
-$0.4
-$0.1
-$0.4
$61.5
$82.8
$61.5
$21.3
-$21.3
-$13.3
-$190.8
2014
-$130.4
-$97.4
-$23.3
-$55.4
-$6.5
-$12.2
-$33.1
-$2.6
-$30.4
-$81.4
-$81.1
-$19.4
-$46.2
-$5.4
-$10.1
$0.0
-$0.2
$0.0
-$0.2
$74.5
$100.3
$74.5
$25.8
-$25.8
-$13.4
-$150.7
2015
-$240.7
-$207.0
-$66.3
-$75.9
-$14.5
-$50.3
-$33.7
-$2.7
-$31.0
-$172.7
-$172.5
-$55.3
-$63.2
-$12.1
-$41.9
$0.0
-$0.2
$0.0
-$0.2
$75.5
$101.7
$75.5
$26.2
-$26.2
-$12.9
-$350.7
2016
-$243.9
-$209.5
-$67.1
-$76.8
-$14.7
-$50.9
-$34.4
-$2.7
-$31.7
-$174.8
-$174.6
-$55.9
-$64.0
-$12.2
-$42.4
$0.0
-$0.2
$0.0
-$0.2
$76.5
$103.1
$76.5
$26.5
-$26.5
-$12.2
-$354.4
(continued)
13-57
-------
Table 13C-1. Time Series of Social Costs (continued)
Consumer Surplus Change, Total
Gasoline, U.S.
PADD I & III
PADDII
PADD IV
PADD V (excludes California)
Gas Cans, U.S.
States With State Regulatory
Programs
States Without State Regulatory
Programs
Producer Surplus Change, Total
Gasoline, U.S.
PADD I & III
PADDII
PADD IV
PADD V (excludes California)
PADD V (California)
Gas Cans, U.S.
States With State Regulatory
Programs
States Without State Regulatory
Programs
Fuel Savings
Consumer Savings
Fuel
Tax
Government Revenue
Vehicle Program
Total Surplus Change
2017
-$247.1
-$212.0
-$67.9
-$77.7
-$14.9
-$51.5
-$35.1
-$2.8
-$32.3
-$176.9
-$176.7
-$56.6
-$64.8
-$12.4
-$42.9
$0.0
-$0.2
$0.0
-$0.2
$77.6
$104.5
$77.6
$26.9
-$26.9
-$11.4
-$357.9
2018
-$250
-$214
-$68
-$78
-$15
-$52
-$35
.5
.7
.8
.7
.0
.2
.8
-$2.9
-$32
-$179
-$179
-$57
-$65
-$12
-$43
$0
-$o
$0
-$o
$78
$105
$78
$27
-$27
-$10
-$361
.9
.2
.0
.3
.6
.5
.5
.0
.2
.0
.2
.6
.9
.6
.3
.3
.7
.8
2019
-$253.8
-$217.3
-$69.6
-$79.7
-$15.2
-$52.8
-$36.5
-$2.9
-$33.6
-$181.3
-$181.1
-$58.0
-$66.4
-$12.7
-$44.0
$0.0
-$0.2
$0.0
-$0.2
$79.7
$107.3
$79.7
$27.6
-$27.6
-$10.6
-$366.0
2020
-$256.8
-$219.6
-$70.4
-$80.5
-$15.4
-$53.4
-$37.2
-$3.0
-$34.3
-$183.3
-$183.0
-$58.6
-$67.1
-$12.8
-$44.5
$0.0
-$0.3
$0.0
-$0.2
$80.7
$108.7
$80.7
$28.0
-$28.0
$0.0
-$359.4
2021
-$259.9
-$221.9
-$71.1
-$81.4
-$15.6
-$53.9
-$38.0
-$3.0
-$35.0
-$185.2
-$185.0
-$59.3
-$67.8
-$13.0
-$44.9
$0.0
-$0.3
$0.0
-$0.2
$81.8
$110.1
$81.8
$28.4
-$28.4
$0.0
-$363.3
2022
-$262.8
-$224.1
-$71.8
-$82.2
-$15.7
-$54.5
-$38.7
$o i
3.1
-$35.7
-$187.0
-$186.7
-$59.8
-$68.5
-$13.1
-$45.4
$0.0
-$0.3
$0.0
-$0.2
$82.9
$111.6
$82.9
$28.7
-$28.7
$0.0
-$367.0
2023
-$265.7
-$226.2
-$72.5
-$82.9
-$15.9
-$55.0
-$39.5
-$3.2
-$36.4
-$188.8
-$188.5
-$60.4
-$69.1
-$13.2
-$45.8
$0.0
-$0.3
$0.0
-$0.2
$83.9
$113.1
$83.9
$29.1
-$29.1
$0.0
-$370.6
2024
-$268.9
-$228.6
-$73.2
-$83.8
-$16.0
-$55.5
-$40.3
-$3.2
-$37.1
-$190.7
-$190.5
-$61.0
-$69.8
-$13.3
-$46.3
$0.0
-$0.3
$0.0
-$0.3
$85.0
$114.5
$85.0
$29.5
-$29.5
$0.0
-$374.6
2025
-$272.0
-$230.9
-$74.0
-$84.7
-$16.2
-$56.1
-$41.1
-$3.3
-$37.8
-$192.7
-$192.4
-$61.6
-$70.5
-$13.5
-$46.8
$0.0
-$0.3
$0.0
-$0.3
$86.1
$116.0
$86.1
$29.9
-$29.9
$0.0
-$378.6
2026
-$275.4
-$233.4
-$74.8
-$85.6
-$16.4
-$56.7
-$41.9
-$3.3
-$38.6
-$194.8
-$194.5
-$62.3
-$71.3
-$13.6
-$47.3
$0.0
-$0.3
$0.0
-$0.3
$87.2
$117.4
$87.2
$30.2
-$30.2
$0.0
-$383.0
13-58
-------
Table 13C-1. Time Series of Social Costs (continued)
Consumer Surplus Change, Total
Gasoline, U.S.
PADD I & III
PADDII
PADD IV
PADD V (excludes California)
Gas Cans, U.S.
States With State Regulatory Programs
States Without State Regulatory Programs
Producer Surplus Change, Total
Gasoline, U.S.
PADD I & III
PADDII
PADD IV
PADD V (excludes California)
PADD V (California)
Gas Cans, U.S.
States With State Regulatory Programs
States Without State Regulatory Programs
Fuel Savings
Consumer Savings
Fuel
Tax
Government Revenue
Vehicle Program
Total Surplus Change
2027
-$278.7
-$236.0
-$75.6
-$86.5
-$16.5
-$57.3
-$42.8
-$3.4
-$39.4
-$196.9
-$196.6
-$63.0
-$72.1
-$13.8
-$47.8
$0.0
-$0.3
$0.0
-$0.3
$88.3
$118.9
$88.3
$30.6
-$30.6
$0.0
-$387.4
2028
-$281
-$238
-$76
-$87
-$16
-$57
-$43
-$3
-$40
-$198
-$198
-$63
-$72
-$13
-$48
$0
-$o
$0
-$o
$89
$120
$89
$31
-$31
$0
-$391
.9
.3
.3
.4
.7
.9
.6
.5
.2
.9
.6
.6
.8
.9
o
.J
.0
.3
.0
.3
.3
o
.J
o
.J
.0
.0
.0
.4
2029
-$285
-$240
-$77
-$88
-$16
-$58
-$44
-$3
-$41
-$201
-$200
-$64
-$73
-$14
-$48
$0
-$o
$0
-$o
$90
$121
$90
$31
-$31
$0
-$395
.3
.8
.2
.3
.9
.5
.5
.5
.0
.0
.7
.3
.6
.1
.8
.0
.3
.0
.3
.4
.8
.4
.4
.4
.0
.9
2030
-$288.5
-$243.2
-$77.9
-$89.1
-$17.0
-$59.1
-$45.4
-$3.6
-$41.8
-$203.0
-$202.7
-$64.9
-$74.3
-$14.2
-$49.2
$0.0
-$0.3
$0.0
-$0.3
$91.5
$123.2
$91.5
$31.7
-$31.7
$0.0
-$400.0
2031
-$292.0
-$245.7
-$78.7
-$90.1
-$17.2
-$59.7
-$46.3
-$3.7
-$42.6
-$205.0
-$204.7
-$65.6
-$75.1
-$14.3
-$49.8
$0.0
-$0.3
$0.0
-$0.3
$92.5
$124.6
$92.5
$32.1
-$32.1
$0.0
-$404.5
2032
-$295.5
-$248.3
-$79.5
-$91.0
-$17.4
-$60.3
-$47.2
-$3.8
-$43.5
-$207.2
-$206.9
-$66.3
-$75.8
-$14.5
-$50.3
$0.0
-$0.3
$0.0
-$0.3
$93.6
$126.0
$93.6
$32.5
-$32.5
$0.0
-$409.1
2033
-$299.0
-$250.8
-$80.4
-$92.0
-$17.6
-$61.0
-$48.2
-$3.8
-$44.3
-$209.4
-$209.0
-$67.0
-$76.6
-$14.6
-$50.8
$0.0
-$0.3
$0.0
-$0.3
$94.6
$127.4
$94.6
$32.8
-$32.8
$0.0
-$413.7
2034
-$302.6
-$253.4
-$81.2
-$92.9
-$17.8
-$61.6
-$49.1
-$3.9
-$45.2
-$211.5
-$211.2
-$67.7
-$77.4
-$14.8
-$51.3
$0.0
-$0.3
$0.0
-$0.3
$95.7
$128.8
$95.7
$33.2
-$33.2
$0.0
-$418.4
2035
-$306.2
-$256.1
-$82.0
-$93.9
-$17.9
-$62.2
-$50.1
-$4.0
-$46.1
-$213.7
-$213.4
-$68.4
-$78.2
-$15.0
-$51.9
$0.0
-$0.3
$0.0
-$0.3
$96.7
$130.2
$96.7
$33.5
-$33.5
$0.0
-$423.2
13-59
-------
Appendix 13D: Overview of Economic Model Equations
We illustrate our approach for addressing conceptual questions of market-level impacts
using a numerical simulation model. Our method involves specifying a set of nonlinear supply
and demand relationships for the affected markets, simplifying the equations by transforming
them into a set of linear equations, and then solving the equilibrium system of equations.37
13D.1 Discussion and Specification of Model Equations
First, we consider the formal definition of the elasticity of supply with respect to changes
in own price:
(D.D
dp/ p
Next, we can use "hat" notation to transform Eq. (D.I) to proportional changes and rearrange
terms:
Qs = percentage change in the quantity of market supply,
es = market elasticity of supply, and
p = percentage change in market price.
As Fullerton and Metcalfe38 note, we have taken the elasticity definition and turned it into a
linear behavioral equation for our market. Similarly, we can specify a demand equation as
follows:
Qd = VdP (D.2)
Qd = percentage change in the quantity of market demand,
r[d = market elasticity of demand, and
p = percentage change in market price.
To introduce the direct impact of the regulatory program, we assume the per-unit cost (c) leads to
a proportional shift in the marginal cost of production. Under the assumption of perfect
competition (price equals marginal cost), we can approximate this shift at the initial equilibrium
point as follows:
MC = —5— = — (D.3)
MC0 Po
13-60
-------
Finally, we specify the market equilibrium conditions in the affected markets. In response
to the exogenous increase in production costs, producer and consumer behaviors are represented
in Eq. (D.la) and Eq. (D.2), and the new equilibrium satisfies the condition that the change in
supply equals the change in demand:
Qs = Qd (D-4)
We now have three linear equations in three unknowns ( p , Q^ , and Qs ) and we can
solve for the proportional price change in terms of the elasticity parameters^ and r^) and the
proportional change in marginal cost:
p = — — — • MC (D.5)
Given this solution, we can solve for the proportional change in market quantity using Eq. (D.2).
13D.2 Consumer and Producer Welfare Calculations
The change in consumer surplus in the affected markets can be estimated using the
following linear approximation method:
ACS = -Qi«Ap + 0.5«AQ«Ap. (D.6)
As shown, higher market prices and reduced consumption lead to welfare losses for consumers.
A geometric representation of this calculation is illustrated in Figure D-l.
For affected supply, the change in producer surplus can be estimated with the following
equation:
APS = Qi • (Ap - c) - 0.5 • AQ • (Ap - c). (D.7)
13-61
-------
Price
Increase
S-,: With Regulation
Unit Cost Increase
S0: Without Regulation
Qi Qo Output
A consumer surplus = -[fghd + dhc]
A producer surplus = [fghd - aehb] - bdc
A total surplus = -[aehb + dhc + bdc]
Figure D-l. Welfare Calculations
Increased regulatory costs and output declines have a negative effect on producer surplus,
because the net price change (Ap - c) is negative. However, these losses are mitigated, to some
degree, as a result of higher market prices. A geometric representation of this calculation is
illustrated in Figure D-l.
13-62
-------
Appendix 13E: Elasticity Parameters
To estimate market equilibrium price and quantity, supply and demand elasticities are
needed to represent the behavior adjustments that are likely to be made by market participants/
Tables 13E-1 and 13E-2 provide a summary of the supply and demand elasticities used to
estimate the economic impact of the rule.
Table 13E-1. Summary of Supply Elasticities Used in the EIA Model
Markets
All Gasoline
Markets
Portable Fuel
Container
Markets
Estimate
0.24
1.50
Source
Considine39
EPA econometric estimate
(see Section 13E.4)
Method
Literature estimate
Cobb-Douglas
production function
Input Data Summary
NA
Bartlesman40; 1980-1996;
SIC 3089
Table E-2. Summary of Demand Elasticities Used in EIA Model
Market
All Gasoline
Markets
Portable Fuel
Container
Markets
Estimate
-0.20
-0.01
FTC
EPA
(see
Source
41
numerical simulation
Section 13E.3)
Method
Literature estimate
Hicks-Allen derived
demand
Input
NA
Data Summary
Described in Section
13E.3
13E.1 Gasoline Market Parameters
Very few studies have attempted to quantify supply responsiveness for individual refined
products, such as gasoline fuel. For example, a study for the California Energy Commission
stated "There do not seem to be credible estimates of gasoline supply elasticity."42 However,
sources agree that refineries have little or no ability to change output in response to price: high
fixed costs compel them to operate as close to their capacity limit as possible. The Federal Trade
Commission (FTC) analysis made this point explicitly.43
Greene and Tishchishyna reviewed supply elasticity estimates available in the
literature.44 The supply elasticity values cited in most of these studies were for "petroleum" or
"oil" production in the United States, which includes exploration, distribution and refining
activities. The lowest short-term numbers cited were 0.02 to 0.05, with long-run values ranging
JThe models equations are described in Appendix A.
13-63
-------
from 0.4 to 1.0. It seems likely that these extremely low numbers are influenced by the limited
domestic supply of crude petroleum and the difficulty of extraction.
A recent paper by Considine provides one of the few supply elasticity estimates for
refining production (excluding extraction and distribution), based on historical price and quantity
data.45 In this study, Considine estimates a refining production supply elasticity of 0.24. This
estimate is for aggregate refinery production and includes distillate and nondistillate fuels.
Because petroleum products are made in strict proportion and refineries have limited ability to
adjust output mix in the short to medium run, it is reasonable to assume that supply is relatively
inelastic and similar across refinery products. This value of 0.24 was used for the supply
elasticity for this market. This estimated elasticity is inelastic, which means that the quantity of
goods and services supplied is expected to be fairly insensitive to price changes.
For demand elasticity estimates, EPA's NESHAP analysis of refinery markets included
the development of a price elasticity of demand elasticity for several refined petroleum
products.46 To compute this elasticity, EPA reviewed the economic literature and found
estimated for the following petroleum products:
• Motor gasoline: -0.5 5 to - 0.82.
• Jet fuel: -0.15.
• Residual fuel oil: -0.61 to -0.74.
• Distillate fuel oil: -0.50 to -0.99.
• Liquefied petroleum gas: -0.60 to -1.00
EPA developed a weighted average elasticity for petroleum products using the midpoints of the
elasticity estimates and production data for 1995. The use of the average value of-0.69 is more
consistent with long-run estimates of the gasoline price elasticity of demand.
However, a better choice for the primary analysis in this ELM is a short- to midterm-run
elasticity of-0.2 cited by the Federal Trade Commission.47 This value is consistent with recent
surveys of the gasoline demand literature.48'49 In addition, recent applied work on the incidence
of gas taxes suggests that the national demand elasticity should approximately equal the negative
of the national supply elasticity.50 Given that the supply elasticity we are using in the economic
model is 0.24, this implies a national gasoline demand elasticity of approximately -0.2.
13E.2 Portable Fuel Container Market Parameters
There are no estimated PFC demand elasticities from current economic literature. As a
result, we estimated this parameter numerically using a Hicks-Allen derived demand approach
(see Section E.3 for discussion) for a class of products that use similar production technologies
(SIC 3089, Plastic Products, Not Elsewhere Classified). Our Monte Carlo simulation and
13-64
-------
generated a mean value of-0.01 for the derived demand elasticity estimate for PFCs. Using this
value, a 1 percent change in the price of PFCs would lead to approximately a 0.014 percent
reduction in the quantity of PFCs demanded by consumers.
There are also no estimated PFC supply elasticities from the economic literature. As a
result, we estimated this parameter econometrically using a production function cost
minimization approach (see Section E.4 for discussion) for a class of products that use similar
production technologies (SIC 3089, Plastic Products, Not Elsewhere Classified). This category
includes manufacturers engaged in manufacturing plastic products not elsewhere classified and
includes such products as plastic containers and plastics drums. Using this approach, we found
the elasticity supply for these products is approximately 1.5, which means a 1 percent change in
the price of PFCs would lead to a 1.5 percent increase in the quantity of PFCs manufacturers
would be willing to sell in the market.
13E.3 Portable Fuel Container Demand Elasticity Estimation Procedure
Portable Fuel Containers are an integral component of any activity involving small
gasoline engines. These activities range from lawn and garden work to recreation use. The
behavioral change in PFC consumption is expected to be quite small in response to an increased
price because PFCs represent a small fraction of overall lawn and garden or recreation
expenditures. In addition, because PFCs are in many cases a necessity for small engine use,
households have limited ability to substitute away from PFCs as their price increases.
However, it is probably not appropriate to assume that the demand elasticity for PFCs is
zero. There will likely be some behavior response to the increased price of PFCs—even though it
is anticipated to be small. Unfortunately, an elasticity of demand for PFCs is not available in the
literature. Nor does the historical price and quantity data exist that would be required to
empirically estimate a demand elasticity for cans.
An alternative approach is to model PFCs as an input in the household production
function for household lawn and garden activities and develop a derived demand for PFCs
through changes in the household for lawn and garden products and services market. Because
over 90 percent of PFCs are used to support lawn and garden activities, we use the lawn and
garden market to derive a demand elasticity for PFCs.
The demand for PFCs is directly linked to the demand for lawn and garden products and
services. When the price of PFCs increases, the cost of the bundled commodity, lawn and garden
products services, also increases. This is illustrated in the supply curve's upward shift in Figure
E-l. This results in a reduced equilibrium quantity in the household lawn and garden services
market. Then, this reduced quantity feeds back into a reduced demand in the PFC market. For
example, if households reduce their purchases by X percent in the lawn and garden service
13-65
-------
market, this translates into the same X percent decrease in PFC purchases, which in turn
determines the derived demand point di in Figure E-1.K
13E.3.1 Numerical Example: Base Case
Because PFCs represent such as small fraction of household expenditures in the lawn and
garden services market, the resulting derived elasticity of demand is very small. As illustrated
below, with average annual household expenditures on lawn and garden services of $500 to
$2,500, and a $5 increase in the price of PFCs because of the regulation, the resulting shift in the
supply function is 1.0 percent to 0.2 percent.
Economic theory states that the elasticity of the derived demand for an input is a function
of the following:51'52'53
• demand elasticity for the final good it will be used to produce,
• the elasticity of supply of other inputs,
• the cost share of the input in total production cost, and
• the elasticity of substitution between this input and other inputs in production.
Using Hicks' formula,
Edc = [ a*(Edf + E^ + C*Esl*(Edf - a)] / [(Edf + ES1) - C*(Edf - a)] (E. 1)
where
Edc = price elasticity of demand for the cans,
Edf = price elasticity of demand for final product,
Esi = price elasticity of supply of other inputs,
C = cost share of cans in total production cost, and
a = elasticity of substitution between cans and all other inputs.
KThis assumes that PFCs are a fixed proportion input into the lawn and garden services market.
13-66
-------
Price Lawn
and Garden
Products
and
Services
Supply
Price Gas
Cans
Lawn and Garden
Products and Services
Supply
Derived Demand
Gas Can Output
Figure 13E-1. Derived Demand for Portable Fuel Containers
Using the parameter values in Table E-3, we conducted a Monte Carlo simulation and generated
the following derived demand elasticity estimate for PFCs:
Mean Value = -0.01
Standard Deviation = 0.004
Using the mean value, a 100 percent change in the price of PFCs would lead to
approximately a 1.0 percent reduction in the quantity of PFCs demanded by consumers.
13E.3.2
Numerical Example: Sensitivity
In the baseline analysis for the EIA, we propose to use a zero elasticity of substitution
between PFCs and all other inputs. This implies that consumers do not substitute away from
PFCs as the price increases. However, we acknowledge that there is a potential for households
13-67
-------
with more than one PFC to reduce the number of multiple can purchases as the price increases
(i.e., they may choose to reduce the number of cans they purchase, giving up the "luxury" of
Table 13E-3. Assumed Parameter Values Used to Generate Derived Demand Elasticity for
Portable Fuel Containers
Parameter
Edf
ESI
C
Type of Distribution
Normal
Uniform
Uniform
Values (range)
Mean = -1.2
StDev = 0.64
Min = 0.5
Max = 2.0
Min = 0.20%
Max= 1.0%
Comments
EPA econometric estimate for consumer
walk behind mowers
Assumed range
Example: $5 increase in cost for PFC, with
household lawn and garden expenditures of
$500 to $2,500
a
Assume fixed proportions technology
having multiple cans in multiple locations, or the capability of filling multiple cans with a single
trip to the gas station). These decisions in effect substitute additional household labor for the
convenience of having more than one PFC.
To investigate the potential impact of substitution in the PFC market, we conducted a
sensitivity analysis. Unfortunately, neither a literature estimate of substitution elasticity for PFCs
nor the data to estimate such elasticities exist. Thus, a substitution elasticity value of a = 0.1 was
used in the sensitivity analysis (see Table E-4). Using this value yields a demand elasticity for
cans with a mean value = -0.25 and a standard deviation = 0.45. This implies that a 100 percent
change in the price of PFCs would lead to approximately a 25 percent reduction in the quantity
of PFCs demanded by consumers. Specific impact estimates were estimated with engineering
cost data.
13E.4 Portable Fuel Container Supply Elasticity Estimation
Our approach assumes that firms minimize costs subject to production technology
constraints. To characterize these constraints, we use a "production function" that describes the
relationship between inputs and outputs of the production process. The functional form (Cobb-
Douglas) of the production function is specified as
(E.2)
13-68
-------
Table 13E-4. Assumed Parameter Values Used to Generate Derived Demand Elasticity for
Portable Fuel Containers
Parameter Type of Distribution Values (range)
Edf Normal
Es; Uniform
C Uniform
a
Mean = -1.2
StDev
Min =
Max =
Min =
Max =
0.1
= 0.64
0.5
2.0
0.20%
1.0%
Comments
EPA econometric estimate for consumer
walk behind mowers
Assumed range
Example: $5 increase in cost for
household expenditures of $500
on lawn and garden services
Used a single value
PFC, with
to $2,500
where
Qt = output in year t,
Kt = real capital consumed in production in year t,
Lt = quantity of labor used in year t,
Mt = material inputs in year t, and
t = a time trend variable to reflect technology changes.
This equation can be written in linear form by taking the natural logarithms of each side of the
equation. The parameters of this model, «K, CŁL, CŁM, can then be estimated using linear regression
techniques:
In Qt = In A + aK In Kt + aL In Lt + aM In Mt + A In t. (E.3)
Under the assumptions of a competitive market and perfect competition, the elasticity of supply
with respect to the price of the final product can be expressed in terms of the parameters of the
production function:
Supply Elasticity = (aL + aM) / (1 - aL - aM).
(E.4)
To maintain the desired properties of the Cobb-Douglas production function, it is
necessary to place restrictions on the estimated coefficients. For example, if aL + aM = 1, then
the supply elasticity will be undefined. Alternatively, if aL + aM > 1, this yields a negative
supply elasticity. Thus, a common assumption is that aK + aL + «M = 1. This implies constant
returns to scale, which is consistent with most empirical studies.
13E.4.1
Data Sets
13-69
-------
The National Bureau of Economic Research-Center for Economic Studies publishes
industry-level data used for the analysis (years 1958 to 1996).54 In cases where a price index was
not available, we used the most recent implicit gross domestic product (GDP) price deflator
reported by the U.S. Bureau of Economic Analysis.55 The following variables were used:56
• value of shipments,
• price index of value shipments,
• production worker wages,
GDP deflator,57
• cost of materials,
• price index for materials, and
• value added.
To provide a measure of capital consumed, a capital variable is calculated as follows:
Capital = (Value added - Production worker wages)/GDP deflator.
The NBER data set is restricted to four-digit SIC codes for the manufacturing industries. As a
result, we selected a class of products that use similar production technologies (SIC 3089, Plastic
Products, Not Elsewhere Classified). This category includes manufacturers engaged in
manufacturing plastic products not elsewhere classified and includes such products as plastic
containers and plastics drums. We also restricted our analysis to years after 1980, the time period
the Consumer Products Safety Commission identified plastic cans were introduced.58 The data
cover the period 1980 through 1996.
13E.4.2 Results of Supply Elasticity Estimation
We used an autoregressive error model to estimate Eq. (E.3). SAS procedure PROC
AUTOREG was used to compute a linear regression corrected for auto correlation. We assume
the error term is AR(2). This approach is identical to the one used successfully for the Nonroad
CI Engines and Equipment EIA completed in 2003, with some of the independent variables
updated with the most recent data.59 In addition, we also tested the assumption of constant error
variance using a Goldfeld-Quandt test and could not reject the hypothesis of homoskedasticity.
Using this model, we estimate a supply elasticity of 1.5 for this industry (see Table E-5).
Table 13E-5. Supply Elasticity Estimate for SIC 3089, Plastic Products, Not Elsewhere
Classified: 1980-1996
Supply elasticity = 1.5
Number of observations = 17
R-squared = 99.79
13-70
-------
Goldfeld-Quandt F(4,4) = 2.62 (p-value = 0.187)
dDW= 1.40
di = 0.90
du= 1.71
Variable Estimate t-value p-value
Intercept -0.3544
InK 0.4048 4.07 0.0019
InL 0.4404 3.21 0.0083
InM 0.1548 1.26 0.2339
InT 0.5087 7.27 <0.0001
13-71
-------
Appendix 13F: Initial Market Equilibrium - Price Forecasts
The EIM analysis begins with current market conditions: equilibrium supply and
demand. To estimate the economic impact of a regulation, standard practice uses projected
market equilibrium (time series of prices and quantities) as the baseline and evaluates market
changes from this projected baseline. Consequently, it is necessary to forecast equilibrium prices
and quantities for future years.
Equilibrium quantity forecasts are driven by projected activity factors and this approach
implicitly incorporates changes in production capacity during the period of analysis into the
baseline.
S/Q
•win
Bnpplj
Figure 13.3-1. Prices and Quantities in Long Run Market Equilibrium
Equilibrium price forecasts typically use one of two approaches.60 The first assumes a
constant (real) price of goods and services over time. The second models a specific time series
where prices may change over time due to exogenous factors.
In the absence of shocks to the economy or the supply of raw materials, economic theory
suggests that the equilibrium market price for goods and services should remain constant over
time. As shown in Figure 13.3-1, demand grows over time, in the long run, capacity will also
grow as existing firms expand or new firms enter the market and eliminate any excess profits.
This produces a flat long run supply curve. Note that in the short to medium run time frame the
supply curve has a positive slope due to limitations in how quickly firms can react.
If capacity is constrained (preventing the outward shift of the baseline supply curve) or if
the price of production inputs increase (shifting the baseline supply curve upward over time),
then prices may trend upward reflecting that either the growth in demand is exceeding supply or
the commodity is becoming more expensive to produce.
13-72
-------
It is very difficult to develop forecasts events (such as those mentioned above) that
influence long run prices. As a result, the approach used in this analysis is to use a constant 2003
observed price for PFCs and gasoline prices.
Nevertheless, there are forecasts of future gasoline prices, such as those provided by the
Annual Energy Outlook. To take these forecasts into account we performed a sensitivity
analysis using AEO forecasted prices for gasoline markets (see Appendix 13G).
13-73
-------
Appendix 13G: Sensitivity Analyses
The economic impact analysis presented in this Chapter 13 is based on an economic
impact mode (EIM) developed specifically for this analysis. This EIM reflects certain
assumptions about behavioral responses (modeled by supply and demand elasticities), how
compliance costs are treated by refiners, and how prices will behave in the future. This
Appendix presents several sensitivity analyses in which various model parameters are varied to
examine how different values for these parameters would affect model results. Four parameters
are examined:
• Scenario 1: alternative market supply and demand elasticity parameters
• Scenario 2: alternative ways to treat fuel market compliance costs
• Scenarios: alternative ways to project future gasoline prices
• Scenario 4: alternative social discount rates
The results of these sensitivity analyses are presented below. The results for the first two
scenarios are presented for 2015. The results for the other two scenarios are presented for 2007
through 203 5.
In general, varying the model parameters does not significantly change the estimated net
impacts on economic welfare. The estimated net surplus loss in 2015 for the program is about
$350.7 million. The net surplus losses (consumer plus producer) across the sensitivity analysis
scenarios are all about $350 million. The exceptions are the alternative fuel market compliance
cost scenarios. The results of those scenarios suggest the rule will result in a substantial
consumer loss that is expected to be captured by refiners in the form of excess profits and
resulting in a net gain for producers. In those cases, the net surplus losses are $322.8 million and
$333.9 million.
With regard to how the compliance costs are expected to be shared, the alternative fuel
market compliance cost scenarios result in significant wealth transfers from consumers to
producers. For the elasticity scenarios, even if expected net surplus losses are similar across
most scenarios, varying the model parameters has an impact on how costs would be distributed
between producers and consumers. Varying the supply elasticity in Scenario 1, for example,
results in the producer share of the gasoline fuel program varying from $34.5 million (9.1
percent) to $316.2 million (83.3 percent), compared to $172.5 million (45.5 percent) for the
primary analysis. Finally, the alternative gasoline prices in Scenario 3 do not substantially affect
the distribution of costs between consumers and producers.
13G.1 Scenario 1: Model Elasticity Parameters
The supply and demand price elasticities are key parameters in the EIM. They
characterize the behavioral responses of producers and consumers in the gasoline fuel and PFC
markets. Demand and supply elasticities measure the responsiveness of producers and
consumers to a change in price: how much the quantity demanded or supplied is expected to
13-74
-------
change. A detailed discussion regarding the estimation and selection of the elasticities used in
the EIM is provided in Appendix 13E. In this section we examine the impact of changes in the
selected values of the elasticity parameters, holding other parameters constant. The goal is to
determine whether alternative elasticity values significant alter the conclusions of the primary
analysis.
13G.1.1
Alternative Demand and Supply Elasticities
The values of the demand and supply elasticities for the gasoline fuel and PFC markets is
important because the distribution of regulatory costs depends on the relative supply and demand
elasticities used in the analysis. For example, consumers will bear less of the regulatory burden
of a program if they are more responsive to prices than producers (demand is relatively more
elastic). Similarly, producers will bear less of the regulatory burden if they are more responsive
(supply is relatively more elastic).
Table 13G.1-1 reports the upper- and lower-bound values of the values of the elasticity
parameters (supply and demand) used in this sensitivity analysis.
Table 13G.1-1. Sensitivity Analysis of the Supply and Demand Elasticities
Market/Parameter
Elasticity
Source
Lower Bound
Base Case
Upper Bound
Gasoline Market
Supply
Demand
Clean Air
Nonroad Diesel
rule61
Federal Trade
Commission 62
0.04
-0.10
0.24
-0.20
2.0
-0.40
Portable Fuel Container Market
Supply
Demand
EPA estimate
EPA estimate
0.7
N/A
1.5
-0.01
3.9
-0.25
For the gasoline market, the upper- and lower-bounds of the demand and supply
elasticities are those reported in the literature. It should be noted that these are these ranges do
not include long-run elasticity estimates. As explained in Section 13.2.3, the EIM uses an
intermediate time frame, during which producers have some resource immobility which may
cause them to suffer producer surplus losses. In the long run, in contrast, all factors of
production are variable and producers can adjust production in response to cost changes. This
allows them to shift more of the burden of the rule to consumers.
The elasticites for the PFC market are estimated econometrically. The sensitivity ranges
13-75
-------
are derived by estimating a 90 percent confidence interval around the estimated elasticities, using
the coefficient and standard error values from the econometric analysis (See Appendix 13E).
Because PFC expenditures are only such a small portion of total household production inputs,
households are not expected to switch their preferences for PFCs due to the standards. The
sensitivity analysis reflects a hypothetical assumption that 10 percent of demand is substituted
away from PFCs, a fairly large assumption since it is not clear what consumers would use
instead of PFCs for such a significant share of their consumption. This forms the upper bound of
the sensitivity analysis. Such a household behavioral change would increase the demand
elasticity for PFCs to -0.25 from -0.01. In other words, a 1.0 percent increase in the price of
PFCs will result in a 0.25 percent decrease in the quantity demanded.
13G.1.2 Results
The results of the sensitivity analysis for the demand and supply elasticities are reported
in Tables 13G.1-2 and 13G.1-3.
In the gasoline fuel case, price increases are the highest for the upper-bound supply
elasticity and lower-bound demand elasticity. In other words, when producers are more able to
respond to cost increases (more elastic supply elasticity) they can adjust their production and
pass more of the costs on to producers. Similarly, when consumers are less able to respond to
price increases (less elastic demand elasticity) they cannot reduce their demand and must
accommodate higher prices, resulting in their bearing more of the costs of the program. It is
important to note, however, that none of these estimated price increases are very large, with the
smallest being about 0.02 cent per gallon and the largest about 0.9 cent per gallon, as compared
to 0.08 to 0.54 cent per gallon in the primary case.
In the PFC case, changes in the elasticity parameters have no impacts on the price of
PFCs. This is not surprising given that the alternative elasticities are perfectly inelastic
(elasticity of zero) or very inelastic (elasticity of-0.25), meaning that consumers are not expected
to alter their purchases very much, if at all, in response to a change in price.
With regard to how the compliance costs of the program are distributed among producers
and consumers in the gasoline fuel market, producers bear a larger portion of the burden when
supply elasticity is less elastic (producers are less responsive to price changes) or the demand
elasticity is more elastic (consumers are more responsive to price changes), ranging from about
63 percent to 83 percent compared to the primary analysis of 45 percent. Similarly, consumers
bear a larger portion of the burden when the supply elasticity is more elastic (producers are more
responsive to price changes) or the demand elasticity is less elastic (consumers are less
responsive to price changes), ranging from 71 percent to 91 percent compared to the primary
analysis of about 55 percent.
In the PFC case, however, varying the demand and supply parameters does not vary the
results, with consumers expected to bear most of the burden across all cases. The sole exception
is the demand upper-bound, in which the consumer burden decreases from 99 percent in the
13-76
-------
primary case to 85 percent. Again, this is because the alternative elasticities are also highly
inelastic.
Finally, the overall expected social costs of the program across scenarios do not change,
and are always about $350 million.
Table 13G.1-2. Application Market Sensitivity Analysis for Supply Elasticities"'b
Supply Lower Bound
Base Case
Supply Upper Bound
Scenario
Gasoline Fuel
Price (0/q)
PADD I+III
PADDII
PADD IV
PADD V (w/out CA)
Change in Consumer
Surplus ($106/yr)
Change in Producer
Surplus ($106/yr)
Gas Cans
Price ($/q)
States w/Programs
States w/out
Programs
Change in Consumer
Surplus ($106/yr)
Change in Producer
Surplus ($106/yr)
Subtotal Social Costs
Fuel Savings
Vehicle Program
Total Social Costs
($106/yr)
Absolute
0.020
0.050
0.080
0.170
-$63.2
-$316.2
$0.20
$1.50
-$33.5
-$0.5
-$413.4
$75.5
-$12.9
-$350.8
Relative c
0.02%
0.03%
0.05%
0.10%
16.7%
83.3%
1.8%
32.2%
98.6%
1.4%
Absolute
0.080
0.170
0.270
0.540
-$207.0
-$172.5
$0.20
$1.52
-$33.7
-$0.2
-$413.4
$75.5
-$12.9
-$350.7
Relative c
0.05%
0.11%
0.17%
0.33%
54.5%
45.5%
1.9%
32.5%
99. 3%
0.7%
Absolute
0.140
0.280
0.460
0.910
-$344.9
-$34.5
$0.21
$1.52
-$33.9
-$0.1
-$413.3
$75.5
-$12.9
-$350.7
Relative c
0.09%
0.18%
0.29%
0.55%
90.9%
9.1%
1.9%
32.6%
99. 7%
0.3%
a Sensitivity analysis is presented for 2015.
b Figures are in 2003 dollars.
0 For "prices" rows the "relative" column refers to the relative change in price (with regulation) from the baseline
price. For "Surplus" rows, the "relative" column contains the percent distribution between consumer and
producer surplus.
13-77
-------
Table 13G.1-3. Application Market Sensitivity Analysis for Demand Elasticities"'b
Scenario
Gasoline Fuel
Price (0/q)
PADD I+III
PADDII
PADD IV
PADD V (w/out CA)
Change in Consumer
Surplus ($106/yr)
Change in Producer
Surplus ($106/yr)
Gas Cans
Price ($/q)
States w/Programs
States w/out
Programs
Change in Consumer
Surplus ($106/yr)
Change in Producer
Surplus ($106/yr)
Subtotal Social Costs
Fuel Savings
Vehicle Program
Total Social Costs
($106/yr)
Demand
Absolute
0.110
0.220
0.350
0.700
-$267.8
-$111.6
$0.21
$1.53
-$34.0
$0.0
-$413.4
$75.5
-$12.9
-$350.8
Lower Bound
Relative c
0.07%
0.14%
0.22%
0.42%
70.6%
29.4%
1.9%
32.7%
100.0%
0.0%
Base Case
Absolute Relative c
0.080 0.05%
0.170 0.11%
0.270 0.17%
0.540 0.33%
-$207.0 54.5%
-$172.5 45.5%
$0.20 1.9%
$1.52 32.5%
-$33.7 99.3%
-$0.2 0.7%
-$413.4
$75.5
-$12.9
-$350.7
Demand
Absolute
0.060
0.120
0.190
0.370
-$142.3
-$237.1
$0.18
$1.31
-$28.2
-$4.7
-$412.3
$75.5
-$12.9
-$349.7
Upper Bound
Relative c
0.04%
0.08%
0.12%
0.22%
37.5%
62.5%
1.6%
28.0%
85.7%
14.3%
a Sensitivity analysis is presented for 2015.
b Figures are in 2003 dollars.
0 For "prices" rows the "relative" column refers to the relative change in price (with regulation) from the baseline
price. For "Surplus" rows, the "relative" column contains the percent distribution between consumer and producer
surplus.
13G.2
Scenario 2: Fuel Market Compliance Costs
13G.2.1
Scenarios Modeled
13-78
-------
Section 13.2 discusses alternative approaches to shifting the supply curve in the market
model. Three alternatives for the fuel market supply shift are investigated in this sensitivity
analysis:
• Total average (variable + fixed) cost shift—the results presented in Section 13.1 and the
appendices are generated using this cost shift.
Total maximum (variable + fixed) cost shift
• Variable maximum cost shift
Figure 13G2-1 High Cost Producer Drives Price Increases
Mnax
High Cost Supplier
Aggregate Remaining
Suppliers
Fuel Market
While it may seem reasonable to estimate costs based on maximum variable or maximum
total costs, it should be noted that both of those scenarios assume that refiners with the highest
benzene compliance costs are also the highest-cost gasoline producers absent benzene control.
We do not have information on the highest gasoline cost producers to be able to examine
whether these refineries are also expected to have the highest benzene control costs. However,
we believe this is an extreme assumption.
To model the total and variable maximum cost scenarios, the high-cost producer is
represented by a separate supply curve as shown in Figure 13G-1. The remainder of the market
is represented as a single aggregate supplier. The high-cost producer's supply curve is then
shifted by Cmax (either total or variable), and the aggregate supply curve is shifted by Cagg.
Using this structure, the high-cost producer will determine price as long as
• the decrease in market quantity does not shut down the high-cost producer, and
13-79
-------
• the supply from aggregate producers is highly inelastic (i.e., remaining producers are
operating close to capacity); thus, the aggregate producers cannot expand output in
response to the price increase.
Note that the aggregate supply curve is no longer shifted by the average compliance costs
but slightly less than the average because the high-cost producer has been removed. The
adjusted average aggregate cost shift (Cagg) is calculated from the following:
c *n = c * n + c *
^ave Vtot '-'max Vmax ' ^agg
(13G.1)
where Cave is the average control cost for the total population; Qmax, Cmax, and Qagg, Cagg are the
baseline output and cost shift for the maximum cost producer; and the baseline output and cost
shift for the remaining aggregate producers, respectively.
13G.2.2
Compliance Costs
This analysis is based on the alternative compliance costs set out in Tables 13G.2-1 and
13G.2-2.
Table 13G.2-1 Gasoline Fuel Compliance Costs - Maximum Variable Cost Scenario by
Region (cVgallon, 2003$)
Year
2007
2008
2009
2010
2011
2012
2013
2014
2015+
PADD
1&3
0.3230
0.3230
0.3230
0.4240
0.4240
5.6700
5.6700
5.6700
5.6700
PADD 2
0.2430
0.2430
0.2430
0.4730
0.4730
3.5380
3.5380
3.5380
5.8900
PADD 4
0.6090
0.6090
0.6090
0.1760
0.1760
2.4640
2.4640
2.4640
5.6230
PADD 5
(w/out
California)
0.3340
0.3340
0.3340
0.3340
0.3340
3.3680
3.3680
3.3680
4.2900
13-80
-------
Table 13G.2-2. Gasoline Fuel Compliance Costs - Maximum Variable Cost Scenario
by Region (cVgallon, 2003$)
Year
2007
2008
2009
2010
2011
2012
2013
2014
2015+
PADD
1 &3
0.3230
0.3230
0.3230
0.3420
0.3420
4.5660
4.5660
4.5660
4.5660
PADD 2
0.2430
0.2430
0.2430
0.3510
0.3510
3.0180
3.0180
3.0180
4.4150
PADD 4
0.6090
0.6090
0.6090
0.6090
0.6090
2.0140
2.0140
2.0140
4.2710
PADD 5
(w/out
California)
0.3340
0.3340
0.3340
0.3340
0.3340
2.7530
2.7530
2.7530
3.3360
13G.2.3
Results
The results of the sensitivity analysis for the fuel compliance scenarios reported in Table
13G.2-1. According to these results, market prices are sensitive to changes in assumptions about
compliance costs. The way in which the cost burden is shared across producers and consumers
is also sensitive to changes in these assumptions.
With regard to prices, the Maximum Total Cost and Maximum Variable Cost scenarios
both lead to larger estimated price increases. In the primary case (Total Average Cost scenario),
prices are expected to increase between 0.08 to 0.54 cents per gallon, depending on the PADD.
In the Maximum Total Cost scenario, prices are expected to increase from 4.3 to 5.9 cents per
gallon. In the Maximum Variable Cost scenario, the estimated prices increases range from 3.3 to
4.4 cents per gallon.
With regard to how the burden is shared, both the Maximum Total Cost and Maximum
Variable Cost scenarios lead to a significant outcome: producers are expected to benefit from
the regulations and consumers are expected to experience a much larger surplus loss. In the
Maximum Total Cost scenario, producers would benefit by about $7,308 million, while
consumers surplus would decline by about $7,659 million. In the Maximum Variable Cost
scenario, producers would benefit by about $5,596 million and consumers surplus would decline
by about $5,958 million..
13-81
-------
Table 13G.2-3. Sensitivity Analysis to Cost Shifts in the Gasoline Fuel Market (2015)
a,b
Scenario
Gasoline Fuel
Price (0/q)
PADD I+III
PADDII
PADD IV
PADD V (w/out CA)
Change in Consumer
Surplus ($106/yr)
Change in Producer
Surplus ($106/yr)
Gas Cans
Price ($/q)
States w/Programs
States w/out
Programs
Change in Consumer
Surplus ($106/yr)
Change in Producer
Surplus ($106/yr)
Subtotal Social Costs
Fuel Savings
Vehicle Program
Total Social Costs
($106/yr)
Total Average Scenario
Absolute Relative0
0.080 0.05%
0.170 0.11%
0.270 0.17%
0.540 0.33%
-$207.0
-$172.5
$0.20 1.9%
$1.52 32.5%
-$33.7 99. 3%
-$0.2 0.7%
-$413.4
$75.5
-$12.9
-$350.7
Maximum Total Scenario
Absolute Relative0
5.30 3.6%
5.90 3.9%
5.60 3.6%
4.30 2.6%
-$7,659.0
$7,307.5
$0.20 1.9%
$1.52 32.5%
-$33.7 99. 3%
-$0.2 0.7%
-$385.5
$75.5
-$12.9
-$322.8
Maximum Variable
Scenario
Absolute Relative0
4.20 2.8%
4.40 2.9%
4.30 2.7%
3.30 2.0%
-$5,958.4
$5,595.8
$0.20 1.9%
$1.52 32.5%
-$33.7 99. 3%
-$0.2 0.7%
-$396.6
$75.5
-$12.9
-$333.9
a Sensitivity analysis is presented for 2015.
b Figures are in 2003 dollars.
c For "prices" rows the "relative" column refers to the relative change in price (with regulation) from the
price. For "Surplus" rows, the "relative" column contains the percent distribution between consumer and
surplus
baseline
producer
Under the base case (Total Average Cost scenario), refiners are expected to pass more
than half of the average compliance costs on to consumers, and the net decrease in producer
surplus for refiners is about $172.5 million, or 45 percent of the gasoline program social costs.
Under this scenario, prices are expected to increase less than 0.4 percent. Note that these are
industry averages, and individual refiners will gain or lose because compliance costs vary across
individual refineries.
In the Total Maximum Cost scenario, the highest operating cost refinery determines the
13-82
-------
new market price through the impacts on both fixed and variable costs. This refinery has the
highest per-unit supply shift, which leads to a higher price increase relative to the Total Average
Cost scenario. As a result, all refiners except the highest cost refiner are expected to benefit from
the rule, with an increase in producer surplus of about $7,308 million. This would occur because
the change in market price exceeds the additional per-unit compliance costs for most of the
refineries (i.e., most refiners have costs less than the costs for the highest operating cost
refinery). Consequently, in this scenario gasoline fuel consumers are expected to bear a larger
share of the total cost of the program: $7,659 million compared to $207 million in the base case.
The Variable Maximum Cost scenario is similar to the Total Maximum Cost scenario in
that the highest cost refinery determines the with-regulation market price. However, the
Variable Maximum Cost scenario leads to an expected price increase that is smaller than the
Total Maximum Cost scenario because the refiner supply shift includes only variable compliance
costs. In other words, the refiners do not pass along any fixed costs; they absorb the fixed costs.
Refiners also experience a net surplus gain in this scenario, about $5,596 million, because the
change in market price (driven by the Maximum Variable Cost supply curve shift) exceeds the
additional per-unit compliance costs for many refineries (i.e., many refiners still have total costs
less than the costs for the highest operating cost refinery in this scenario). The net surplus gain
for refiners is smaller than the Total Maximum Cost scenario ($5,596 million compared to
$7,308 million) because refiners absorb fixed costs, and the projected market price increase is
smaller. Again, gasoline fuel consumers are expected to bear a larger share of the total cost of the
program, about $5,958 million.
The results of this sensitivity analysis suggest that the expected impacts on producers and
consumers are affected by how refinery costs are modeled. In the ELM these costs are modeled
based on the Average Total Cost scenario (variable + fixed), reflecting a competitive market
situation in all regional markets. However, if the highest cost refinery drives the new market
price, then prices are expected to increase more (up to 3.9 percent in PADD 2) and output is
expected to contract more. In both of the maximum cost scenarios, gasoline fuel consumers are
expected to bear more than the cost of the rule and refiners will bear less than in the base case.
13G.3 Scenario 3: Alternative Gasoline Price
Appendix F discusses two ways to handle future prices in the Economic Impact Analysis.
The first assumes a constant (real) price of goods and services over time. The second approach
allows prices change over time.
The primary analysis reflects the first alternative, and prices are held constant. As
explained in Appendix F, this is a reasonable assumption because in a competitive market as
demand grows over time production capacity will also grow as existing firms expand or new
firms enter the market and eliminate any excess profit. If, however, capacity is constrained or if
the price of inputs increases, then prices may change over time. In this sensitivity analysis we
relax the constant price assumption and allow prices to change over time.
13-83
-------
This sensitivity analysis examines the constant price assumption for the gasoline fuel
market. We do not examine the impacts of relaxing the constant price assumption for the PFC
market because there are no publicly available price forecasts for that market. Gasoline price
forecasts are available through the Annual Energy Outlook's Reference (DoE 2006,
Supplemental Table 20).63 The AEO forecasted gasoline prices are national averages and are
reported in dollar per million btu. To compute prices per gallon, we convert the AEO price data
into an index (assume 2003 price as 1.00) and multiply this index by the appropriate 2003
baseline gasoline price. For example, the calculation for PADD II gasoline price in 2010 is:
2003 price ($/gallon) x 2010 AEO Price ($/million btu)/2003 AEO Price($/million btu)
= $1.51 x [16.52/13.31] = $1.87
The resulting indexes were applied to the individual PADD prices presented in Table
13.3-4 (2003 price multiplied by the index). The resulting price forecasts by PADD are
presented in Table 13G.3-1. Because the final year of the AEO projections is 2030, it is
necessary to estimate projected prices through 2035. This was done by applying a linear growth
rate based on the average annual growth Rate between 2021 and 2030.
Gasoline fuel forecast prices are presented in Figure 13G-2. This graph shows that
prices are initially expected to decrease from 2007 to about 2014, and then gradually increase
after 2014. The trends in fluctuations in gas prices reported in the AEO 2006 forecast have
changed when compared with the AEO 2005 forecasts (forecasted gasoline prices used in the
analysis). For example, annual growth in motor gasoline prices between 2003 and 2025 is higher
(0.6 versus -0.0). In addition, absolute gasoline prices are substantially higher (approximately 50
cent per gallon) in the latest forecast.
13-84
-------
Figure 13G3-lForecast Motor Fuel Prices (Includes Federal and State Taxes, 2003$)
210
205
_ 200
ra
o>
^
Q. 195
0)
0
190
185
180
2007 2009 2011 2013 2015 2017 2019 2021 2023 2025 2027 2029 2031 2033 2035
Year
13-85
-------
Table 13G.3-1. Forecast Gasoline Prices (2003$)
Year
Constant Price
(Primary Case)
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
PADD 1
&3
$1.48
PADD 2
$1.51
PADD 4
$1.57
PADD 5
$1.66
Forecast Prices
$2.02
$1.98
$1.92
$1.84
$1.84
$1.83
$1.82
$1.81
$1.82
$1.83
$1.84
$1.85
$1.87
$1.89
$1.91
$1.92
$1.93
$1.94
$1.95
$1.96
$1.96
$1.97
$1.98
$1.99
$2.00
$2.01
$2.02
$2.02
$2.03
$2.06
$2.02
$1.95
$1.87
$1.88
$1.87
$1.86
$1.85
$1.85
$1.87
$1.88
$1.89
$1.91
$1.93
$1.94
$1.96
$1.97
$1.98
$1.98
$2.00
$2.00
$2.01
$2.02
$2.03
$2.04
$2.05
$2.06
$2.06
$2.07
$2.14
$2.10
$2.03
$1.95
$1.95
$1.94
$1.93
$1.92
$1.93
$1.94
$1.95
$1.97
$1.98
$2.01
$2.02
$2.04
$2.05
$2.05
$2.06
$2.07
$2.08
$2.09
$2.10
$2.11
$2.12
$2.13
$2.14
$2.15
$2.15
$2.27
$2.22
$2.15
$2.06
$2.07
$2.05
$2.04
$2.03
$2.04
$2.05
$2.06
$2.08
$2.10
$2.12
$2.14
$2.15
$2.16
$2.17
$2.18
$2.19
$2.20
$2.21
$2.22
$2.24
$2.24
$2.25
$2.26
$2.27
$2.28
13-86
-------
The results of this sensitivity analysis are presented in Table 13G.3-2. Results are
reported for 2015, 2020, and 2030, for each PADD. These results suggest there is no measurable
difference between holding the price of gasoline constant or allowing it to vary in terms of the
impact of the standard on gasoline prices or in the distribution of social welfare costs among
producers and consumers of gasoline fuel. Relative gasoline price changes are slightly smaller
because the baseline price of gasoline in the variable price scenario is substantially higher. This
is not surprising, since the estimated compliance costs are the same for both the constant price
and variable price scenarios and are small, and the difference in fuel prices between the two
scenarios is small, less than five cents per gallon for all PADDs.
13-87
-------
Table 13G3.2. Sensitivity Analysis Constant and Variable Prices3
Scenario
Gasoline Fuel
Price (0/q)
PADD I+III
PADD II
PADD IV
PADD V (w/out
CA)
Change in
Consumer Surplus
($106/yr)
PADD I+III
PADD II
PADD IV
PADD V (w/out
CA)
Change in Producer
Surplus ($106/yr)
PADD I+III
PADD II
PADD IV
PADD V (w/out
CA)
Total Gasoline Fuel
Social Costs
2015
Constant Price
Absolute
Relative
Variable Price
Absolute
Relative
2020
Constant Price
Absolute
Relative
Variable Price
Relative
Relative
2030
Constant Price
Absolute
Relative
Variable Price
Absolute
Relative
0.080
0.170
0.270
0.540
-$66.3
-$75.9
-$14.5
-$50.3
-$55.3
-$63.2
-$12.1
-$41.9
-$379.4
0.05%
0.11%
0.17%
0.33%
17.5%
20.0%
3.8%
13.3%
14.6%
16.7%
3.2%
11.0%
100.0%
0.080
0.170
0.270
0.540
-$66.3
-$75.9
-$14.5
-$50.3
-$55.3
-$63.2
-$12.1
-$41.9
-$379.4
0.04%
0.09%
0.14%
0.27%
17.5%
20.0%
3.8%
13.3%
14.6%
16.7%
3.2%
11.0%
100.0%
0.080
0.170
0.270
0.540
-$70.4
-$80.5
-$15.4
-$53.4
-$58.6
-$67.1
-$12.8
-$44.5
-$402.6
0.05%
0.11%
0.17%
0.33%
17.5%
20.0%
3.8%
13.3%
14.6%
16.7%
3.2%
11.0%
100.0%
0.080
0.170
0.270
0.540
-$70.4
-$80.5
-$15.4
-$53.4
-$58.6
-$67.1
-$12.8
-$44.5
-$402.6
0.04%
0.09%
0.14%
0.26%
17.5%
20.0%
3.8%
13.3%
14.6%
16.7%
3.2%
11.0%
100.0%
0.080
0.170
0.270
0.540
-$77.9
-$89.1
-$17.0
-$59.1
-$64.9
-$74.3
-$14.2
-$49.2
-$445.8
0.05%
0.11%
0.17%
0.33%
17.5%
20.0%
3.8%
13.3%
14.6%
16.7%
3.2%
11.0%
100.0%
0.080
0.170
0.270
0.540
-$77.9
-$89.1
-$17.0
-$59.1
-$64.9
-$74.3
-$14.2
-$49.2
-$445.8
0.04%
0.08%
0.13%
0.24%
17.5%
20.0%
3.8%
13.3%
14.6%
16.7%
3.2%
11.0%
100.0%
a Figures are in 2003 dollars.
bFor "prices" rows the "relative" column refers to the relative change in price (with regulation) from the baseline price. For "Surplus" rows, the "relative"
column contains the percent distribution between consumer and producer surplus
13-Ł
-------
13-89
-------
13G.4
Scenario 4: Alternative Social Discount Rates
Future benefits and costs are commonly discounted to account for the time value of
money. Pursuant to Circular A-4, we provide present value estimates using real discount rates of
3 percent and 7 percent, in Table 13G.4-1. According to OMB Circular A-4, "the 3 percent
discount rate represents the 'social rate of time preference'... [which] means the rate at which
'society' discounts future consumption flows to their present value"; "the seven percent rate is an
estimate of the average before-tax rate of return to private capital in the U.S. economy ... [that]
approximates the opportunity cost of capital."64 The net present value of the social costs through
2035 using the 3 percent discount rate is $5,354 million. Using a seven percent social discount
rate, the present value of total social costs is $2,900 million.
Table 13G.4-1. Net Present Value of Cumulative Estimated Social Costs Through 2035
(discounted to 2006; Smillion; 2003$)
Market
Gasoline, U.S.
PADD 1 & 3
PADD2
PADD 4
PADD 5 (w/out
CA)
Portable Fuel
Containers US
States with
existing programs
States without
existing programs
Subtotal
Fuel Savings
Vehicle Program
Total
Change in
Consumer
Surplus
Change in
Producer
Surplus
Total
Net Present Value 3%
-$959.7
-$1,260.4
-$210.8
-$684.5
-$78.7
-$676.2
-$3870.3
59.8%
-$799.8
-$1,050.4
-$175.6
-$570.4
-$0.5
-$4.5
-$2,601.2
40.2%
-$1,759.5
-$2,310.8
-$386.4
-$1,254.8
-$79.3
-$680.7
-$6,471.6
$1,208.0
-$91.1
-$5,354.6
Change in
Consumer
Surplus
Change in
Producer
Surplus
Total
Net Present Value 7%
-$499.2
-$699.6
-$109.6
-$343.7
-$50.7
-$399.8
-$2,102.7
60.4%
-$416.0
-$583.0
-$91.3
-$286.5
-$0.3
-$2.7
-$1,379.8
39.6%
-$915.3
-$1,282.6
-$200.9
-$630.2
-$51.1
-$402.5
-$3,482.5
$647.3
-$64.6
-$2,899.8
13-90
-------
References
1 U.S. EPA "EPA Guidelines for Preparing Economic Analyses, EPA 240-R-00-003." September
2000, p 113. A copy of this document can be found at
http://yosemite.epa.gov/ee/epa/eed.nsf/webpages/Guidelines.html
2 See Table A-2 and Supplemental Table 34 of U.S. Department of Energy, Energy Information
Administration "Annual Energy Outlook 2004 with projections to 2025.". DOE/EIA-0383. 2004.
A copy of this document is available at http://www.eia.doe.gov/oiaf/archive/aeo04/index.html
3 California Environmental Protection Agency, Air Resources Board "Hearing Notice and Staff
Report; Initial Statement of Reasons for Proposed Rule Making Public Hearing to Consider the
Adoption of Portable Fuel Container Spillage Control Regulation." Sacramento, CA: California
Environmental Protection Agency, Air Resources Board (August 6, 1999). A copy of this
document is available at http://www.arb.ca.gov/regact/spillcon/isor.pdf
4 Research Triangle Institute "Characterizing Gasoline Markets: A Profile." Prepared for the U.S.
EPA by RTI. August, 2004. EPA Contract No. 68-D-99-024.
5 Research Triangle Institute "Characterizing Gas Can Markets: A Profile." Prepared for the U.S.
EPA by RTI. August, 2004. EPA Contract No. 68-D-99-024.
6 U.S. EPA "OAQPS Economic Analysis Resource Document." Research Triangle Park, NC:
EPA 1999 A copy of this document can be found at
http://www.epa.gov/ttn/ecas/econdata/6807-305.pdf
7 Bingham, T.H., and T. J. Fox "Model Complexity and Scope for Policy Analysis." Public
Administration Quarterly 23(3). 1999.
8 Berck, P., and S. Hoffmann "Assessing the Employment Impacts." Environmental and
Resource Economics 22:133-156. 2002.
9 Office Management and Budget "Executive Analysis of Federal Regulations Under Executive
Order 12866." Executive Office of the President, Office Management and Budget. January 11,
1996. A copy of this document is available at
http://www.whitehouse.gov/omb/inforeg/print/riaguide.html
10 U.S. EPA "EPA Guidelines for Preparing Economic Analyses, EPA 240-R-00-003."
September 2000, p 126. A copy of this document can be found at
http://yosemite.epa.gov/ee/epa/eed.nsf/webpages/Guidelines.html
11 Research Triangle Institute "Characterizing Gasoline Markets: A Profile." Prepared for the
U.S. EPA by RTI. August, 2004. EPA Contract No. 68-D-99-024.
13-91
-------
12 Baumol. W. "Contestable Markets: An Uprising in the Theory of Industry Structure."
American Economic Review 72:1-15. 1982.
13 Baumol, W., J. Panzer, and R. Willig "Contestable Markets and the Theory of Industry
Structure." San Diego, CA: Harcourt, Brace, Jovanovich. 1982.
14 Research Triangle Institute "Characterizing Gas Can Markets: A Profile." Prepared for the
U.S. EPA by RTI. August, 2004. EPA Contract No. 68-D-99-024.
15 U.S. EPA "OAQPS Economic Analysis Resource Document." Research Triangle Park, NC:
EPA 1999. A copy of this document can be found at
http://www.epa.gov/ttn/ecas/econdata/6807-305.pdf
16 Research Triangle Institute "Characterizing Gasoline Markets: A Profile." Prepared for the
U.S. EPA by RTI. August, 2004. EPA Contract No. 68-D-99-024.
17 Research Triangle Institute. Characterizing Gas Can Markets: A Profile. Prepared for the U.S.
EPA by RTI. August, 2004. EPA Contract No. 68-D-99-024.
18 California Environmental Protection Agency, Air Resources Board "Hearing Notice and Staff
Report, Initial Statement of Reasons for Proposed Rule Making Public Hearing to Consider the
Adoption of Portable Fuel Container Spillage Control Regulation." Sacramento, CA: California
Environmental Protection Agency, Air Resources Board (August 6, 1999). A copy of this
document is available at http://www.arb.ca.gov/regact/spillcon/isor.pdf
19 Hicks, J.R. "Marshall's Third Rule: A Further Comment." Oxford Economic Papers. 13:262-
65, 1961.
20 Hicks, J.R. The Theory of Wages. 2nd Ed. New York: St. Martin's Press 1966.
21 Allen, R.G.D. Mathematical Analysis for Economists. New York: St. Martin's Press 1938.
22 California Environmental Protection Agency, Air Resources Board "Hearing Notice and Staff
Report, Initial Statement of Reasons for Proposed Rule Making Public Hearing to Consider the
Adoption of Portable Fuel Container Spillage Control Regulation." Sacramento, CA: California
Environmental Protection Agency, Air Resources Board (August 6, 1999). A copy of this
document is available at http ://www. arb. ca. gov/regact/spillcon/i sor. pdf
23 U.S. Department of Transportation, Federal Highway Administration. "Highway Statistics
2002." A copy of this document can be found at
http://www.fhwa.dot.gov/policy/ohim/hs02/index.htm
13-92
-------
24 Final Report of the Federal Trade Commission, Midwest Gasoline Price Investigation, March
29, 2001. A copy of this report is available at http://www.ftc.gov/os/2001/03/mwgasrpt.htm
25 See Section 4 of Research Triangle Institute "Characterizing Gas Can Markets: A Profile."
Prepared for the U.S. EPA by RTI. August, 2004. EPA Contract No. 68-D-99-024.
26 See Table 48 of U.S. Department of Energy, Energy Information Administration Petroleum
Marketing Annual, August 2004 (August 2005). DOE/EIA-0487-2004. A copy of this document
is available at
http://www.eia.doe.gov/pub/oil_gas/petroleum/data_publications/petroleum_marketing_annual/h
istorical/2004/pdf/pmatab48 .pdf
27 U.S. Department of Energy, Energy Information Administration. Annual Energy Outlook 2005
with Projections to 2025, Report #: DOE/EIA-0383 (2005). A copy of this document is available
at http://www.eia.doe.gov/oiaf/archive/aeo05/results.html
28 Research Triangle Institute. Memo to Chi Li, U.S. EPA, from Brooks Depro, RTI,
"Calculation of Motor Gasoline Prices in MSAT-EEVI." December 26, 2005. Prepared for the
U.S. EPA by RTI. EPA Contract No. 68-D-99-024.
29 See Table 31 of U.S. Department of Energy, Energy Information Administration. Petroleum
Marketing Annual, August 2004. DOE/EIA-0487-2003. A copy of this document is available at
http://tonto.eia.doe.gov/FTPROOT/petroleum/048703.pdf
30 See Table MF-121T of U.S. Department of Transportation, Federal Highway Administration.
"Highway Statistics 2003." A copy of this document can be found at
http://www.fhwa.dot.gov/policy/ohim/hs03/index.htm
31 See Table 31 of U.S. Department of Energy, Energy Information Administration. Petroleum
Marketing Annual, August 2004. DOE/EIA-0487-2003. A copy of this document is available at
http://tonto.eia.doe.gov/FTPROOT/petroleum/048703.pdf
32 Considine, T. J. 2002. "Inventories and Market Power in the World Crude Oil Market."
Working paper, Department of Energy, Environmental, and Mineral Economics, The
Pennsylvania State University, University Park, PA. A copy of this document is available at
http://www.personal.psu.edU/faculty/c/p/cpw/resume/InventoriesMarketPowerinCrudeOilMarket
s.pdf
33 Bartlesman, E., R. Becker, and W. Gray. NBER-CES Manufacturing Industry Database. 2000.
A copy of this document can be found at http://www.nber.org/nberces/nbprod96.htm
13-93
-------
34 Federal Trade Commission. Final Report of the Federal Trade Commission: Midwest
Gasoline Price Investigation (March 29, 2001). A copy of this document is available at
http: //www.ftc.gov/os/2001/03/mwgasrpt.htm
35 Berck, P., and S. Hoffmann. 2002. "Assessing the Employment Impacts." Environmental and
Resource Economics 22:133-156.
36 Fullerton, D. and G. Metcalf March 2002. "Tax Incidence." NBER Working Paper 8829.
http://www.nber.org/papers/w8829. As obtained June 18, 2004.
37 Fullerton, D. and G. Metcalf. March 2002. "Tax Incidence." NBER Working Paper 8829.
http://www.nber.org/papers/w8829. As obtained June 18, 2004.
38 Fullerton, D. and G. Metcalf. March 2002. "Tax Incidence." NBER Working Paper 8829.
http://www.nber.org/papers/w8829. As obtained June 18, 2004.
39 Considine, T. J. 2002. "Inventories and Market Power in the World Crude Oil Market."
Working paper, Department of Energy, Environmental, and Mineral Economics, The
Pennsylvania State University, University Park, PA. A copy of this document is available at
http://www.personal.psu.edU/faculty/c/p/cpw/resume/InventoriesMarketPowerinCrudeOilMarket
s.pdf
40 Bartlesman, E., R. Becker, and W. Gray. NBER-CES Manufacturing Industry Database. 2000.
A copy of this document can be found at http://www.nber.org/nberces/nbprod96.htm
41 Federal Trade Commission. Final Report of the Federal Trade Commission: Midwest
Gasoline Price Investigation (March 29, 2001). A copy of this document is available at
http://www.ftc.gov/os/2001/03/mwgasrpt.htm.
42 Finizza, A. Economic Benefits of Mitigating Refinery Disruptions: A Suggested Framework
and Analysis of a Strategic Fuels Reserve. Study conducted for the California Energy
Commission pursuant to California State Assembly Bill AB 2076. (P600-02-018D, July 4,
2002). A copy of this document is available at
http://www.energv.ca.gov/reports/2002-07-08 600-02-018D.PDF.
43 Federal Trade Commission. "Final Report of the Federal Trade Commission: Midwest
Gasoline Price Investigation," March 29, 2001. A copy of this document is available at
http://www.ftc.gov/os/2001/03/mwgasrpt.htm.
44 Greene, D.L. and N.I. Tishchishyna "Costs of Oil Dependence: A 2000 Update." Study
prepared by Oak Ridge National Laboratory for the U.S. Department of Energy under contract
DE-AC05-OOOR22725. ORNL/TM-2000/152. May 2000. This document can be accessed at
http://www.ornl.gov/~webworks/cpr/v823/rpt/107319.pdf.
13-94
-------
45 Considine, T. J. "Inventories and Market Power in the World Crude Oil Market." Working
paper, Department of Energy, Environmental and Mineral Economics. Pennsylvania State
University, University Park, PA. 2002. Copy of this document available at
http://www.personal.psu.edU/facultv/c/p/cpw/resume/InventoriesMarketPowerinCrudeOilMarket
s.pdf.
46 See pages 3-19 of U.S. EPA "Industry Profile for the Petroleum Refinery NESHAP." Draft for
EPA by Methtech and Pechan & Associates. February 1997. EPA Contract No. 68-D4-0107,
WA No. 11-17. A copy of this document can be found at
http://www.epa.gov/ttnecasl/regdata/IP s/Petroleum%20Refinerv%20(Sulfur%20Recovery%20U
nits.%20Catalvtic%20Crackin.pdf
47 Federal Trade Commission. Final Report of the Federal Trade Commission: Midwest
Gasoline Price Investigation. March 29, 2001. A copy of this document is available at
http://www.ftc.gov/os/2001/03/mwgasrpt.htm.
48 Graham, D. and S. Glaister. "The Demand for Automobile Fuel: A Survey of Elasticities."
Journal of Transport Economics and Policy 36:1-26, 2002.
49 Espey, M. "Gasoline Demand Elasticities Revisited: An International Meta-Analysis of
Elasticities." Energy Economics 20:273-295, 1998.
50 See page 57 of Chouinard, H. and J.M. Perl off "Incidence of Federal and State Gasoline
Taxes." Economics Letters 83:55-60, 2004.
51 Hicks, J.R. "Marshall's Third Rule: A Further Comment." Oxford Economic Papers. 13:262-
65, 1961.
52 Hicks, J.R. The Theory of Wages. 2nd Ed. New York: St. Martin's Press 1966.
53 Allen, R.G.D. Mathematical Analysis for Economists. New York: St. Martin's Press 1938.
54 Bartlesman, E., R. Becker, and W. Gray. NBER-CES Manufacturing Industry Database. 2000.
A copy of this document can be found at http://www.nber.org/nberces/nbprod96.htm.
55 See Table 1.1.9 of U.S. Bureau of Economic Analysis "Implicit Price Deflators for Gross
Domestic Product," BEA Quarterly., 2002 through 2004. A copy of this information can be found
at http://www.bea.gov/bea/dn/nipaweb/SelectTable.asp. All values are expressed in $1987.
Please note that the GDP deflators have been updated since the previous estimation of engine
and lawn and garden supply elasticities used for the Clean Air Nonroad Diesel rule (see Chapter
10 of the Final Regulatory Analysis for the Clean Air Nonroad Diesel Rule, EPA 420-R-04-007,
May 2004; http://www.epa.gov/nonroad-diesel/2004fr/420r04007.pdf) As a result, the supply
elasticity estimates are the same; however, the coefficient estimates may vary slightly.
13-95
-------
56 NBER-CES. National Bureau of Economic Research and U.S. Census Bureau, Center for
Economic Research. 2002. NBER-CES Manufacturing Industry Database, 1958-1996.
http://www.nber.org/nberces/nbprod96.htm.
57 See Table 1.1.9 of U.S. Bureau of Economic Analysis "Implicit Price Deflators for Gross
Domestic Product," BEA Quarterly, 2002 through 2004. A copy of this information can be found
at http://www.bea.gov/bea/dn/nipaweb/SelectTable.asp.
58 U.S. Consumer Product Safety Commission "Report on the Safety of Portable Fuel Containers
(Gas Cans)." Washington, DC: U.S. Consumer Product Safety Commission. 2003. This
document is available at http://www.cpsc.gov/LIBRARY/FOIA/FOIA03/os/Gascans.pdf.
59 U.S. Environmental Protection Agency (EPA 2004). Final Regulatory Analysis for the Clean
Air Nonroad Diesel Rule, EPA 420-R-04-007, May 2004; see Chapter 10. This document is
available at http://www.epa.gov/nonroad-diesel/2004fr/420r04007.pdf
60 See pages 5-25 of U.S. EPA "OAQPS Economic Analysis Resource Document." Research
Triangle Park, NC: EPA 1999 A copy of this document can be found at
http://www.epa.gov/ttn/ecas/econdata/6807-305.pdf
61 See Chapter 10 of the Final Regulatory Analysis for the Clean Air Nonroad Diesel Rule, EPA
420-R-04-007, May 2004; http://www.epa.gov/nonroad-diesel/2004fr/420r04007.pdf
62 Federal Trade Commission. Final Report of the Federal Trade Commission: Midwest
Gasoline Price Investigation. March 29, 2001. A copy of this document is available at
http://www.ftc.gov/os/2001/03/mwgasrpt.htm.
63 See Appendix A of U.S. Department of Energy, Energy Information Administration. Annual
Energy Outlook 2005 with Projections to 2025, Report #: DOE/EIA-0383. 2005. A copy of this
document is available at http://www.eia.doe.gov/oiaf/archive/aeo05/results.html.
64Office of Management and Budget (OMB). Circular A-4 September 17, 2003. Pg33.
13-96
-------
Final Regulatory Impact Analysis
Chapter 14: Table of Contents
CHAPTER 14: Small-Business Flexibility Analysis 2
14.1 Overview of the Regulatory Flexibility Act 2
14.2 Need for the Rulemaking and Rulemaking Objectives 3
14.3 Definition and Description of Affected Entities 3
14.3.1 Description of Highway Light-Duty Vehicle Manufacturers 4
14.3.1.1 Vehicle Manufacturers 5
14.3.1.2 Independent Commercial Importers 5
14.3.1.3 Alternative Fuel Vehicle Converters 6
14.3.2 Description of Gasoline Refiners 6
14.3.3 Description of Portable Fuel Container Manufacturers 7
14.4 Issues Raised by Public Comments 7
14.5 Projected Reporting, Recordkeeping, and Other Compliance Requirements of the
Regulation 9
14.6 Steps to Minimize Significant Economic Impact on Small Entities 9
14.6.1 Regulatory Alternatives and Hardship Provisions for Highway Light-Duty Vehicle
Manufacturers 10
14.6.1.1 Panel Recommendations 10
14.6.1.2 What We Proposed 11
14.6.1.3 Provisions Being Finalized in this Rule 12
14.6.2 Regulatory Alternatives and Hardship Provisions for Gasoline Refiners 13
14.6.2.1 Panel Recommendations 13
14.6.2.2 What We Proposed 15
14.6.2.3 Provisions Being Finalized in This Rule 16
14.6.3 Portable Fuel Container Manufacturers 18
14.6.3.1 Panel Recommendations 18
14.6.3.2 What We Proposed 20
14.6.3.3 Provisions Being Finalized in This Rule 20
14.7 Related Federal Rules 20
14.8 Conclusions 21
14-1
-------
Final Regulatory Impact Analysis
CHAPTER 14: Small-Business Flexibility Analysis
This chapter discusses our Final Regulatory Flexibility Analysis, which evaluates the
potential impacts of new standards on small entities. The Regulatory Flexibility Act, as amended
by the Small Business Regulatory Enforcement Fairness Act of 1996 (SBREFA), generally
requires an agency to prepare a regulatory flexibility analysis of any rule subject to notice-and-
comment rulemaking requirements under the Administrative Procedure Act or any other statute,
unless the agency certifies that the rule will not have a significant economic impact on a
substantial number of small entities. Prior to issuing a proposal for this rulemaking, we analyzed
the potential impacts of these regulations on small entities. As a part of this analysis, we
convened a Small Business Advocacy Review Panel (SBAR Panel, or 'the Panel'). During the
Panel process, we gathered information and recommendations from Small Entity Representatives
(SERs) on how to reduce the impact of the rule on small entities, and those comments are
detailed in the Final Panel Report which is located in the public record for this rulemaking
(Docket EPA-HQ-OAR-2005-0036).
14.1 Overview of the Regulatory Flexibility Act
In accordance with section 609(b) of the Regulatory Flexibility Act, we convened an SBAR
Panel before conducting the Regulatory Flexibility Analysis. A summary of the Panel's
recommendations can be found in our proposal. Further, the Final Panel Report contains a
detailed discussion of the Panel's advice and recommendations (as well as the SER
recommendations). The regulatory alternatives that are being adopted in this final rule are
described below.
Section 609(b) of the Regulatory Flexibility Act further directs the Panel to report on the
comments of small entity representatives and make findings on issues related to identified
elements of the Regulatory Flexibility Analysis under section 603 of the Regulatory Flexibility
Act. Key elements of a Regulatory Flexibility Analysis are:
- a description and, where feasible, an estimate of the number of small entities to which
the proposed rule applies;
projected reporting, record keeping, and other compliance requirements of the
proposed rule, including an estimate of the classes of small entities that would be
subject to the rule and the type of professional skills necessary to prepare reports or
other records;
an identification, to the extent practicable, of all other relevant federal rules that may
duplicate, overlap, or conflict with the proposed rule;
- any significant alternatives to the proposed rule that accomplish the stated objectives
of applicable statutes and that minimize any significant economic impact of the
proposed rule on small entities.
The Regulatory Flexibility Act was amended by SBREFA to ensure that concerns regarding
small entities are adequately considered during the development of new regulations that affect
those entities. Although we are not required by the Clean Air Act to provide special treatment to
14-2
-------
Final Regulatory Impact Analysis
small businesses, the Regulatory Flexibility Act requires us to carefully consider the economic
impacts that our rules may have on small entities. The recommendations made by the Panel may
serve to help lessen these economic impacts on small entities when consistent with Clean Air Act
requirements.
14.2 Need for the Rulemaking and Rulemaking Objectives
A detailed discussion on the need for and objectives of this rule are located in the preamble
to the final rule. As previously stated, controlling emissions from light-duty highway vehicles,
gasoline, and portable fuel containers has important public health and welfare benefits.
Section 202(1)(2) of the Clean Air Act (CAA) authorizes EPA to promulgate standards to
control emissions of mobile source air toxics (MSATs) from new motor vehicles and fuels.
Specifically, this section states that EPA must:
...promulgate (and from time to time revise) regulations under subsection (a)(l) or section
211(c)(l) containing reasonable requirements to control hazardous air pollutants from
motor vehicles and motor vehicle fuels. The regulations shall contain standards for such
fuels or vehicles, or both, which the Administrator determines reflect the greatest degree of
emission reduction achievable through the application of technology which will be
available, taking into consideration the standards established under subsection (a), the
availability and costs of the technology, and noise, energy, and safety factors, and lead
time....The regulations shall, at a minimum, apply to emissions of benzene and
formaldehyde.
Thus, EPA must determine the maximum amount of emission reduction possible through
application of technology, and further assess the reasonableness of these reductions after
considering cost, lead time, and the other enumerated factors. Controls on NMHC (a surrogate
for organic mobile source air toxics) for light-duty vehicles, and benzene emissions from
gasoline, implement this provision. In addition, many prior rules (including the Tier 2 standards
and the highway and nonroad diesel engine standards) control toxics emitted by motor vehicles.
In addition, section 183(e) directs EPA to study, list, and regulate consumer and
commercial products that are significant sources of VOC emissions. The final rule for portable
fuel containers implements this provision. Regulations under section 183(e) must require the
"best available control," considering technological and economic feasibility and health,
environmental, and energy impacts.
14.3 Definition and Description of Affected Entities
Small entities include small businesses, small organizations, and small governmental
jurisdictions. For the purposes of assessing the impacts of the proposed rule on small entities, a
small entity is defined as: (1) a small business that meets the definition for business based on the
Small Business Administration's (SBA) size standards (see Table 14.3-1); (2) a small
governmental jurisdiction that is a government of a city, county, town, school district or special
district with a population of less than 50,000; and (3) a small organization that is any not-for-
14-3
-------
Final Regulatory Impact Analysis
profit enterprise which is independently owned and operated and is not dominant in its field.
Table 14.3-1 provides an overview of the primary SBA small business categories potentially
affected by this regulation.
The following sections discuss the small entities directly regulated by this final rule—namely
light-duty manufacturers, gasoline fuel refiners, and portable fuel container manufacturers. We
conducted preliminary industry profiles to identify the universe of small entities in each sector.
Table 14.3-1. Small Business Definitions
Industry
Light-duty vehicles:
- vehicle manufacturers (including
small volume manufacturers)
- independent commercial importers
- alternative fuel vehicle converters
Gasoline fuel refiners
Portable Fuel Container
Manufacturers:
- plastic container manufacturers
- metal fuel container manufacturers
Defined as small entity
by SBA if less than or
equal to:
1,000 employees
$6 million annual sales
100 employees
1,000 employees
$6 million annual sales
1,500 employees b
500 employees
1,000 employees
NAICSa Codes
336111
811111,811112,811198
424720
335312
811198
324110
326199
332431
a North American Industrial Classification System
b We have included in past fuels rulemakings a provision that, in order to qualify for the small refiner flexibilities, a
refiner must also have a company-wide crude refining capacity of no greater than 155,000 barrels per calendar day.
We have included this criterion to qualify for the small refiner provisions for this program as well.
14.3.1 Description of Highway Light-Duty Vehicle Manufacturers
To assess how many small entities would be directly affected by the rule, EPA first created a
database comprised of firms specified in its Certification and Fuel Economy Information System
(CFEIS) and EPA's independent commercial importers (ICIs) and converters lists. Sales and
employment data for the parent companies of these firms was then found using the Dunn and
Bradstreet (and Hoover's) and ReferenceUSA databases. Due to the range of manufacturers and
ICIs, there are several NAICS codes in which these businesses report their sales, but the majority
of the manufacturers and ICIs are listed under the following major groups, respectively: 3361 Ix -
Automobile and Light Duty Motor Vehicle Manufacturing and Slllxx - Automotive Repair and
Maintenance. For alternative fuel converters, there did not appear to be a prominent NAICS
code, and the codes range from 3 3 5 312 - Motor and Generator Manufacturing (and/or 336312-
Gasoline Engine and Engine Parts Manufacturing) to 811198- All Other Automotive Repair and
Maintenance.
14-4
-------
Final Regulatory Impact Analysis
Based on the preliminary industry characterization, we identified a total of about 50
businesses that would be covered by the new light-duty vehicle standards. However, due to a
lack of sales or employment data, a few of these entities could not be confirmed for
consideration in EPA's analysis. Out of these 50 businesses, 21 entities (or 42 percent) fit the
SBA criterion of a small business. EPA estimates that these entities comprise about 0.02 percent
of the total light-duty vehicle sales in the U.S. for the year 2004.A
In addition to major vehicle manufacturers, three distinct categories of businesses
characterize the above 50 total entities (and the subset of 21 small businesses): small volume
manufacturers (SVMs), ICIs, and alternative fuel vehicle converters. The below discussion gives
more detail on these categories.
14.3.1.1 Vehicle Manufacturers
In most cases, new standards for light-duty vehicles would minimally increase the costs
of vehicle manufacturers to produce these vehicles. In addition to major vehicle manufacturers,
SVMs are companies that sell less than 15,000 vehicles per year, as defined in past EPA
regulations, and this status allows vehicle models to be certified under a slightly simpler
certification process.
Using information from a preliminary assessment of the industry, EPA identified a total
of 30 businesses that manufacture vehicles (including about 14 SVMs). The top 10 vehicle
manufacturers comprise 97 percent of the U.S. total market (there were about 16.9 million total
U.S. sales for the year 2004), while the other 20 manufacturers (including SVMs), ICIs, and
converters make up the remaining 3 percent. Of the 30 manufacturers (14 SVMs included), 5
SVMs fit the SBA definition of a small entity. These five small businesses comprise about 0.01
percent of the total vehicle sales for the year 2004. Also, these businesses produce vehicles for
small niche markets, and nearly all of these entities manufacture limited production, high
performance cars. In addition, there are four other SVMs that EPA believes meet the SBA
small-entity criterion, but since they are foreign businesses, they cannot be considered in the
SBREF A work.
14.3.1.2 Independent Commercial Importers
ICIs are companies that hold a Certificate (or Certificates) of Conformity permitting them
to import nonconforming vehicles and to modify these vehicles to meet U.S. emission standards.
ICIs are not required meet the emission standards in effect when the vehicle is modified, but
instead they must meet the emission standards in effect when the vehicle was originally produced
(with an annual production cap of a total of 50 light-duty vehicles and trucks).8 ICIs would
likely have minimal increased cost from the new standards.
A Sales information used for this analysis was 2004 data.
To prevent entities from circumventing Tier 2 light-duty vehicle standards, EPA capped at 50 each Id's annual
production of vehicles meeting the original production (OP) year standards when OP year standards are less stringent than
standards that apply during the year of modification. This does not impact the number of vehicles an ICI may produce that
are certified to the standards that apply during the year of modification.
14-5
-------
Final Regulatory Impact Analysis
Currently 10 ICIs hold EPA certificates, and EPA believes all 10 of these businesses
would meet the small-entity criteria as defined by SB A. In 2004, collectively they had total U.S.
sales of about 300 vehicles, and thus, they comprised about 0.002 percent of the total vehicle
sales. ICIs modify vehicles for a small niche market, and many of these vehicles are high
performance cars.
14.3.1.3 Alternative Fuel Vehicle Converters
Alternative fuel vehicle converters are businesses that convert gasoline or diesel vehicles
to operate on alternative fuel (e.g., compressed natural gas), and converters must seek a
certificate for all of their vehicle models. Model year 1993 and newer vehicles that are
converted are required to meet the standards applicable at the time the vehicle was originally
certified. Converters would likely have minimal increased cost from the new light-duty vehicle
standards.
As with SVMs and ICIs, converters serve a small niche market, and these businesses
primarily convert vehicles to operate on compressed natural gas (CNG) and liquefied petroleum
gas (LPG), on a dedicated or dual fuel basis. Based on information from a preliminary
assessment, EPA identified a total of 10 alternative fuel vehicle converters. Together these 10
businesses had about 0.02 percent of the total vehicle sales in the U.S. for the year 2004. Out of
these 10 businesses, 6 meet the SB A small-entity criteria. These 6 converters represent about
0.01 percent of the total vehicle sales. In addition, EPA believes three of the other converters fit
the SB A small-entity definitions, but since they are foreign businesses, they cannot be
considered in the SBREFA work.
14.3.2 Description of Gasoline Refiners
Information about the characteristics of gasoline refiners comes from sources including
the Energy Information Administration within the U.S. Department of Energy, oil industry
literature, and industry searches using Hoover's and Dun and Bradstreet. These refiners fall
under the Petroleum Refineries category, NAICS code 324110.
Using our preliminary industry characterization, coupled with 2003 gasoline production
data, we believe that there are about 116 domestic refineries producing gasoline (however, due to
a lack of publicly available sales or employment data, some of these entities could not be
confirmed for consideration in the analysis). Our current assessment is that 14 refiners, owning
16 refineries, meet SBA's employee count criterion of having 1,500 employees or less. Due to
dynamics in the refining industry (i.e., mergers and acquisitions) and decisions by some refiners
to enter or leave the gasoline market, the actual number of refiners producing gasoline (and, thus,
the number of small refiners that ultimately qualify for small refiner status under this program)
could be much different than these estimates.
14-6
-------
Final Regulatory Impact Analysis
14.3.3 Description of Portable Fuel Container Manufacturers
For manufacturers of portable fuel containers, the SB A size thresholds are 500 employees
for manufacturers of plastic containers and 1,000 employees for metal fuel containers. The
NAICS codes are 326199 - All Other Plastics Product Manufacturing and 332431 - Metal Can
Manufacturing. Discussions with industry and searches in databases such as LexisNexis
Academic and ReferenceUSA (electronic resources) enabled EPA to determine how many
businesses would be impacted by the proposed rule and may meet the small-entity criteria. The
latter two sources provided sales and employment data for the parent companies of these
businesses.
As discussed earlier, annual sales nationwide of portable fuel containers are about 21
million units. 98 percent are plastic containers, and 2 percent are metal. Blow molding
equipment is relatively costly and large production volumes are necessary to operate profitably.
These factors seem to limit the number of companies engaged in producing fuel containers. EPA
has identified 9 domestic manufacturers and 1 foreign manufacturer. Of these 9 U.S.
manufacturers, 8 meet the SB A definition of a small entity. One small business accounted for
over 50 percent of the U.S. sales in 2002, and the other small entities comprised about 10 percent
of U.S. sales.
14.4 Issues Raised by Public Comments
During the public comment period we received numerous comments regarding various
aspects of the proposed rule; however, we did not receive many comments on our proposed small
business provisions. The comments relating to the small business provisions were mainly
focused on those provisions proposed for small refiners, and are summarized below. More
information on these comments can be found in the Final Summary and Analysis of Comments,
which is a part of the rulemaking record.
We received comments from small refiners generally supporting the small refiner provisions.
We also received comments from a few stakeholders regarding the small refiner employee count
and crude capacity criteria. These commenters stated that they believed that EPA's criteria fail
to provide relief to a small number of refiners whom they believe are similar in many respects to
those refiners that will qualify as small under our criteria. The commenters pointed to recent
Congress!onally-enacted programs, specifically the Energy Policy Act of 2005 and the American
Jobs Creation Act of 2004, which use definitions that are different from SBA's definition, and
from the criteria that EPA is adopting in this rule. The Energy Policy Act focuses on refinery
size rather than company size, and the American Jobs Creation Act focuses on refinery-only
employees rather than employees company-wide. EPA has established the criteria for qualifying
for small refiner relief based on the Small Business Administration's (SB A) small business
definition (13 CFR 121.201). Further, we have used these criteria in previous and current fuels
programs and we believe it is prudent to retain the criteria of 1,500 employees and 155,000 bpcd
crude capacity limit for consistency with these programs.
14-7
-------
Final Regulatory Impact Analysis
We do not believe that it would be appropriate to change the small refiner employee count or
crude capacity limit criteria to fit either the Energy Policy Act or the Jobs Creation Act
definitions. Further, SBA established the small business standards to set apart those companies
which were at an inherent economic disadvantage due to their size. We agree with SBA's
assessment that refiners of this size should be afforded special consideration under regulatory
programs that have a significant economic impact on them. We continue to believe that it is
most appropriate to remain consistent with our previous fuels programs and retain the small
refiner criteria that have been used in the past (with some minor clarifications to avoid
confusion).
We also received comments from representatives of small refiners which stated that a
maximum average benzene standard changes the economics of small refiner compliance and that
it should (and must) be considered by an SBAR Panel before a rule is finalized. The commenters
stated that they believe that the imposition of a 1.3 vol% refinery maximum average violates the
Regulatory Flexibility Act because the Panel did not have the opportunity to review the impacts
of such a standard on small businesses. The commenter stated that EPA needed to present the
maximum average provision to the Panel for its consideration prior to including it as part of a
final rule. The commenters added that the possibility of a maximum average was never raised
during the Panel process and that had it been, the small refiner SERs would have opposed the
concept as greatly damaging to their segment of the industry. The commenters expressed
concerns with the 1.3 vol% refinery maximum average, and requested that small refiner
provisions allowing flexibility in meeting this maximum average be included in the final rule.
The commenters also expressed concerns such as maintaining octane levels, costs for
transportation of extracted benzene, and ability to locate other treatment facilities. Lastly, the
commenters stated that they have serious concerns about inability to use credits to meet levels
above 1.3, thus they suggested that EPA should allow small refiners to use credits for
compliance with the 1.3 vol% refinery maximum average, with either a PADD restriction on
credit trading or discounting credits used to meet the 1.3 vol% standard.
We understand the commenters' concerns with regard to the comments on the small refiners'
difficulty in meeting the 1.3 vol% refinery maximum average. As discussed further in section VI
of the preamble to the final rule, as well as chapter 4 of the Summary and Analysis document, we
disagree that adopting a refinery maximum average in the final rule without specifically
presenting the option for consideration by the Panel, or without reconvening that panel, violates
the requirements of the Regulatory Flexibility Act. EPA complied with all requirements under
SBREFA, and we note that the statute in fact contemplates that there will be changes between
proposed and final rules, and states that EPA's only procedural requirement in such a case is to
describe that change in the Final Regulatory Flexibility Analysis. Further, EPA requested
comment on the option of adopting a 1.3 vol% maximum average (71 FR 15869, 15903) and
received comment on the issue (including from small refiners).
We do not agree with the suggestion for PADD-restricted trading. Such geographic
restrictions on credit use can prove to be very problematic, and would necessitate that we set
different standards in different PADDs, due to the different level of benzene reductions
achievable considering cost and other factors in those PADDs. This would reduce the liquidity
14-8
-------
Final Regulatory Impact Analysis
of the credit trading market, and thus drive up the costs of the program. We believe that even
with a maximum average standard, the combination of provisions that we are finalizing will
minimize the likelihood of extreme hardship for small refiners. As discussed below in section
14.6, we are finalizing several significant relief provisions that apply specifically to small
refiners, namely four years of additional lead-time to meet the 1.3 vol% maximum average (until
July 1, 2016). Further, the hardship provisions that we are finalizing are available to all refiners,
and these provisions could apply to situations that the commenters identified may still occur.
14.5 Projected Reporting, Recordkeeping, and Other Compliance
Requirements of the Regulation
For highway light-duty vehicles, EPA is continuing the reporting, recordkeeping, and
compliance requirements prescribed for this category in 40 CFR part 86. These requirements
include certification requirements and provisions related to reporting of production, emissions
information, flexibility use, etc. The types of professional skills required to prepare reports and
keep records are also similar to the types of skills set out in 40 CFR part 86.
For any fuel control program, EPA must have assurance that fuel produced by refiners meets
the applicable standard, and that the fuel continues to meet this standard as it passes downstream
through the distribution system to the ultimate end user. The recordkeeping, reporting and
compliance provisions we are finalizing are fairly consistent with those currently in place for
other fuel programs. For example, reporting will include the submission of pre-compliance
reports, which are already required under the highway and nonroad diesel fuel programs, to give
EPA general information on refiners' plans and projected credit availability. Refiners will be
required to submit refinery batch reports under the MSAT2 program, as they currently are for our
other fuel programs. As with previous fuel regulations, small refiners will be required to apply
for small refiner status and small refiner baselines. Lastly, we are requiring that all records be
kept for at least five years. This recordkeeping requirement should impose little additional
burden, as five years is the applicable statute of limitations for current fuel programs.
For portable fuel containers, requirements are similar to those in the California program,
such as submitting emissions testing information, reporting of certification families, and use of
transition provisions. For more information on the specific compliance provisions that are being
finalized today, please see section VII.D of the preamble to the final rule.
Section XI.B of the preamble to the final rule includes a discussion of the estimated burden
hours and costs of the recordkeeping and reporting that will be required by this final rule.
Detailed information on the reporting and recordkeeping measures associated with this
rulemaking are described in the Information Collection Requests (ICRs), also located in the
preamble to this rulemaking: EPA ICR #0783.50 for light-duty vehicles, EPA ICR #1591.20 for
fuel-related items, and EPA ICR #2213.01 for portable fuel containers.
14.6 Steps to Minimize Significant Economic Impact on Small Entities
14-9
-------
Final Regulatory Impact Analysis
As a part of the SBREFA process, we conducted outreach to a number of small entities
representing the various sectors covered in this rulemaking and convened a Panel to gain
feedback and advice from these representatives. Prior to convening the Panel, we held outreach
meetings with the SERs to learn the needs of small businesses and potential challenges that these
entities may face. The outreach meetings also helped to provide the SERs an opportunity to gain
a better understanding of the upcoming standards. The feedback that we received from SERs as
a result of these meetings was used during the Panel process to develop regulatory alternatives to
mitigate the impacts of the rulemaking on small businesses. General concerns raised by SERs
during the SBREFA process were potential difficulty and costs of compliance with the upcoming
standards.
The Panel consisted of members from EPA, the Office of Management and Budget (OMB),
and the Small Business Administration's Office of Advocacy. Following the Panel convening, a
Final Panel Report detailing all of the alternatives that were recommended by the Final
Regulatory Support Document Panel (as well as individual Panel members) was issued. We
either proposed or requested comment on the various recommendations put forth by the Panel.
Below we discuss those flexibility options recommended in the Panel Report, our proposed
regulatory alternatives, and those provisions which are being finalized. We are finalizing many
of the provisions recommended by the Panel, with exceptions noted below. We believe that the
provisions that we are finalizing will help to mitigate the burden imposed upon small entities in
complying with this rule.
14.6.1 Regulatory Alternatives and Hardship Provisions for Highway Light-Duty Vehicle
Manufacturers
The Panel developed a wide range of regulatory alternatives to mitigate the impacts of the
rulemaking on small businesses, and recommended that we propose and seek comment on the
flexibilities. Described below are the flexibility options recommended by the Panel and our
proposed regulatory alternatives.
14.6.1.1 Panel Recommendations
For certification purposes, SVMs include ICIs and alternative fuel vehicle converters
since they sell less than 15,000 vehicles per year. Similar to the flexibility provisions
implemented in the Tier 2 rule, the Panel recommended that we allow SVMs (includes all
vehicle small entities that would be affected by this rule, which are the majority of SVMs) the
following flexibility options for meeting cold temperature VOC standards and evaporative
emission standards:
For cold VOC standards, the Panel recommended that SVMs simply comply with the
standards with 100 percent of their vehicles during the last year of the four-year phase-in period.
For example, if the standard for light-duty vehicles and light light-duty trucks (0 to 6,000 pounds
GVWR) were to begin in 2010 and end in 2013 (25%, 50%, 75%, 100% phase-in over 4 years),
the SVM provision would be 100 percent in 2013. If the standard for heavy light-duty trucks and
14-10
-------
Final Regulatory Impact Analysis
medium-duty passenger vehicles (greater than 6,000 pounds GVWR) were to start in 2012 (25%,
50%, 75%, 100% phase-in over four years), the SVM provision would be 100 percent in 2015.
In regard to evaporative emission standards, the Panel recommended that since the
evaporative emissions standards will not have phase-in years, we allow SVMs to simply comply
with standards during the third year of the program (we have implemented similar provisions in
past rulemakings). For a 2009 start date for light-duty vehicles and light light-duty trucks, SVMs
would need to meet the evaporative emission standards in 2011. For a 2010 implementation date
for heavy light-duty trucks and medium-duty passenger vehicles, SVMs would need to comply in
2012.
In addition, the Panel recommended that hardship flexibility provisions be extended to
SVMs for the cold temperature VOC and evaporative emission standards. The Panel
recommended that SVMs be allowed to apply (EPA would need to review and approve
application) for up to an additional 2 years to meet the 100 percent phase-in requirements for
cold VOC and the delayed requirement for evaporative emissions. Appeals for such hardship
relief must be made in writing, must be submitted before the earliest date of noncompliance,
must include evidence that the noncompliance will occur despite the manufacturer's best efforts
to comply, and must include evidence that severe economic hardship will be faced by the
company if the relief is not granted.
14.6.1.2 What We Proposed
For cold VOC standards, we proposed the Panel's recommendation that SVMs comply
with the standards with 100 percent of their vehicles during the last year of the four-year phase-
in period, which would be 100 percent in model year 2013. Also, since the proposed standard
for heavy light-duty trucks and medium-duty passenger vehicles would start in 2012 (25%, 50%,
75%, 100% phase-in over four years), we proposed that the SVM provision would be 100
percent in model year 2015.
We agreed with the Panel's recommendation regarding evaporative emission standards,
therefore, for a 2009 model year start date for light-duty vehicles and light light-duty trucks, we
proposed that SVMs meet the evaporative emission standards in model year 2011. For a model
year 2010 implementation date for heavy light-duty trucks and medium-duty passenger vehicles,
we proposed that SVMs comply in model year 2012.
Although the SBAR panel did not specifically recommend it, we also proposed to allow
ICIs to participate in the averaging, banking, and trading program for cold temperature NMHC
fleet average standards (as described in Table VI.B-1 of the preamble), but with appropriate
constraints to ensure that fleet averages will be met. The existing regulations for ICIs
specifically bar ICIs from participating in emission related averaging, banking, and trading
programs unless specific exceptions are provided (see 40 CFR 85.1515(d)). The concern is that
they may not be able to predict their sales and control their fleet average emissions because they
are dependent upon vehicles brought to them by individuals attempting to import uncertified
vehicles. However, an exception for ICIs to participate in an averaging, banking, and trading
14-11
-------
Final Regulatory Impact Analysis
program was made for the Tier 2 NOx fleet average standards, and thus we proposed to apply a
similar exception for the cold temperature NMHC fleet average standards.
If an ICI is able to purchase credits or to certify a test group to a family emission level
(PEL) below the applicable cold temperature NMHC fleet average standard, we would permit the
ICI to bank credits for future use. Where an ICI desires to certify a test group to a FEL above the
applicable fleet average standard, we would permit them to do so if they have adequate and
appropriate credits. Where an ICI desires to certify to an FEL above the fleet average standard
and does not have adequate or appropriate credits to offset the vehicles, we would permit the
manufacturer to obtain a certificate for vehicles using such a FEL, but would condition the
certificate such that the manufacturer can only produce vehicles if it first obtains credits from
other manufacturers or from other vehicles certified to a FEL lower than the fleet average
standard during that model year.
We do not believe that ICIs can predict or estimate their sales of various vehicles well
enough to participate in a program that would allow them leeway to produce some vehicles to a
higher FEL now but sell vehicles with lower FELs later, such that they were able to comply with
the fleet average standard. We also cannot reasonably assume that an ICI that certifies and
produces vehicles one year would certify or even be in business the next. Consequently, we
proposed that ICIs not be allowed to utilize the deficit carry-forward provisions of the proposed
ABT program.
We proposed the Panel recommendation that hardship provisions be extended to SVMs
for the cold temperature NMHC and evaporative emission standards as an aspect of determining
the greatest emission reductions feasible. These entities could, on a case-by-case basis, face
hardship more than major manufacturers (manufacturers with sales of 15,000 vehicles or more
per year). We proposed this provision to provide what could prove to be a needed safety valve
for these entities, and we are proposing that SVMs would be allowed to apply for up to an
additional 2 years to meet the 100 percent phase-in requirements for cold NMHC and the
delayed requirement for evaporative emissions. As with hardship provisions for the Tier 2 rule,
we proposed that appeals for such hardship relief must be made in writing, must be submitted
before the earliest date of noncompliance, must include evidence that the noncompliance will
occur despite the manufacturer's best efforts to comply, and must include evidence that severe
economic hardship will be faced by the company if the relief is not granted.
14.6.1.3 Provisions Being Finalized in this Rule
We are finalizing, as proposed, that the SVM provision will be 100 percent in model years
2013 and 2015. For a 2009 model year start date for LDVs and LLDTs, we are finalizing that
SVMs must meet the evaporative emission standards in model year 2011. For a model year 2010
implementation date for HLDTs and MDPVs, we are finalizing that SVMs must comply in
model year 2012.
We are also finalizing the proposed provision that ICIs may participate in the averaging,
banking, and trading program for cold temperature NMHC fleet average standards, but with
14-12
-------
Final Regulatory Impact Analysis
appropriate constraints to ensure that fleet averages will be met. Further, we are finalizing that
ICIs not be allowed to utilize the deficit carry-forward provisions of the ABT program.
Lastly, we are finalizing the proposed hardship provisions described above. Sections V.E. 1
through V.E.3 of the preamble to the final rule contain more detailed discussions on provisions
for small volume manufacturers.
14.6.2 Regulatory Alternatives and Hardship Provisions for Gasoline Refiners
14.6.2.1 Panel Recommendations
Discussed below are the options that the Panel recommended during the SBREFA
process.
Delay in Standards
The Panel recommended that a four-year delay period should be proposed for small
refiners. Such a delay would be needed in order to allow for a review of the ABT
program, as discussed below, to occur one year after implementation but still three years
prior to the small refiner compliance deadline. It was also noted that a delay option
would also allow for small refiners to be able to expand their production capacity. The
Panel supported allowing for refinery expansion and recommended that refinery
expansion be provided for in the rule.
Early ABT Credits
The Panel recommended that early credit generation be afforded to small refiners that
take some steps to meet the benzene requirement prior to the effective date of the
standard. Depending on the start date of the program, and coupled with the four-year
delay option, a small refiner could have a total credit generation period of five to seven
years. The Panel also stated that it supports allowing refiners (small, as well as non-
small, refiners) to generate credits for reductions to their benzene emissions levels (unlike
prior fuels programs which have given early credits only to refiners who have fully met
the applicable standard early).
Extended Credit Life
The Panel recommended that EPA propose a program that does not place a limit on credit
life. During Panel discussions, it was noted that some Panel members were not in
support of limited credit life for the general program. When the Final Panel Report was
written, EPA intended to proceed with a proposal that did not place a limit on credit life;
therefore the Panel did not make a specific recommendation on the concept of extended
credit life. However, based on discussions during the Panel process, the Panel would
have recommended that extended credit life be offered to small refiners if the general
ABT program were to include a limit on credit life.
Program Review
14-13
-------
Final Regulatory Impact Analysis
The Panel recommended a review of the credit trading program and small refiner
flexibility options one year after the general program starts. Such a review could take
into account the number of early credits generated, as well as the number of credits
generated and sold during the first year of the program. Further, requiring the submission
of pre-compliance reports from all refiners would likely aid EPA in assessing the ABT
program prior to performing the review. The Panel noted that, combined with the
recommended four-year delay, a review after the first year of the program would still
provide small refiners with the three years that it was suggested would be needed for
these refiners to obtain financing and perform engineering and construction for benzene
reduction equipment. Should the review conclude that changes to either the program or
the small refiner provisions are necessary, the Panel recommended that EPA also
consider some of the suggestions provided by the small refiners (their comments are
located in Appendix E of the Final Panel Report), such as:
• the general MSAT program should require pre-compliance reporting (similar to
EPA's highway and nonroad diesel rules);
• following the review, EPA should revisit the small refiner provisions if it is found
that the credit trading market does not exist, or if credits are only available at a cost
that would not allow small refiners to purchase credits for compliance; and,
• the review should offer ways either to help the credit market, or help small
refiners gain access to credits (e.g., EPA could 'create' credits to introduce to the
market, EPA could impose additional requirements to encourage trading with small
refiners, etc.).
In addition, the Panel recommended that EPA consider in this rulemaking establishing an
additional hardship provision to assist those small refiners that cannot comply with the
MSAT with a viable credit market. (This suggested hardship provision was also
suggested by the small refiners in their comments, located in Appendix E of the Final
Panel Report). This hardship provision could address concerns that, for some small
refineries, compliance may be technically feasible only through the purchase of credits
and it may not be economically feasible to purchase those credits. This flexibility could
be provided to a small refiner on a case-by-case basis following the review and based on
a summary, by the refiner, of technical or financial infeasibility (or some other type of
similar situation that would render its compliance with the standard difficult). This
hardship provision might include further delays and/or a slightly relaxed standard on an
individual refinery basis for a duration of two years; in addition, this provision might
allow the refinery to request, and EPA grant, multiple extensions of the flexibility until
the refinery's material situation changes. The Panel also stated that it understood that
EPA may need to modify or rescind this provision, should it be implemented, based on
the results of the program review.
During the Panel process, we stated that we intended to propose the extreme unforeseen
circumstances hardship and extreme hardship provisions (for all gasoline refiners and importers),
similar to those in prior EPA fuels programs. A hardship based on extreme unforeseen
circumstances would provide short term relief due to unanticipated circumstances beyond the
control of the refiner, such as a natural disaster or a refinery fire. An extreme hardship would
14-14
-------
Final Regulatory Impact Analysis
provide short-term relief based on extreme circumstances (e.g., extreme financial problems,
extreme operational or technical problems, etc.) that impose extreme hardship and thus
significantly affect a refiner's ability to comply with the program requirements by the applicable
dates. The Panel agreed with the proposal of such provisions and recommended that we include
them in the MSAT rulemaking.
14.6.2.2 What We Proposed
In general, we proposed the Panel's recommended regulatory flexibility provisions. The
following is a discussion of the proposed provisions, as well as an additional provision that we
proposed based on additional analysis following the SBREFA Panel process.
Delay in Standards
We proposed the Panel's recommendation that small refiners be allowed to postpone
compliance with the proposed benzene standard until January 1, 2015, which is four
years after the general program begins. While all refiners are allowed some lead time
before the general proposed program begins, we believe that in general small refiners
would still face disproportionate challenges. Previous EPA fuel programs have included
two to four year delays in the start date of the effective standards for small refiners,
consistent with the lead time we believe appropriate here. The proposed four-year delay
for small refiners would help mitigate these challenges. Further, a four-year delay would
be needed in order to allow for a review of the ABT program, as discussed below, to
occur one year after the general MSAT program implementation but still roughly three
years prior to the small refiner compliance deadline.
Early ABT Credit Generation Opportunities
We are proposing the Panel's recommendation that early credit generation be afforded to
small refiners that take steps to meet the benzene requirement prior to their effective date.
While we have anticipated that many small refiners would likely find it more economical
to purchase credits for compliance, some have indicated they will make reductions to
their gasoline benzene levels to meet the proposed benzene standard. Further, a few
small refiners indicated that they would likely do so earlier than would be required by the
January 1, 2015 proposed small refiner start date. Small refiner credit generation would
be governed by the same rules as the general program, described in the preamble to the
proposed rule in Section VILE. The only difference is that small refiners would have an
extended early credit generation period of up to seven years. Early credits could be
generated by small refiners making qualifying reductions from June 1, 2007 through
December 31, 2014, after which program credits could be generated indefinitely for those
that over-comply with the standard.
Extended Credit Life
As discussed in the preamble, we proposed a limit on credit life. However, in order to
encourage the trading of credits to small refiners and increase the certainty that credits
would be available (as it would provide a viable outlet for credits facing expiration), we
proposed that the useful life of credits be extended by 2 years if they are generated or
14-15
-------
Final Regulatory Impact Analysis
used by small refiners. This is meant to directly address concerns expressed by small
refiners during the Panel process that they would be unable to rely on the credit market to
avoid large capital costs for benzene control. While this flexibility option was not
specifically recommended by the Panel, we believe that the Panel would be in support of
such an option.
ABTProgram Review
We proposed the Panel's recommendation that a review of the ABT program be
performed within the first year of the general MSAT program (i.e., by 2012). To aid the
review, we also proposed the requirement that all refiners submit refinery pre-compliance
reports annually beginning June 1, 2008. In order for EPA to carry out this review, we
believe that refiners' 2011 annual compliance report would also need to contain
additional information, including credits generated, credits used, credits banked, credit
balance, cost of credits purchased, and projected credit generation and use through 2015.
When combined with the four-year delay option, this would afford small refiners with the
knowledge of the credit trading market's status before they would need to invest capital.
As suggested by the Panel, we requested comment on elements to be included in the ABT
program review, and suggested actions that could be taken following such a review.
Such elements could include:
• Revisiting the small refiner provisions if it is found that the credit trading market
does not exist to a sufficient degree to allow them to purchase credits, or that credits
are only available at a cost-prohibitive price.
Options to either help the credit market, or help small refiners gain access to credits.
In addition, we proposed the Panel's recommendation of the inclusion of an additional
hardship provision that could be applied for following, and based on the results of, the
ABT program review.
We did in fact propose the two hardship provisions stated above that the Panel recommended
(the extreme unforeseen circumstances hardship and extreme hardship provisions). In addition,
we proposed that these hardship provisions would be available to all refiners, regardless of size.
These provisions would, at our discretion, permit a refiner to seek a temporary waiver from the
MSAT benzene standard under certain rare circumstances.
14.6.2.3 Provisions Being Finalized in This Rule
We are finalizing a four-year period of additional lead time for small refiners to comply
with the 0.62 vol% benzene requirement, until January 1, 2015. Consistent with the general
program allowance of an additional 18 months (beyond the 0.62 vol% benzene standard
compliance date) for compliance with the 1.3 vol% refinery maximum average, we are also
finalizing 18 months of additional lead-time for small refiners to comply with the 1.3 vol%
maximum average, until July 1, 2016 (and thus, small refiners will also receive an additional four
years of lead-time from the general program start date for the 1.3 vol% refinery maximum
14-16
-------
Final Regulatory Impact Analysis
average). We believe that this lead-time will provide these refiners with sufficient time to
complete any necessary capital projects.
We are also finalizing the early credit generation provision for small refiners. This is similar
to the general early credit generation provision that is provided to all refiners, except that small
refiners may generate early credits until January 1, 2015. As discussed further in section
VI.A.2.b.ii of the preamble to the final rule, refineries must reduce their 2004-2005 benzene
levels by at least ten percent to generate early credits. This ten percent threshold is being set to
ensure that changes in gasoline benzene levels are representative of real refinery process
improvements, not just normal fluctuations in benzene level at a given refinery (allowed under
MSAT1). The small refiner early credit generation period will be from June 1, 2007 to
December 31, 2014, after which credits may be generated indefinitely for those that
overcomplied with the standard. We are finalizing a modified version of the proposed extended
credit life provision. The two-year credit life extension will pertain to standard credits only
(since refiners already have an incentive to trade early credits to small refiners), and the
extension will only apply to those standard credits traded to small refiners. There is no need to
extend credit life for credits generated by small refiners, because in this event, the small refiner
would already have the utmost certainly that the credits would be available for use.
We are also finalizing as proposed the ABT program review after the first year of the overall
program. In part to support this review, we are requiring that refiners submit pre-compliance
reports, similar to those required under the highway and nonroad diesel programs. If, following
the review, EPA finds that the credit market is not adequate to support the small refiner
provisions, we will revisit the ABT provisions to determine whether or not they should be altered
or whether EPA can assist the credit market (and small refiners' access to credits) to enable a
successful ABT program. We are finalizing an additional hardship provision to assist small
refiners if it is found that some small refiners still cannot comply with the benzene standard even
with a viable credit market. This hardship provision would be for the case of a small refiner for
which compliance with the 0.62 vol% benzene standard would be feasible only through the
purchase of credits, but it was not economically feasible for the refiner to do so. This hardship
provision will only be afforded to a small refiner on a case-by-case basis, and will only be
available following the ABT program review. The hardship application must be based on a
summary by the refiner of the practical or financial difficulty with compliance with the 0.62
vol% benzene standard (or some other type of similar situation that would render its compliance
with the standard) difficult. The relief offered under this hardship provision is a further delay, on
an individual refinery basis, for up to two years. Following the two years, a small refiner will be
allowed to request one or more extensions of the hardship until the refinery's material situation
has changed.
We are finalizing the extreme hardship provision and the extreme unforeseen
circumstances hardship provision with some modifications, as this final rule includes a 1.3 vol%
refinery maximum average benzene standard. As discussed in more detail in section VI.A.3.b of
the preamble to the final rule, relief will be granted on a case-by-case basis, however it may
differ somewhat depending upon whether a refiner applies for hardship relief for the 0.62 vol%
benzene standard or for the 1.3 vol% refinery maximum average standard. This is partly due to
14-17
-------
Final Regulatory Impact Analysis
the fact that a refiner may use credits to meet the 0.62 vol% benzene standard, but credits cannot
be used for compliance with the 1.3 vol% refinery maximum average.
Extreme hardship circumstances could exist based on severe economic or physical lead time
limitations of the refinery to comply with the required benzene standards at the start of the
program. For relief from the 0.62 vol% benzene standard in extreme hardship circumstances,
relief will likely be in the form of an extension of the one-year deficit carry-forward allowed by
the rule. Hardship relief from the 1.3 vol% refinery maximum average benzene standard in
extreme hardship circumstances would consist of additional time to comply with the 1.3 vol%
refinery maximum average. Refiners must apply by January 1, 2008 (or, January 1, 2013 for
approved small refiners) for extreme hardship relief from the 1.3 vol% refinery maximum
average, as this provision is intended to address unusual circumstances that should be apparent
now or well before the effective date of the standard.
The extreme unforeseen circumstances hardship is available to both refiners and importers,
and is intended to provide relief in extreme and unusual circumstances outside the refiner or
importer's control that could not have been avoided through the exercise of due diligence.
Hardship relief for the 0.62 vol% benzene standard will allow a deficit to be carried forward for
an extended, but limited, time period (more than the one year allowed by the rule). Hardship
relief from the 1.3 vol% refinery maximum average benzene standard based on unforeseen
circumstances will be granted on a case-by-case basis, following an assessment of the hardship
application.
14.6.3 Portable Fuel Container Manufacturers
14.6.3.1 Panel Recommendations
Since nearly all portable fuel container manufacturers are small entities and they account
for about 60 percent of sales, the Panel suggested that the flexibility options be offered to all
portable fuel container manufacturers. The flexibilities that the Panel recommended are detailed
below.
Design Certification
The Panel recommended that we propose to permit portable fuel container manufacturers
to use design certification in lieu of running any or all of the durability aging cycles.
Manufacturers could demonstrate the durability of their portable fuel containers based in
part on emissions test data from designs using the same permeation barriers and
materials. Under a design-based certification program a manufacturer would provide
evidence in the application for certification that their container would meet the applicable
standards based on its design (e.g., use of a particular permeation barrier). The
manufacturer would submit adequate engineering and other information about its
individual design such that EPA could determine that the emissions performance of their
individual design would not be negatively impacted by slosh, UV exposure, and/or
14-18
-------
Final Regulatory Impact Analysis
pressure cycling (whichever tests the manufacturer is proposing to not run prior to
emissions testing).
Broaden Certification Families
This approach would relax the criteria used to determine what constitutes a certification
family. It would allow small businesses to limit their certification families (and therefore
their certification testing burden), rather than testing all of the various size containers in a
manufacturer's product line. Some small entities may be able to put all of their various
size containers into a single certification family. Manufacturers would then certify their
containers using the "worst case" configuration within the certification family. To be
grouped together, containers would need to be manufactured using the same materials
and processes even though they are of different sizes. The Panel recommended that EPA
propose this approach.
Additional Lead-time
It was recognized that time would be needed for the portable fuel container SERs to
gather information to fully evaluate whether or not additional lead-time might be needed
beyond the proposed 2009 start date, the Panel recommended that we discuss lead-time in
the proposal and request comment on the need for additional lead-time to allow
manufacturers to ramp up to a nationwide program.
Product Sell-through
As with past rulemakings for other source sectors, the Panel recommended that EPA
propose to allow normal sell through of portable fuel containers as long as manufacturers
do not create stockpiles of noncomplying portable fuel containers prior to the start of the
program.
Following the SBREFA process, the Panel recommended that we propose two types of
hardship programs for small portable fuel container manufacturers. These suggested provisions
were:
• Allow small manufacturers to petition EPA for limited additional lead-time to
comply with the standards. A manufacturer would have to make the case that it has
taken all possible business, technical, and economic steps to comply but the burden of
compliance costs or would have a significant adverse effect on the company's
solvency. Hardship relief could include requirements for interim emission reductions.
The length of the hardship relief would be established during the initial review and
would likely need to be reviewed annually thereafter.
• Permit small manufacturers to apply for hardship relief if circumstances outside
their control cause the failure to comply (i.e., supply contract broken by parts
supplier) and if failure to sell the subject containers would have a major impact on the
company's solvency. The terms and timeframe of the relief would depend on the
14-19
-------
Final Regulatory Impact Analysis
specific circumstances of the company and the situation involved. As part of its
application, a company would be required to provide a compliance plan detailing
when and how it would achieve compliance with the standards under both types of
hardship relief.
14.6.3.2 What We Proposed
Based upon the comments received from portable fuel container small entity
representatives during the SBREFA Panel process, we decided to propose the Panel-
recommended flexibility and hardship provisions for portable fuel container manufacturers. As
stated previously, nearly all portable fuel container manufacturers (8 of 10 manufacturers as
defined by SB A) are small entities and they account for about 60 percent of sales, the Panel
recommended to extend the flexibility options and hardship provisions to all portable fuel
container manufacturers, thus we proposed that these flexibilities be offered to all portable fuel
container manufacturers. Moreover, implementation of the program would be much simpler by
doing so.
Further, we proposed that the two types of hardship provisions recommended by the
Panel be extended to portable fuel container manufacturers.
14.6.3.3 Provisions Being Finalized in This Rule
We are finalizing, as proposed, the flexibility provisions described above for portable fuel
container manufacturers. We are also finalizing the hardship provisions described above for
these entities. These entities could, on a case-by-case basis, face hardship, and we are finalizing
these provisions to provide what could prove to be needed safety valves for these entities. For
both types of hardship provisions, the length of the hardship relief will be established, during the
initial review, for not more than one year and will be reviewed annually thereafter as needed.
Section VII.F of the preamble to the final rule contains a more detailed discussion of these
hardship provisions.
14.7 Related Federal Rules
The primary federal rules that are related to this rule are the first mobile source air toxics rule
(66 FR 17230, March 29, 2001), the Tier 2 Vehicle/Gasoline Sulfur rulemaking (65 FR 6698,
February 10, 2000), the fuel sulfur rules for highway diesel (66 FR 5002, January 18, 2001) and
nonroad diesel (69 FR 38958, June 29, 2004), the Reformulated Gasoline and Anti-dumping rule
(59 FR 7813 and 59 FR 7860, February 16, 1994), and the Cold Temperature Carbon Monoxide
Rulemaking (57 FR 31888, July 17, 1992).
In addition, the Evaporative Emissions Streamlining Direct Final Rulemaking was issued on
December 8, 2005 (70 FR 72917). For portable fuel containers, the Occupational Safety and
Health Organization (OSHA) has safety regulations for gasoline containers used in workplace
settings. Containers meeting OSHA requirements, commonly called safety cans, are exempt
14-20
-------
Final Regulatory Impact Analysis
from the California program, and EPA is planning to exempt them from the EPA program.
Section 1501 of the Energy Policy Act of 2005 (EPAct) requires that EPA implement a
Renewable Fuels Standard (RFS) program. Beginning in 2006, this program will require
increasing volumes of renewable fuel to be used in gasoline, until a total of 7.5 billion gallons is
required in 2012. The most prevalent renewable fuel to be used in gasoline is expected to be
ethanol.
There are a wide variety of potential impacts of ethanol blending on MS AT emissions that
will be evaluated as part of the RFS rulemaking process. In general, as ethanol use increases,
other sources of octane in gasoline can decrease. Depending on these changes, the impact on
benzene emissions will vary. The specific effects of ethanol on benzene are addressed in this
Regulatory Impact Analysis, and will also be addressed and in future rulemakings such as the
RFS rule.
14.8 Conclusions
Throughout the entire rulemaking process, we conducted substantial outreach— including
convening a Panel during the SBREFA process as well as meetings with other stakeholders— to
gather information about the effect of this final rule on small entities. We used this information,
and performed cost-to-sales ratio tests (a ratio of the estimated annualized compliance costs to
the value of sales per company) to determine the impacts of the rule on small entities.
In regard to the highway light-duty manufacturers, we found that small vehicle entities
(which include manufacturers, ICIs and converters) in general would likely be impacted
similarly as large entities. As we discussed earlier in Chapter 5 (Vehicle Feasibility) and Chapter
8 (Vehicle Costs), we are aligning the EPA evaporative emission standards with California LEV
II standards, and essentially all manufacturers certify 50-state evaporative systems that meet both
sets of standards. We do not expect additional costs from this requirement since we expect that
manufacturers will continue to produce 50-state evaporative systems. In limited cases where
vehicle small entities may not currently produce 50-state systems, the flexibilities and hardship
relief for small entities, as described earlier, will reduce the burden on these entities.
In addition, as described earlier in Chapters 5 and 8, the cold temperature exhaust (VOC)
emission standards for light-duty vehicles can be achieved through calibration alone. It will only
require up-front research and development costs, and certification burden is likely to be small
due to existing cold carbon monoxide testing requirements. Therefore, the new cold temperature
VOC standard is expected to add less than $1 on average to the cost of vehicles. In general,
small vehicle entities will likely experience similar impacts as large entities. Also, as described
earlier, the flexibility and hardship provisions will reduce the burden of the new cold VOC
standard on small vehicle entities.
With respect to small refiners, these entities in general would likely experience a significant
and disproportionate financial hardship in complying with the requirements in this rule. Refinery
14-21
-------
Final Regulatory Impact Analysis
modeling (of all refineries), indicates higher refining costs for small refiners. Chapter 9 of this
RIA contains a detailed discussion of our analysis and projected costs for U.S. refiners in
complying with the benzene control program.
Of the small refiners with publicly available sales data, we were able to estimate annual
costs, and use this information to complete a cost-to-sales ratio test. Our current estimate for the
14 small refiners (owning 16 refineries) that we believe will be subject to this rulemaking is as
follows: 37.5 percent (6 refineries) would be affected at less than 1 percent of their sales (i.e., the
estimated costs of compliance with the proposed rule would be less than 1 percent, of their
sales), 37.5 percent (6 refineries) would be affected at greater than 1 percent but less than 3
percent, and 25 percent (4 refineries) would be affected at greater than 3 percent of their sales.
Therefore, we believe that the flexibility provisions are necessary to help mitigate these impacts
to small refiners. Our cost analysis, however, does not consider benzene control options which
could dramatically reduce compliance costs for these small refineries, particularly those
refineries affected by the 1.3 vol% maximum average standard. The costs for these small
refineries are high because of their poorer economies of scale for installed capital. We believe
that these refiners can avoid high per-gallon costs by installing a reformate splitter. The
reformate splitter is a relatively low capital and operating cost unit that would allow them to
remove a benzene-rich stream from the rest of their reformate, resulting in a final gasoline that
would be in compliance with the maximum average standard. The benzene-rich stream can be
sold to another refinery with gasoline benzene levels below the cap standard and so can absorb
this small benzene-rich volume. This sort of trading is similar to the credit trading program,
except that actual benzene is being traded instead of paper credits.
For portable fuel containers, as discussed earlier, nearly all manufacturers are small
entities, thus the flexibility and hardship provisions afforded in this rule will be offered to all
portable fuel container manufacturers. Moreover, small portable fuel container manufacturers
will likely be impacted by the new standards similarly as the large manufacturers.
Automatically-closing spouts and permeation control are expected to be utilized to meet the
evaporative emissions standard for portable fuel containers. As discussed in Chapters 10
(Portable Fuel Container Costs) and Chapter 13 (Economic Impact Analysis), all portable fuel
containers range in price from $3 to $7, and the added variable and fixed costs for the new
portable fuel containers with auto-close spouts and permeation control is estimated to be about
$2.70 per unit on average. We continue to believe that manufacturers will be able to pass on
these costs without a significant impact on portable fuel container sales. In addition, the
flexibilities and hardship relief for all portable fuel container manufacturers would reduce the
burden of the new standards on small and large manufacturers.
14-22
------- |