Draft Regulatory Impact Analysis
Control of Gasoline Volatility
and
Evaporative Hydrocarbon Emissions
from New Motor Vehicles
July 1987
U.S. Environmental Protection Agency
Office of Air Radiation
Office of Mobile Sources
-------
TABLE OF CONTENTS
Page
Chapter 1: Introduction 1-1
I. Background 1-1
II. Control Options 1-3
A. Gasoline RVP Control Options 1-4
B. Alcohol Blend RVP Control Options 1-5
C. Mid-Range Volatility Control Options . . .1-6
III. Structure of this Report 1-6
Chapter 2: In-Use Motor Vehicle-Related Hydrocarbon
Emissions 2-1
I. Ozone Noncompliance Situation 2-2
A. Non-Attainment Areas 2-2
B. Seasonal Nature of Ozone Non-Compliance. .2-5
II. Effect of Volatility on HC Emission Factors. . .2-10
A. Gasoline-Fueled Vehicles 2-10
B. Alcohol Blend-Fueled Vehicles 2-127
C. Gasoline Distribution System 2-154
III. Benzene Emission from Gasoline-Fueled Vehicles
as a Function of Fuel Parameters 2-156
A. Literature Review 2-157
B. Exhaust Emission Analysis 2-161
C. Evaporative Emission Analysis 2-164
D. Refueling Emission Analysis 2-169
E. Effect of RVP Control on Fuel Composition.2-169
F. Results 2-170
Appendix 2-A: Breakdown of Motor Vehicle Evaporative
Emissions Into Their Components
Appendix 2-B: Detailed Benzene Emission Data
Chapter 3: Environmental Impacts 3-1
I. Description of Model and Inputs 3-1
A. MOBILE3: Calculation of Emission Factors.3-2
B. Calculation of Inventories 3-17
C. EKMA: Ozone Air Quality Analysis 3-38
II. Nationwide and Non-Attainment Area Inventory
Projections 3-43
A. Nationwide Inventory Projections 3-45
B. Non-Attainment Area Inventory Project ions.3-52
III. Ozone Air Quality Analysis 3-54
IV. Benzene Emissions and Incidence Analysis . . . .3-63
A. Benzene Emissions 3-63
B. Incidence Analysis 3-67
Appendix 3-A: Butane Reactivity
Chapter 4: Vehicle Controls 4-1
I. Technology 4-1
A. Canister Modifications 4-1
-i-
-------
TABLE OF CONTENTS (cont'd)
B. Purge Requirements 4-12
II. Costs 4-22
A. Vehicle Hardware 4-22
B. Effects of Increased Canister Size on
Operating Costs 4-30
C. Body Modification Costs 4-34
III. Conclusions 4-38
Appendix 4-A: Detailed Derivation of Evaporative ECS
Component Costs
Chapter 5: Technological Feasibility and Cost of In-Use
Volatility Control 5-1
I. Refinery Control of Gasoline Volatility 5-2
II. The Bonner and Moore Studies 5-3
A. Bonner and Moore's Refinery and
Photochemical Modeling System 5-3
III. Refinery Costs of Volatility Control 5-11
A. RVP Control 5-11
B. Control of % i 6o 5-18
IV. Effect of RVP Control on the Butane Market . . .5-23
V. Increased Energy Density and Evaporative
Emissions Recovery 5-24
A. Volatility and Heat of Combustion 5-25
B. Heat of Combustion and Fuel Economy. . . .5-28
C. Overall Relationship Between Gasoline
Volatility and Fuel Economy 5-58
D. Economic Credit from Evaporative HC
Recovery/Prevention 5-59
VI. Effect of RVP Control on Driveability 5-61
A. Studies Performed on the Subject of
Driveability 5-65
B. Industry Data on Current Volatility-
Related Problems 5-72
C. Costs Associated with High Volatility
Driveability Problems 5-76
D. Conclusion 5-84
VII. Enforcement Costs 5-85
Chapter 6: Analysis of Alternatives: Gasoline 6-1
I. RVP Control 6-3
A. Methodology 6-3
B. Cost Effectiveness Comparison of RVP
Control Scenerios 6-13
II. %iso Control 6-31
Appendix 6-A: Selection of a Certification Fuel RVP
Appendix 6-B: Evaluation of an Inspection/Maintenance
Program for Evaporative Emission
Control Systems
-l i-
-------
TABLE OF CONTENTS (cont'd)
Chapter 7: Analysis of Alternatives: Alcohol Blends . . .7-1
I. Industry Economics and Blend RVP 7-1
A. Gasohol 7-1
B. Methanol Blends 7-2
II. Emission Effects of Blend Usage 7-3
III. Effect of RVP Control on Alcohol Blends 7-4
A. Effect on Emissions 7-4
B. Effect on Refinery and Distribution
Costs 7-4
IV. Gasohol Control Options 7-6
V. Methanol-Blend Control Options 7-9
Appendix 7-A: Effect of Gasohol RVP Allowance on
Chicago VOC Inventory
-iii-
-------
CHAPTER 1
Introduction
I. Background
High concentrations of ozone have become an increasing
problem for urban areas across the United States. Over one
hundred million Americans now live in areas which do not meet
the National Ambient Air Quality Standard for ozone -- even as
the Clean Air Act (CAA) deadline for attainment by all areas
approaches (December 31, 1987). Increasing concerns about the
harmful human health effects of exposure to ozone has
heightened the seriousness of this ozone non-attainment problem.
Hydrocarbon (HC) and other volatile organic carbon (VOC)
emissions are the primary precursors in the creation of ozone
in the atmosphere. Evaporative HC emissions from
gasoline-fueled vehicles continue to contribute substantially
to total HC emissions. Cost effective recovery and control of
these emissions can result in significant reductions in urban
ozone levels.
Evaporative HC emissions from motor vehicles originate
from two basic components of the vehicle's fuel system — the
fuel tank and the carburetor. Evaporative emissions from the
fuel tank -- known as "diurnal" losses -- occur as the gasoline
vapors expand in response to daily ambient temperature
increases. The other type of vehicle evaporative emissions --
referred to as "hot soak" losses — occur just after the engine
is turned off, when residual engine heat causes the evaporation
of some of the fuel remaining in the carburetor bowl and fuel
lines. In many newer vehicles, some hot-soak losses also
originate from the fuel tank as well, probably due to
recirculation of gasoline that has been heated by the engine.
Currently, all new gasoline-fueled vehicles and trucks are
equipped with evaporative control systems designed to capture
the majority of these diurnal and hot-soak losses. A typical
system consists of a canister filled with charcoal granules
which adsorb the HC vapors generated in the fuel tank and the
carburetor. Later, while the engine is operating, the
evaporative canister is periodically purged with air and the
collected HCs are stripped from the canister and burned in the
engine.
Beginning in 1971, light-duty gasoline vehicles and
light-duty gasoline trucks with gross vehicle weights (GVWs)
less than 6000 lbs. were subject to an evaporative hydrocarbon
-------
1-2
standard. The original standard of 6 grams/test in 1971 and 2
grams/test in 1972 was based on a rather lenient, canister-only
test*. Beginning in 1978, the standard was adjusted to 6.0
grams/test, to be measured according to the Sealed Housing
Erussion Determination, or SHED, test**. Also in 1978,
light-duty gasoline trucks over 6000 lbs. GVW became subject to
the same standard. Later, in the 1981 model year, this
standard was lowered to the current 2.0 grams/test value for
both passenger cars and light trucks. Most recently, in 1985,
EPA began to regulate evaporative emissions from heavy-duty
gasoline vehicles. Two standards apply, depending on truck
weight: 3.0 grams/test ( 14 ,000 lbs. GVW or less) and 4.0
grams/test (greater than 14,000 lbs. GVW).
Evaporative control systems are designed to meet these HC
standards when the vehicle is fueled with certification test
gasoline, which has a typical Reid Vapor Pressure (or RVP, a
common measure of volatility) of 9.0 psi. Although this level
of volatility was representative of commercial fuels in the
early 1970's when certification test fuel specifications were
developed, the RVP of commercial gasoline has risen steadily
since then as refiners have increased the butane content of
gasoline. Adding butane has been a response to rising energy
costs (butane is less expensive than gasoline) and has been
made possible by newer vehicles' ability to handle higher RVP
f uel.
Another factor increasing m-use fuel RVP has been the
introduction of blends of alcohols -- primarily ethanol but
also methanol — with gasoline. Adding alcohols increases the
RVP of the "base gasoline" significantly. Ethanol-gasol1ne
blends ("gasohol") have been free from volatility constraints,
while the RVP of methanol-plus-cosolvent gasoline blends (or
methanol blends) has been restricted. Methanol blend RVP
cannot exceed the ASTM limits of straight gasoline ("ASTM
limits" are defined in Section II.A. below). Since gasoline
RVP is essentially at the ASTM limits today on average,
methanol must be blended with special lower-RVP gasoline if
blends are to meet RVP limits. With no current volatility
limits, the gasohol industry is able to simply "splash-blend"
ethanol with commercially available gasoline, resulting in a
1.0 psi RVP increase.
Under normal summertime conditions, higher RVP results in
nigher evaporative emissions. EPA's testing of actual in-use
vehicles on typical ASTM Class C fuel (i.e., 11.5 psi RVP)
That is, hydrocarbons emitted from the evaporative
emissions control charcoal canister, rather than those
from the whole vehicle, were measured.
The SHED test uses an enclosure to measure evaporative
emissions from the entire vehicle.
-------
1-3
indicates that evaporated vapors frequently exceed the capacity
of most vehicle evaporative emission control systems. As a
result, evaporative emissions from vehicles operating on
commercial fuels are well above the certification standards.
EPA's testing has also revealed that the majority of
problem-free in-use carbureted vehicles are unable to meet the
evaporative standards even while operating on certification
test fuel (9.0 psi RVP), though the degree of exceedance is
small compared to that on 11.5 psi RVP fuel. This suggests
that insufficient design related to aspects of the current test
procedure — in addition to higher RVP fuels — also contribute
to the higher evaporative emissions seen in in-use vehicles.
Problem-free fuel-injected vehicles (a minority of today's
fleet, but expected to dominate late 1980's sales) perform well
on certification test fuel, but greatly exceed 2 grams/test on
commercial fuel.
Based on these findings, EPA has concluded that the
majority of vehicles being driven in the field today are
exceeding the current evaporative HC standards and will
continue to do so into the future, even as more fuel-injected
vehicles progressively replace carbureted vehicles.
Because the evaporative HC emission problem is widespread
and serious, a clear potential exists for significant ozone
reductions through successful evaporative HC control efforts.
EPA has been exploring a broad range of control scenarios
designed to accomplish various levels of emissions reductions.
This report provides detailed descriptions and analyses of
these scenarios. This report updates and supersedes a November
1985 report entitled "Study of Gasoline Volatility and
Hydrocarbon Emissions From Motor Vehicles," (EPA-AA-SDSB-85-5).
11. Control Options
There are three independent sets of control options
available to EPA for reducing evaporative HC emissions and
hence, urban ozone levels. All available options are attempts
to better match the volatility of actual in-use fuels with the
volatility of certification test fuel. We will touch on each
set of options briefly here and more throughly in later
chapters of this Regulatory Impact Analysis (henceforth
"R.I.A.")
The three sets of possible control options are: 1) those
which relate to gasoline volatility control, 2) those which
relate to alcohol blend volatility control, and 3) those which
relate to "mid-range" volatility control of gasoline and/or
alcohol blends.
-------
1-4
A. Gasoline RVP Control Options
There are two fundamental approaches to gasoline RVP
control; as a practical matter, all control options are
actually combinations of the two. The first approach would
focus on improving the design of vehicle evaporative systems to
better capture the excess vapors from current and future
commercial fuels. EPA would revise the evaporative emissions
certification test procedure to require the use of a test fuel
representative of in-use fuels and a more sophisticated test
sequence. Passing the new certification test for evaporative
HC would then require the design of an improved control system
which would in actual use capture the excess evaporative HCs.
The second approach would focus on reducing the volatility
of the commercial gasoline actually used in vehicles. Refiners
would change the composition of their gasolines to lower their
RVP values, primarily by adding less butane. The American
Society of Testing and Materials (ASTM) has developed a
classification system which assigns a non-binding volatility
limit to gasoline supplied to each state for each month of the
year. These limits are linked to climatic variations and
although they exist primarily to reduce vapor lock and starting
problems, they also appear to be a satisfactory basis from
which to apply in-use volatility controls.
When refiners supply gasoline which meets the five ASTM
classifications (ranging from 9.0 psi RVP in Class A to 15.0
psi for Class E) the effect is to roughly equalize the actual
volatilities of in-use gasolines throughout the country,
regardless of climate, altitude, and time of year. The EPA
test procedure most closely simulates the temperature ranges in
Class C areas, and our analysis uses Class C areas as a point
of reference. If EPA established an RVP reduction (for
example, a 2.5 psi RVP reduction from the 11.5 psi RVP Class C
limit) we would also make proportional reductions in the
volatility of gasolines supplied to all other areas in order to
achieve approximately the same degree of emission control
natlonwide.
All gasoline RVP options we consider in this study are
combinations of vehicle- and fuel-based controls which match
certification fuel RVP to Class C in-use fuel RVP. The
evaluated options range from matching certification and in-use
fuels at 11.5 psi RVP (which would primarily consist of vehicle
control with a very small amount of fuel control) to matching
the fuels at 8.0 psi (which would be a primarily fuel control
program with very slight vehicle improvements). Under all of
these options new vehicles would be designed to meet the
evaporative HC standard under in-use conditions. This study
evaluates the emission reductions, costs and cost effectiveness
-------
1-5
associated with matching certification and m-use fuels at each
half-psi RVP increment from 11.5 psi to 8.0 psi.
In-use fuel RVP control produces emission reductions
sooner than does vehicle control. That is because fuel turns
over in the production/distribution system very fast (about one
month), while a fleet of new vehicles with improved controls
reaches significant numbers only after a number of years. This
fact presents the option of additional short-term reduction of
RVP below the revised certification level in the early years of
the program. Such supplementary control would achieve
additional evaporative emission reductions from existing older
vehicles and the gasoline distribution system. For each of the
long-term fuel control scenarios considered in this report, we
also evaluate a range of possible short-term supplemental fuel
controls.
B. Alcohol Blend RVP Control Options
Another set of alternatives that this report evaluates
relate to the special case of controlling the volatility of
alcohol blends. Since adding ethanol or methanol to gasoline
increases the RVP, in-use fuel controls would create unique
problems for alcohol blenders. The effects would be most
pronounced for the gasohol industry because their product has
been free from volatility controls throughout the industry's
development. This situation has made it possible for gasohol
blenders to use commercially-available finished gasolines as a
base stock for blending with ethanol. This "splash-blending"
is a less expensive process than using special low-RVP base
gasoline in order to reduce the RVP of the final blend.
Methanol blends, on the other hand, have been and are currently
subject to volatility controls; as a result, producers have
never had the flexibility to use commercial gasolines as base
stock.
Three regulatory options for gasohol are evaluated in this
report: 1) entirely exempting gasohol from RVP control, 2)
providing a 1.0 psi RVP "allowance" above the gasoline
standards and 3) requiring gasohol to meet the same RVP
standard as gasoline. Also evaluated are variations of Option
3 which would require blends to meet the gasoline RVP standards
only in non-attainment areas or only after a delay until 1993
(the year the special federal tax status of the ethanol
industry is set to expire). Even though methanol blend RVP has
never been allowed to exceed that of gasoline, those options
appearing to be most viable for gasohol are also evaluated with
respect to methanol blends.
-------
1-6
C. Mid-Range Volatility Control Options
In addition to options relating to the control of RVP,
this report addresses the question of whether mid-range
volatility of in-use fuel should be controlled and what the
effects of such controls would be. The issue arises because at
the elevated temperatures associated with "hot-soak"
conditions, additional heavier components of gasoline begin to
evaporate as well as the butane. This report evaluates the
options of fuel-based and vehicle-based control of mid-range
volatility.
III. Structure of This Report
The primary analysis contained in this report focuses on a
comparison of cost effectiveness values for a number of long-
and short-term evaporative control strategies, as described
above. (Cost effectiveness analysis relates control costs to
amount of emission control achieved in terms of dollars per ton
of pollutant reduced.) The bulk of the report develops a basis
for these results.
Chapter 2 presents the detailed technical evaluations of
how volatility affects the HC emissions. The chapter begins
with a discussion of the U.S. ozone non-compliance situation
and then describes how higher volatility fuel increases
evaporative, exhaust and refueling emissions. The improved
vehicle evaporative emissions model presented incorporates the
effects of fuel "weathering", tank fill level, temperature, and
RVP variability. Also a part of this chapter are analyses of
how volatility affects emissions from alcohol-blend fueled
vehicles and from the gasoline distribution system. Finally,
Chapter 2 includes a discussion of the value of controlling
mid-range volatility, as mentioned above.
Next, Chapter 3 translates these emissions/RVP
relationships into environmental impacts across the country for
each control scenario. Nationwide VOC emissions reductions,
effect on ozone non-attainment, and effect on benzene-related
cancer incidences are evaluated in this chapter. Hydrocarbon
emissions reductions developed here become one of two primary
components of the final cost-effectiveness analyses.
Costs are the other component in evaluating cost-
effectiveness, and Chapters 4 and 5 develop cost data for the
two basic control approaches described earlier. Chapter 4
presents an in-depth analysis of the cost of improving
evaporative controls on vehicles themselves. Chapter 5
completes the foundation of the report by developing the range
of costs and savings which would result from reducing in-use
gasoline RVP to various levels.
-------
1-7
The final step in our analysis is to compare the various
control scenarios among themselves and Chapters 6 and 7 fulfill
this role. Chapter 6, the more quantitative of the two,
combines the actual calculated emissions and cost results from
the earlier chapters to arrive at numerical cost effectiveness
values for each of the alcohol-free gasoline scenarios (both
RVP and mid-range volatility). Then in Chapter 7 the report
concludes by discussing the alternatives for alcohol-blend
fuels described above. This discussion is necessarily less
quantitative than that in Chapter 6 because of the issues
dependence on the detailed workings of the gasoline
distribution system and the commercial interactions between the
alcohol industry, gasoline refiners and marketers, and the
various governmental entities providing tax credits.
Nonetheless, the comparison of alternatives is as detailed and
thorough as currently available information allows.
A complete discussion of how the Agency used this study in
selecting specific options for proposal is found in the
Preamble to the associated Notice of Proposed Rulemaking
itself. At the time this study was done, 1988 appeared to be a
feasible model year for implementing fuel RVP controls.
However, the actual timeline of the rulemaking process has made
it necessary to instead propose fuel controls for 1989. The
analyses here are equally applicable to either year of
implementation.
-------
Chapter 2
In-Use Motor Vehicle Related Hydrocarbon Emissions
The primary goal of this chapter is to determine the
relationship between fuel volatility and hydrocarbon (HC)
emissions, which will be used later for air quality modeling in
Chapter 3. Section I is more general in nature, describing the
current ozone noncompliance situation. It's emphasis is on
establishing the need for HC control and on identifying the
seasonal and geographical extent of this need. The remaining
sections of the chapter are an analysis of the effect of fuel
volatility and m-use conditions on both hydrocarbon and
benzene emissions, both from vehicles and from gasoline
distribution systems. Consideration was given to the effect of
fuel volatility and m-use conditions on hydrocarbon and
benzene emissions from each possible emission source
independently.
Section II presents the effect of volatility on
hydrocarbon emissions from gasoline-fueled and alcohol
blend-fueled vehicles and also the emissions from the gasoline
distribution system. The vehicle emissions discussed are
evaporative, exhaust, and refueling emissions, with the primary
emphasis on evaporative emissions. The discussion on
evaporative emissions presents the data upon which the
evaporative emissions model was based, develops the evaporative
emissions model, explains the process by which the model is
adjusted for in-use conditions, and also analyzes the effects
of fuel distillation properties on evaporative emissions.
Vehicle emissions of benzene, a known carcinogen, were
also investigated. Work done to quantify evaporative and
exhaust benzene emissions as they relate to fuel volatility is
presented in Section III. Vehicle refueling is another source
of benzene emissions but is not dealt with here in depth since
this subject has been treated elsewhere sufficiently.[1,2]
Chapter 2 concludes with two appendices. Appendix 2-A
presents a more detailed description of how the Emission Factor
vehicle evaporative emissions data were broken down into their
components, and how vehicle evaporative emissions were
predicted based on the emission components. Appendix 2-B
presents the data which the benzene analysis of Section III is
based on.
-------
2-2
I. Ozone Noncompliance Situation
A. Non-Attainment Areas
The National Ambient Air Quality Standard (NAAQS) for
ozone is a concentration of 0.12 parts per million (ppm), and
is attained when the expected number of calendar days per year
with maximum hourly concentrations exceeding the standard is
less than or equal to one. EPA determines whether an area is
in compliance with the ozone NAAQS on the basis of the area's
ozone design value, defined as the fourth highest daily maximum
one-hour measured ozone concentration over a three-year
period.* Based on 1982-84 design values, the most recent
available at the time of this analysis, EPA's Office of Air
Quality Planning and Standards (OAQPS) currently designates 73
urban areas** (12 of them m California) as non-attainment
areas for ozone. This compares to the 54 such areas (seven in
California) that were designated as non-attainment areas based
on 1981-83 design values, as described in EPA's original
volatility study.[3]
The analyses presented here focus on the 61 non-California
urban non-attainment areas. For reasons that are described in
Chapter 3, the 12 urban non-attainment areas within California
were not modeled individually, although the nationwide
emissions inventories presented in Chapter 3 include the state
of California in the totals. These 61 areas and their
associated ozone design values are presented in Table 2-1.[4]
For additional information on the development of the ozone
design values presented in Table 2-1, the reader is referred to
the report, "Standard Metropolitan Statistical Areas Regulatory
Analysis Air Quality Data Base, 1982-845 ]
The serious nature of the ozone noncompliance situation
becomes clearer when the design values in Table 2-1 are
examined. Many of the 61 urban areas listed have design values
of 0.15 ppm or greater. Recent air quality modeling, the
results of which are presented in Chapter 3, indicates that it
is very unlikely that any of the areas having design values of
0.15 ppm or greater can achieve attainment with the ozone NAAQS
by the statutory deadline of December 31, 1987. In fact, in
the absence of any additional programs to reduce emissions of
ozone precursors, projections indicate that 40 of the 61 areas
will continue to be in non-attainment status at the deadline.
Or the third highest from two years of data, or the second
highest from a single year's data.
Metropolitan Statistical Areas (MSA), or combinations of
contiguous MSAs.
-------
2-3
Table 2-1
61 Non-California Urban Ozone Non-Attainment
Areas and Associated Design Values (ppm) [2]
Design
Area Value
EPA Region 1
Boston Metro Area 0.19
Greater Connecticut Metro Area 0.23
New Bedford, MA 0.19
Portland, ME 0.15
Portsmouth-Dover-Rochester, NH-ME 0.13
Providence, RI 0.16
Springfield, MA 0.19
Worcester, MA 0.14
EPA Region 2
Atlantic City, NJ 0.19
New York Metro Area 0.23
Vineland-Millville-Bridgeton, NJ 0.14
EPA Region 3
Allentown-Bethlehem, PA 0.15
Baltimore, MD 0.17
Erie, PA 0.13
Harrisburg-Lebanon-Carlisle, PA 0.13
Lancaster, PA 0.14
Philadelphia Metro Area 0.18
Pittsburgh, PA 0.14
Reading, PA 0.13
Richmond-Petersburg, VA 0.14
Scranton-Wilkes Barre, PA 0.13
Washington, DC-MD-VA 0.16
York, PA 0.13
EPA Region 4
Atlanta, GA 0.17
Birmingham, AL 0.15
Charlotte-Gastonia-Rock Hill, NC-SC 0.13
Chattanooga, TN-GA 0.13
Huntington-Ashland, WV-KY-OH 0.14
Louisville, KY-IN 0.15
Memphis, TN-AR-MS 0.13
Miami-Hialeah, FL 0.13
Nashville, TN 0.13
Tampa-St. Petersburg-Clearwater, FL 0.13
-------
2-4
Table 2-1 (cont'd)
Design
Area Value
EPA Region 5
Akron, OH 0.13
Canton, OH 0.13
Chicago Metro Area 0.20
Cincinnati Metro Area 0.15
Cleveland, OH 0.14
Dayton-Springfield, OH 0.13
Detroit, MI 0.14
Grand Rapids, Ml 0.13
Indianapolis, IN 0.13
Milwaukee Metro Area 0.17
Muskegon, MI 0.14
EPA Region 6
Baton Rouge, LA 0.17
Beaumont-Port Arthur, TX 0.21
Brazoria, TX 0.14
Dallas-Fort Worth, TX 0.16
El Paso, TX 0.17
Galveston-Texas City, TX 0.17
Houston, TX 0.25
Lake Charles, LA 0.15
Longview-Marshal1, TX 0.15
New Orleans, LA 0.15
San Antonio, TX 0.14
Tulsa, OK 0.13
EPA Region 7
Kansas City, MO-KS 0.14
St. Louis, MO-IL 0.17
EPA Region 8
Denver-Boulder, CO 0.14
Salt Lake City-Ogden, UT 0.15
EPA Region 9
Phoenix, AZ 0.15
-------
2-5
The severity of the ozone non-attainment problem can also
be seen when the populations of the non-attainment areas are
considered. Based on 1980 census data, more than 92 million
people (more than 40 percent of the nation's population) reside
in the 61 urban ozone non-attainment areas outside California.
When all 73 areas (including California) are considered, the
total population living in non-attainment areas is about 115
million, or approximately half the nation.
This information briefly but clearly illustrates the
widespread nature of the ozone noncompliance situation. In the
following section, the seasonal nature of violations of the
ambient ozone standard is discussed.
B. Seasonal Nature of Ozone Noncompliance
Another important perspective on the ozone noncompliance
situation is the seasonal nature of violations of the ozone
NAAQS. Examination of data recorded at ozone monitoring sites
in non-attainment areas has revealed seasonal trends in ozone
violations. As might be expected, the majority of all ozone
violations occur during the summer months, when ambient
conditions (i.e., temperatures and length of daylight) are most
favorable to ozone formation. These seasonal trends are
important in determining during what period (specific months)
hydrocarbon emission reductions would be most valuable. This
is an important consideration with respect to any in-use
volatility control measures, as they have the flexibility to be
implemented throughout the year or only during specific
months. However, other hydrocarbon emission control measures,
such as revisions to certification fuel specifications and test
procedure, would affect vehicle design and thus represent
year-round control.
As is also discussed in Chapter 3, the air quality
analyses included in this study focus on the "design value day"
(i.e., the date on which the ozone design values shown in Table
2-1 were recorded). Ozone air quality modeling is performed
over a one-day period. (Modeling over two or three days is
possible, although these models require inputs too detailed for
application to the 61 areas.) The rationale for focusing on
the design value day is that it represents the worst-case
situation m terms of ozone compliance. If the projected
ambient ozone concentration for an area, based on the
conditions that existed on the design value day and additional
emission controls, can be brought below the level of the NAAQS,
it is likely that the area will comply with the standard. Of
course, as with any nationwide analysis, there are many
parameters that cannot be estimated on a city-specific basis.
This could introduce significant error in projecting the
compliance status of any individual area. However, the total
-------
2-6
number of non-attainment areas can be estimated with much more
confidnce.
Table 2-2 illustrates the temporal distribution of the
design value days for the 61 non-California urban non-
attainment areas. Fifty-one of the 61 days, or 84 percent,
occurred in the three-month summer period of June through
August, and all but 6 of the design values (90 percent of the
total) were recorded in the June-through-September period. It
is worth noting that the four design values recorded in April
all occurred on April 27th or later. In fact, three of the
four April design value days occurred on April 28 or 29, 1983.
These four values are all associated with abnormally warm
spring days (maximum temperatures of 80° F or higher), and all
are for areas in the northeastern part of the country. (A
complete listing of the design value dates and the temperatures
recorded on those dates is presented in Chapter 3.)
A second, broader analysis is available which examines the
seasonal occurrence of all ozone violations for 47 non-
California urban areas. [3] Only 47 areas (rather than 61) were
examined because at the time of the analysis, areas were
designated non-attainment on the basis of 1981-83 (rather than
1982-84) design values. The data for the analysis were taken
from the SAROAD (Storage and Retrieval of Atmospheric Data)
system, and included ozone data recorded at all monitoring
sites within each of the urban areas. Reports of daily maximum
one-hour ozone concentrations at each monitoring site, recorded
for each day of the year, were obtained for 1980 and 1983:
1980 representing a recent year with a relatively high number
of ozone violations, and 1983 representing the most recent
complete data set at the time of that analysis. These years
are still appropriate here, 1980 for the reason cited above and
1983 because it is the base year on which the environmental
impact analysis (see Chapter 3) is based. Also, all of the
current 61 non-attainment areas with design values well above
the ozone NAAQS are included in the 47-area set. Those areas
present only in the 47-area set or the 6l-area set are areas
with ambient ozone concentrations very close to the NAAQS.
Thus, the conclusions drawn regarding the seasonality of the
ozone noncompliance problem from this analysis remain valid.
Table 2-3 presents the results of the seasonal analysis
for both 1980 and 1983. The data summarized in the upper
portion of the table describe the distribution of total annual
violations by month and by various combinations of summer
months. "Total annual violations" are defined as the total
number of days in which a one-hour average ozone concentration
exceeding 0.125 ppm at any given site was recorded; since
violations in a specific urban area are summed over all
monitoring sites within that area, the total monthly violations
for any specific non-attainment area can be more than 31.
-------
2-7
Table 2-2
Distribution of Ozone Desiqn Value
Days
Number of Ozone
Percent
Month
Desiqn Value Days
of Total
January
0
0
February
0
0
March
0
0
April *
4
7
May
0
0
June
16
26
July
15
25
August
20
33
September
4
7
October
1
2
November
0
0
December
1
2
Total 61
100
The four design values recorded in April all occurred in
the last 4 days of the month.
-------
2-8
Table 2-3
Seasonal Trends in Ozone Violations
Percent of Total Annual Violations
All 47
Areas
All 47
Areas
Except Houston*
Months
1980
1983
1980
1983
April
1
1
1
1
May
5
5
3
1
June
13
22
14
23
July
39
26
43
28
August
27
31
28
34
September
10
13
9
13
October
2
1
1
0
July & August
66
57
71
62
June - August
79
79
85
85
June - September
89
92
94
98
May - September
94
97
97
99
Average
of Area-Specific Monthly
Peaks (ppm)
All 03 Levels
Months
1980
1983
January
.055
.050
February
.075
. 073
March
. 082
. 080
Apr i 1
. 101
. 099
May
. 123
. 103
June
. 151
. 144
July
. 162
. 149
August
.151
. 156
September
. 134
. 133
October
. 101
. 102
November
.084
. 061
December
. 067
. 053
Houston is excluded because of its anomalously large
number of non-summer violations.
-------
2-9
As indicated in the top half of Table 2-3, ozone
violations are heavily concentrated in the warmer months, when
temperatures and sunlight duration are most conducive to ozone
formation. On the basis of the 1983 data, 80 percent of the
non-attainment areas (38 of 47) experienced all ozone
violations during a five-month period (i.e., May through
September inclusive). Further, nearly all of the remaining
areas recorded at least 80 percent of their violations during
these 5 months. The major exceptions to the latter statement
experienced very few ozone violations in 1983 (Scranton, PA had
three violations, one in April; Miami had only one violation,
also in April).
In both 1980 and 1983, Houston experienced a relatively
large number of ozone violations, recording 193 and 217 annual
exceedances (respectively) over its 13 monitoring sites. In
1980, 25 percent of these violations occurred outside of the
May-September period; the 1983 figure is slightly lower at 18
percent non-summer violations. The fact that Houston recorded
ozone concentrations at or above the NAAQS in every month of
both years, while no other area recorded such levels in more
than 8 different months in either year, further illustrates the
unique nature of the ozone problem in that area.
In evaluating various seasonal options for in-use fuel
volatility control, the most important factor is consideration
of the period of time during which ozone reductions are needed
most. The percentage distribution shown in the upper part of
Table 2-3 is based on total violations at all monitors; thus,
areas with more monitoring sites contribute more heavily to the
distribution. Since Houston recorded a relatively high number
of non-summer violations and also has more monitoring sites
than most of the areas, the results are presented both with and
without the Houston data.
As shown in Table 2-3, the month during which the highest
percentage of all 1980 violations occurred was July, with 39
percent of the total; in 1983, August was the highest with 31
percent of total annual violations. In addition to individual
months, the two-, three-, four-, and five-month periods
recording the highest percentages of total violations are
presented. The top portion of the table shows that in-use fuel
volatility control between May and September (inclusive) would
potentially impact 94 to 97 percent of ozone violations. If
Houston is excluded, this five-month period encompasses
virtually all ozone episodes. Four-month control
(June-September) would have an impact on just slightly less of
the ozone season — 80 to 92 percent of total violations in all
areas and 94 to 98 percent if the Houston data are excluded.
-------
2-10
The lower section of Table 2-3 summarizes seasonal trends
in peak ozone concentrations, showing the average of all areas'
maximum ozone levels by month, As shown, average peak ozone
concentrations show a definite trend, from highs of 0.156 to
0.162 ppm in July or August to lows of 0.050 to 0.055 ppm in
January. Further, the average of the peak ozone concentrations
exceeds the level of the NAAQS in May through September in
1980, and in June through September in 1983. Taken together
with the information shown in the upper portion of the table,
the seasonal nature of ozone NAAQS violations is clear.
The discussion above shows that, for those ozone control
strategies that can be seasonally focused (such as volatility
control), there are two primary choices for the period of
control: June through September and May through September.
The occurrence of four of the current 61 design value days in
late April, and the greater potential to impact ozone
violations, leads to the selection of May through September
(inclusive) as the seasonal control period.
II. Effect of Volatility on HC Emission Factors
A. Gasoline Fueled Vehicle
1. Evaporative Emissions
Evaporative HC emissions from motor vehicles can be
separated into the two basic categories of "diurnal" and
"hot-soak" losses which result from different processes.
Diurnal emissions consist of HCs both evaporated and displaced
from the vehicle's fuel tank as the vehicle tracks the diurnal
swing in ambient temperatures. Each day, as the fuel in the
tank and the vapor above the fuel heat up, more of the liquid
fuel evaporates and the vapor itself expands, with both
phenomena causing HCs to be released into the atmosphere
(unless captured by a control system). Fuel volatility, size
of vapor space, initial ambient temperature, magnitude of the
diurnal temperature swing, and driving patterns all impact the
level of evaporative emissions from a vehicle's fuel tank.
Hot-soak emissions occur during the period immediately
following engine shut-down (i.e., at the end of each vehicle
trip), These losses will occur both as distillation from the
fuel metering system (either a carburetor or a fuel injector)
and as evaporation from the fuel tank. Evaporative emissions
from the fuel metering system occur as part of a different
process than that described for diurnal losses. When the
engine is shut off, so is the cooling system. The engine block
and surrounding area heat the engine coolant (which is no
longer circulating) and other engine components, usually kept
cool by the circulating coolant, before natural cooling begins
-------
2-11
to take effect. Any fuel remaining in the carburetor bowl, or
leaking from a malfunctioning fuel injector, will undergo a
distillation process during this time and vapors will be
released to the atmosphere unless captured by a control
device. Since the majority of hot-soak emissions arise from
distillation, the volatility characteristics of the fuel and
engine temperature have the greatest impact on the level of
hot-soak emissions.
The fuel metering system was previously considered the
only source of hot-soak emissions, but there have been recent
indications that the fuel tank in fuel-injected vehicles can
undergo a temperature change during vehicle operation as a
result of recirculation of fuel heated by the engine. Vapor
production in the tank would be the same type as that described
above for diurnal losses. Because these tank losses are not in
response to ambient temperature changes, they are classified as
hot-soak emissions.
a. Factors Affecting Evaporative Emissions
i. Evaporative Control System
Present evaporative emission control systems are composed
of: 1) an activated carbon canister which adsorbs hydrocarbon
vapor emitted from the vehicle fuel tank and carburetor bowl,
and 2) the associated plumbing and hardware which control the
loading and purging of the canister. Additionally, some
carbureted systems use an air cleaner with an integrated
charcoal element to further adsorb carburetor bowl and intake
manifold vapors. When the engine is running, the stored vapor
is desorbed, or purged, by drawing air through the canister to
the engine intake system. The purge rate is controlled by the
source of the vacuum (air cleaner, carburetor (above throttle
blade) or intake manifold), the pressure drop through the
canister and the size of the controlling orifice (located in
either the canister or purge port). A diagram of a typical
evaporative control system is presented in Figure 2-1.
The actual mass of gasoline vapors that an evaporative
control system will continually adsorb and desorb during
operation is referred to as the "system working capacity".
This working capacity will be dependent upon many factors, some
of which are internal to the system and others that are
external. Among the internal factors are the volume of
charcoal in the canister, the physical characteristics of the
charcoal, the canister configuration, and the volume of purge
air drawn through by the control system. Factors that also
play a role in determining system working capacity, but which
are currently subject to little or no control, include the
temperature and humidity of purge air, and the vapor
-------
Figure 2-1
Typical Evaporative Control System Canister
Cart). Bowl
Carb.
Bawl Valves
0.047" Bleed
To fuel Tank
Purge
Valve. Drilled
To 0.180"
7
i i
Carton -
Foe*
Foan
ire Mesh
Fiberglass Air FUter
valve
STEM
TO
PORTED
VACUUM sffUNGj
SIGNAL
TO
MANIFOLD
VACUUM
SIGNAL
VAPOR TO
INDUCTION
SYSTEM
DIAPHRAGM,
BOOY
COVER
DIAPHRAGM
I
FILTER
UNDER .
VACUUM
-------
2-13
concentration of the evaporative emissions. The temperature of
the purge air could be controlled but it is not very
practical. The vapor concentration effect was noted in an EPA
report comparing the ability of two different charcoals to
adsorb and desorb HC vapors using fuels of varying volatility,
and will be discussed further in Chapter 4.[6]
Manufacturers can design evaporative emission control
systems to have a specific system working capacity by adjusting
those factors that are subject to their control, with
consideration given to the variations in the external factors
mentioned previously. Current certification tests are
performed using Indolene, which has an average Reid Vapor
Pressure (RVP) of 9.0 psi. Therefore, this is the volatility
level at which current evaporative control systems are designed
to meet the current evaporative HC standard of 2 grams/test.
Actual m-use fuel, however, has an RVP closer to 11.5 psi
during the summer months, which results in higher levels of
evaporative emissions than those encountered by the canister
during the certification test. Thus, there exists a
significant difference in the levels of emissions current
evaporative control systems are designed to capture and
emissions encountered in actual operation.
ii. Fuel Volatility
Proper Measure of Fuel Volatility
A number of different gasoline properties may be used to
indicate gasoline volatility. Each year, in a document
referred to as D-439, the American Society for Testing and
Materials (ASTM) publishes recommended limits for several
gasoline volatility parameters, including Reid Vapor Pressure
(RVP), distillation characteristics (such as the temperature at
which a given percentage of gasoline is evaporated), and the
temperature corresponding to a given gasoline vapor-to-1iguid
ratio (V/L). For every month of the year, each state is
assigned a "volatility class" that represents ASTM's
recommended limits on the volatility of gasoline sold in that
state. The five gasoline volatility classes are designated as
A, B, C, D, and E with a corresponding volatility limits as
shown in Table 2-4. As indicated. Class A is the least
volatile and Class E is the most volatile. It should be noted
that these ASTM specifications are merely recommended levels
and are not legally binding for gasoline refiners. A number of
states have adopted ASTM standards as legal limits. However,
the effectiveness of these limits depends directly on
enforcement and it is questionable how strictly these limits
are enforced in many states (California being an exception).
-------
2-14
Table 2-4
ASTM D-439 Gasoline Volatility Specifications
Distillation Temp. (°F) at
Given Percent Evaporated
ASTM
Volatility
Class
Max. RVP
(psi)
Max.
@ 10%
(T10)
Min.
e 50%
(Tso)
Max.
@ 50%
(Tso)
Min. Temp.
<3 V/L = 20
-------
2-15
The major reason behind ASTM's assignment of volatility
limits is the prevention of vapor lock at high ambient
temperatures and problems of starting the engine under colder
conditions. As shown in Table 2-4, values for RVP,
vapor-liquid ratio, and distillation temperatures are defined
for each of the five fuel volatility classes (A through E). In
turn, minimum and maximum ambient temperatures under which no
problems with cold startability or vapor lock would occur were
calculated for each volatility class. Then, temperature data
for each geographical area are used to select the appropriate
volatility classes for each month of the year. A U.S. map
indicating ASTM's state volatility class recommendations for
the month of July is provided in Figure 2-2.
In order to determine the most significant gasoline
volatility parameters with respect to the magnitude of
evaporative HC losses, it is important to consider the source
of the evaporative emissions. The measure of fuel volatility
relevant to diurnal emissions should ideally reflect volatility
at temperatures typically associated with a vehicle's fuel
tank. The most widely accepted measure of volatility in
relation to diurnal emissions is RVP, a measure of the fuel's
vapor pressure at 100°F.[7,8]
Hot-soak emissions originate both from the fuel tank and
the fuel metering system, as indicated previously. The process
occurring in the fuel tank is much the same as with diurnal
emissions. Therefore, RVP is an appropriate measure of
volatility for this portion of the hot-soak emissions. On the
other hand, since the temperatures experienced in the fuel
metering system can be much higher than 100°F, RVP may not be
ideal as an overall indicator of hot-soak emission levels. The
relationship between temperature, volatility and evaporative
losses is not as clear in a carburetor or fuel injector as in
the fuel tank due to the complex interactions that occur. [9]
Only limited work in this area has been done with fuel-injected
vehicles, but carbureted vehicles have been studied fairly
extensively.
Hot-soak losses from a carbureted vehicle are generally
felt to be related also to the mid-range volatility of the
fuel, given as the percent of fuel volume distilled in an ASTM
D216 distillation at the peak temperature in the carburetor
bowl. [7,9-12] This peak temperature will vary from vehicle to
vehicle and may also be affected somewhat by the ambient
temperature. An average value for the peak bowl temperature is
around 160°F and, therefore, the percent of fuel evaporated at
160°F (%16 o) / in addition to RVP, may be a relevant fuel
parameter for estimating hot-soak losses.[7,10 ] EPA has
performed emission tests on vehicles with fuels of equal RVP
and varying %16o values. The results of this program are
presented later in this section.
-------
Figure 2-2
ASTM
,'s July Volatility Classes
INOWTM OMtOTA
i 0 \ /
f"~~ ¦ • ¦¦ * — • m .4 r 1
[SOUTH DAKOTA ? )
i V
r
/"c*£r
par
B
• «
\
B
B
,
B
f C,-«T
•V i
yJ Sggal
C/B
>
i A i
— J
B/A i B I B ^ fciiCc^ „
i b i v r^5^
m5J^ T±™3«SSa V-J—JI—2,&Siwcttift
l'rEXA3 j iAIWAHSAS ^ C^.r.X-f^CAWOOH^-.
A j j B j B ,
• \oui»55^ C/B | c/B ^ C/B /
A = 9.0 psi
B = 10.0 psx
C = 11.5 psi
V/ \
V
-------
2-17
Theoretically, because total evaporative HC emissions are
represented by the sum of hot-soak and diurnal losses, the
ideal measure of fuel volatility to be used in evaluating
various evaporative emission control strategies would most
likely be some combination of both RVP and %i6o- The
weighting given each of these factors would be expected to vary
with vehicle type, as fuel-injected vehicles have significantly
lower hot-soak emissions and, thus, their relative losses would
be more dependent upon RVP. Also, operating temperatures can
vary from model to model, so 160°F may not be the appropriate
point on the distillation curve for all vehicles.
Some attempts have been made to incorporate both RVP and
various distillation points into a volatility index. In an
effort to correlate evaporative emissions to the volatility of
gasolines and methanol/gasoline blends, DuPont developed the
Evaporative Index (EI), as shown below:
EI = 0.8 5(RVP) + 0.14(%2Oo) - 0.32(%,0o ) . [13]
In their application for a waiver of methanol blends, DuPont
showed a correlation of EI versus evaporative emissions with an
Rz value of 0.86.[13] However, some criticisms have been
raised with respect to the DuPont analysis, such as their
combining the results of two independent testing programs
without normalizing the results and the lack of higher
volatility gasolines and blends in the analysis.[14,15]
The Front End Volatility Index (FEVI), which was developed
for purposes other than emissions estimation (primarily the
control of vapor lock), is defined as:
FEVI = RVP + 0.13(%,Sb).[16]
This index essentially includes the two terms most relevant to
evaporative emissions (%iss is very close to %iso)-
However, it is currently unknown if the relative weighting of
the two parameters is appropriate.
Another gasoline property used to measure volatility is
the temperature corresponding to a specified gasoline
vapor-to-liquid ratio (V/L) at atmospheric pressure. This V/L
is the volume of vapor formed at atmospheric pressure and test
temperature divided by the initial volume of liquid gasoline
tested. The temperature at which this vapor-to-liquid ratio is
equal to 20 at atmospheric pressure is designated T20v^l-
The Tjov/l parameter is included in the ASTM volatility
specifications (shown in Table 2-4) and, according to API, is
commonly used for blending purposes by refiners. Since
limiting T20v/l can affect the other evaporative-related fuel
parameters, it deserves further discussion here.
-------
2-18
ASTM D-439 provides an empirical equation defining
T20W/l as a function of the following parameters: RVP, the
temperature at which ten percent of the gasoline is evaporated
(T10), and the temperature at which 50 percent of the
gasoline is evaporated (Tso). This equation is:
T20V/l = 114.6 - 4.1(RVP) + 0.2(T,o) + 0.17(TS0>.[17]
In the above equation for T20v/l, RVP contributes
significantly more to T20v/l than does T,0 or Ts<>.
According to survey data, fuels with RVPs ranging from 11.5 to
9.5 psi have T10s ranging from 108-120°F and TS0s ranging
from 210-220°F.[18] This 2-psi RVP range accounts for roughly
an 8°F change in TlovyLr if Ti0 and Tso are held
constant. If the corresponding Ti0s and TS0s for the
different RVPs are used in the TIOv/l equation, then the
TZ0V/i changes by nearly 12°F. This indicates that RVP is
the major factor affecting T20v/l, but T,0 and Tso are
not negligible.
From the above discussion, it appears that RVP and %i6o
are the most relevant of the available fuel parameters to
indicate evaporative emission potential. As previously
mentioned, RVP, a measure of volatility at 100°F, is relevant
because the maximum fuel tank temperature during a diurnal is
less than 100°F. The %l6o is relevant for hot-soak emissions
since the maximum temperature in the carburetor during a
hot-soak is around 160°F. The other parameters, EI, FEVI and
Tjov/l all essentially combine RVP with higher-temperature
volatility indicators, but it is not clear that any of the
three particular combinations adequately represent the overall
evaporative emission potential for motor vehicles. It appears
safer to address the RVP and %,So parameters separately at
this point. However, only limited data is available on the
effect of %ieo on evaporative emissions. This analysis will
focus primarily on RVP as the most relevant measure of fuel
volatility with an independent analysis of the %i6o effect on
hot-soak emissions.
Historical and Future Trends in Gasoline Volatility
Over the past decade, the volatility of commercial
gasoline has gradually, but steadily, been increasing. Trends
in commercial gasoline volatility were traced using results of
two separate fuel surveys prepared each summer by the National
Institute for Petroleum and Energy Research (NIPER) and the
Motor Vehicle Manufacturers Association (MVMA). Summer
gasoline volatility is the focus because, as concluded in
Section I of this chapter, in-use fuel control from May through
September could have an impact on the majority of ozone
violations. Table 2-5 presents RVP trends for the 15
-------
Table 2-5
NIPER Survey Results[19]: Summer Gasoline RVP Trends by Region*
Region
Years
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
over
Northeast
9.8
10.0
10.2
10.7
10.5
10.5
10.8
10.8
10.9
10.5
10.7
11.2
14
Mid-Atlantic Coast
9.3
10.1
10.1
10.4
10.1
10.2
10.3
10.6
10.8
10.7
10.8
11.1
19
Southeast
9.6
9.6
9.7
9.5
9.4
9.6
9.8
9.9
10.1
10.3
10.2
10.6
10
Appalachian
10.5
10.6
10.5
10.4
10.1
10.6
10.5
10.9
11.1
11.4
11.1
11.5
10
Michigan
10.5
10.4
10.6
11.0
11.2
10.9
11.3
10.9
11.2
11.6
11.5
11.8
12
Northern Illinois
9.5
10.5
10.3
11.0
10.8
10.9
10.9
11.1
10.8
11.7
11.3
11.2
18
Central Mississippi
10.1
10.1
9.9
9.9
10.1
10.4
10.3
9.7
10.5
11.0
10.9
11.1
10
Lower Mississippi
9.4
9.6
9.6
9.5
9.8
9.5
9.7
9.3
10.1
10.1
10.0
10.5
12
Northern Plains
—
9.6
9.9
—
—
9.2
9.8
—
11.0
11.0
10.5
11.3
18
Jentral Plains
—
9.1
9.2
9.0
8.8
9.2
9.2
—
10.2
10.0
10.0
10.3
13
Southern Plains
9.2
9.3
9.2
9.3
9.1
9.5
9.2
9.7
10.0
9.8
9.8
10.1
10
Southern Texas
9.1
9.5
9.4
9.6
9.5
9.4
9.2
9.4
10.3
10.2
10.3
10.1
11
Southern Mountain
8.4
8.9
8.7
8.8
8.9
8.7
8.9
8.4
8.8
9.1
8.9
9.4
12
Northern Mountain
8.9
10.1
9.5
9.9
9.9
9.6
9.5
9.2
10.4
10.4
9.7
10.4
17
Pacific Northwest
9.5
9.9
10.6
10.4
10.0
10.3
10.8
11.0
10.8
11.2
10.8
11.4
20
National Average** 9.5
(excluding California)
9.8
9.8 10.0
9.9
9.2
10.0
10.1 10.5 10.6 10.4
10.8
14
Unleaded regular gasoline only (R + M/2 less than 90).
Calculated as a straight arithmetic average of the 15 regional averages listed.
-------
2-20
non-California regions included in the NIPER survey. As shown,
overall non-California averages indicate a 14-percent increase
in summer unleaded fuel RVP levels over the past 11 years.
Increases within individual regions vary between 10 and 20
percent, with the greatest summer increase occurring in the
Pacific Northwest.
To examine volatility trends over the past twenty years,
it is necessary to look at leaded fuel which, of course, has
been in use longer than unleaded fuel. National-average
results of NIPER leaded gasoline surveys are presented
graphically in Figure 2-3. [19] As shown in the top graph, the
most significant summer RVP "boosts" occurred first in the
1972-74 period, and then again in 1981. Prior to 1972, summer
fuel RVP levels were well within the range specified in the
Code of Federal Regulations (CFR) for certification test fuels
(8.7 to 9.2 psi); however, by the time the SHED* test began
with the 1978 model year, the certification fuel RVP had been
exceeded by the national average RVP of in-use fuel
(approximately 9.6 psi). Since then, the CFR specifications
have become even less representative of commercial fuel
volatility, based on a 1985 summer leaded fuel average RVP of
10.5 psi for the nation. Curves for T90, TSo, and Ti0,
also presented in Figure 2-3 support the trend in increasing
gasoline volatility, as well.
Results of the other fuel survey mentioned — conducted by
MVMA — are presented in Table 2-6.[18] Here, instead of
segregating by geographic region, average volatility
characteristics for unleaded regular gasoline (alcohol free)
are shown for ASTM Classes A, B, and C (as defined in Table
2-4). According to these survey results, the average RVP in
Class C areas has increased by almost ten percent over the past
nine years to a level slightly above 11.0 psi. Trends in other
volatility parameters such as Ti0, TSo, %isb/ and
T2ov/l (all defined in the previous section) are also shown
in Table 2-6. These trends also indicate increasing fuel
volatility over the past few years (i.e., lower Ti0, TSo,
and Tzov/u and higher %iSs). It is important to note that
%iss — close to the %i6o parameter associated with
hot-soak emissions in the previous section — has increased
significantly by 28 percent (from 25.5 percent to a current
average level of 32.6 percent) in Class C areas.
Pool average gasoline volatility properties for Class C
gasolines are similar to those of unleaded regular gasoline
described above. From 1977 to 1986, pool average RVP and
Sealed Housing Evaporative Determination, which is the
current test procedure.
-------
2-21
Figure 2-3
Volatility Trends In Leaded Gasoline
(NIPER Survey Results) [19]
1
1
| ! '1 l
! I
1 '
1
_ i
_^l
Summer ] ,
!
1
!
i
V
1 1 1
i i i
1
i
1
230
220
2 10
200
I 90
ISO
140
I 30
120
I 10
100
90
1959 60 61 '62 63 '64 '65 '66 '6 7 '60 '69 '70 '71 '72 '73 *74 75 '76 '77 '78 '79 '00 '8l '82 '83 '84 '85
O
1
1
1
1 1
1 !
j
/_ I
\ t
- 110 V. Eva
1 1
1
®J
1 1
Trends of certain characteristics of leaded (regular) grade gasoline through summer i960;
leaded antiknock index (R+M)/2 below 93 0 grade gasoline beginning winter 1960-61
-------
Table 2-6
MVMA Survey Results(18J:
Summer Gasoline Trends by ASTM Volatility Claa3»
Year
Volatility
Class
No. of
Gasolines
Avg.
RVP
% of Sample
Above ASTM
Avg.
T i o
Avg.
Tso
Avg.
% i s I
Avg.
Tzov/L
% of Sample
Below ASTM
Minimum
1977
A
37
8.53
27.0
131.6
225.9
21.5
144.3
24.3
B
66
8.72
0.0
128.0
218.3
23.4
141.6
4.6
C
121
9.94
0.0
121.9
220.9
25.5
135.8
0.0
1978
A
38
8.25
13.2
129.0
222.9
22.5
144.5
23.7
B
68
8.56
0.0
127.8
219.8
23.2
142.4
4.4
C
123
9.67
0.8
120.8
220.0
25.8
136.5
0.0
1979
A
37
9.79
64.9
124.4
223.7
24.0
137.4
62.2
B
67
10.10
55.2
123.6
219.4
25.0
135.2
47.8
C
120
11.33
37.5
115.2
221.4
28.0
128.8
22.5
1980
A
39
8.27
20.5
123.8
222.5
23.4
143.3
28.1
B
66
8.71
0.0
120.9
217.9
25.6
140.1
4.6
C
124
9.88
1.6
113.0
218.5
28.4
133.8
0.8
1981
A
41
8.65
22.0
122.4
219.1
25.8
140.9
26.8
B
66
9.30
6.1
122.0
218.1
25.9
137.9
10.6
C
126
10.46
1.6
114.6
215.6
29.3
131.3
4.0
1982
A
37
9.16
37.8
123.8
220.0.
25.2
139.2
43.2
B
65
9.79
33.8
122.5
218.4
26.0
136.1
30.8
C
125
11.06
28.8
114.0
215.6
29.6
126.7
11.2
1983
A
39
9.06
33.3
122.4
220.2
25.4
139.4
46.2
B
64
9.65
31.2
120.1
216.6
26.5
135.9
29.7
C
128
10.84
15.6
113.2
214.7
29.6
129.3
10.9
1984
A
39
8.80
28.2
118.7
210.9
28.5
138.1
51.3
B
60
9.54
28.3
117.5
210.7
28.7
134.8
43.3
C
125
10.89
22.4
108.8
206.7
32.7
126.8
30.4
1985
A
30
9.17
43.3
125.0
214.3
25.9
138.4
56.7
B
59
9.89
45.8
121.4
212.7
27.2
134.5
42.4
_C
106
11.28
40.6
114.9
211.1
29.9
127.2
26.4
1986
A
32
9.80
68.8
115.1
203.4
30.3
132.0
81.2
B
60
9.86
45.0
117.5
208.0
29.2
133.0
43.3
C
106
11.06
30.2
108.9
202.4
32.6
125.4
40.6
Alcohol-free unleaded regular gasoline only (R + M/2 less than 90).
-------
2-23
%i58, as determined from MVMA survey data (unleaded regular,
unleaded premium, and leaded regular gasolines weighted 65
percent, 18 percent, and 17 percent, respectively) have both
increased significantly. RVP has increased from 9.94 to 11.05
psi, and %i58 increased from 26.6 to 32.2 percent.
Alcohol blends, as a whole, have higher volatility than do
alcohol--f ree gasolines. Laboratory experiments yield an
increase in RVP of between 0.5 to 1.0 psi for splash-blended
ethanol blends. Based on the MVMA summer surveys, the average
RVP of in-use ethanol blends tends to be closer to 1.0 psi
greater than the average RVP of in-use straight gasolines. The
MVMA survey data reviewed above for alcohol-free gasoline
volatility properties contained data on ethanol blends, but did
not have volatility properties for methanol blends. Average
RVP for eleven 1985 ASTM Class C unleaded regular gasoline
samples containing an average of 9.1 percent ethanol was 12.5
psi. Average %iS« of these blends was 44.0 percent. These
levels are very similar to those of twenty 1983 and 1984
gasoline samples containing an average of 9.4 percent ethanol.
Average RVP from these 1983 and 1984 ethanol blend samples was
12.3 psi, and average %1S8 was 43 percent. Over the three
years, %iS8 ranged from 30 to 55 percent while RVP ranged
from 10.9 to 13.4 psi. These data indicate that Class C
ethanol blends are significantly more volatile than Class C
alcohol-free gasolines.
According to some theories, the increase in fuel
volatility seen in alcohol-free gasoline is linked to the lead
phasedown in gasoline over the past decade. Because of the
reduction and/or elimination of lead in gasoline, the
traditional octane-booster, refiners must process heavier
crudes in order to obtain the clear, high-octane fractions. As
more crude undergoes hydro-cracking, more butane is produced.
Since butane enhances octane, and because the supply is in
excess, it is allowed to remain a component of the gasoline.
Unfortunately, butane is also a major volatility enhancer and
its abundance is most likely the major reason for the increase
in RVP over time.
As mentioned earlier, ASTM's volatility specifications are
not enforceable by law, but are merely levels agreed upon by
members of the refining industry and automobile manufacturers.
However, some states have adopted ASTM's RVP limits as part of
their own gasoline inspection laws, which are enforceable.
Among the states which have adopted fuel volatility
controls, the assigned RVP limits vary from month-to-month (as
they do in the ASTM specifications) according to temperature
conditions.* For the purposes of comparison to ASTM's D-439
The only exception to this is Hawaii, which holds RVP
constant throughout the year at 11.5 psi.
-------
2-24
limits, July was focused on, because of the high number of
ozone violations which occur during this summer month (as
mentioned in Section I of this chapter).
A comparison of non-California state laws versus ASTM
limits on RVP for July is presented in Table 2-7. As shown, 22
of the states do not currently have laws governing the RVP of
gasoline sold within their boundaries.* Another 22 states have
simply adopted ASTM's current year-round D-439 limits as law;
in addition to these, three more states have RVP limits that
correspond to ASTM's specifications during July but not all of
the rest of the year. Among those state laws that differ from
ASTM's D-439, there is a month-to-month variation involved in
the comparison; for example, Arizona is less restrictive than
ASTM m March, April, and between August and November. Since
July is the focus here, the comparison is simplified. During
July, three states are less restrictive than ASTM. No states
are more restrictive than ASTM during July.
In an effort to determine the effectiveness of ASTM's
recommendations and State laws, fuel survey data were compared
to these standards. Part of this comparison was shown
previously m Table 2-6. As indicated there, over 30 percent
of the Class C fuels sampled by MVMA in 1986 exceeded ASTM's
maximum RVP specification of 11.5 psi; of these same samples,
over 40 percent were below ASTM's recommended minimum level for
Tiov/i. •
In Table 2-8, this comparison is put on a state-by-state
basis. Here, the states are divided between those that have
implemented gasoline RVP standards and those that have not. In
turn, 1982-84 NIPER summer survey results for these states were
compared to both ASTM and State RVP standards in effect at that
time for the month of July. As shown, 11 states had average
RVPs above their respective ASTM specifications; ten of these
states had their own RVP standards and the other one did not.
Further, of the 28 states which had their own RVP limits
(Hawaii excluded), roughly one-third of them had average summer
RVPs above these State standards. Therefore, enforcement
appears to be somewhat ineffective.
Several basic conclusions can be made from the information
presented in the above discussions. One, gasoline volatility
has been gradually increasing over the past two decades and no
substantial data exist to indicate that this trend will not
continue. Two, ASTM recommendations and State-implemented
volatility limits appear to be somewhat ineffective in
controlling gasoline volatility. Even if ASTM specifications
Washington, D.C. is included as one of the 22 states.
-------
2-25
Table 2-7
Comparison of July RVP control: State Laws versus ASTM Limits*
States Same as ASTM
Arizona (A)**
Arkansas (B)
Colorado (B/A)
Delaware (C)
Florida (C)
Georgia (C/B)
Hawaii (C)
Idaho (B)
Illinois (C,C/B)
Iowa (B/C)
Louisiana (C/B)
Michigan (C)**
Minnesota (C)
Mississippi (C/B)**
Missouri (B)
Montana (B)
North Carolina (C/B)
North Dakota (B)
Rhode Island (C)
South Carolina (C/B)
South Dakota (B)
Utah (B/A)
Virginia (C)
Wisconsin (C)
Wyoming (B)
States Less
Restrictive than ASTM**
Alabama (C/B)
Indiana (C)
Maryland (C)
States With No
RVP Specifications
Alaska (D)
Connecticut (C)
Washington, D.C. (C)
Kansas (B)
Kentucky (C)
Maine (C)
Massachusetts (C)
Nebraska (B)
Nevada (A,B)
New Hampshire (C)
New Jersey (C)
New Mexico (A)
New York (C)
Ohio (C)
Oklahoma (C)
Oregon (B,C)
Pennsylvania (C)
Tennessee (C/B)
Texas (A,B)
Vermont (C)
Washington (B,C)
West Virginia (C)
July ASTM volatility class specifications for each state given in
parentheses, excluding California.
States whose year-round volatility controls differ from ASTM
specified levels.
Sources:
ASTM's Standard Specification for Automotive Gasoline, D-439-83.
"Summary of State Regulation of Reid Vapor Pressure in the
Forty-Nine Non-California States" prepared for EPA by ESI
International, Inc., July 25, 1986.
-------
2-26
Table 2-8
Comparison of 1982-84 NIPER Survey Results [191 to July ASTM and State RVP Standards
States with RVP Standards States without RVP Standards
A* A/B B B/C C
Max. ASTM RVP (psi) 9.0 10.0 10.0 11.5 11.5
Avg. of State
RVP Standards (psi) 9.0
9.8 10.1 10.5 11.7
A* A/B
9.0 10.0
B B/C C
10.0 11.5 11.5
No. of States**
No. of States
with Average
RVPs above
ASTM Specs.
No. of States
with Average
RVPs above
State Stds.
14
* "A" through "C" designate ASTM volatility classes as defined in ASTM's
D-439 and reviewed in Table 2-4.
** Hawaii and Alaska excluded due to lack of fuel survey data.
-------
2-27
were currently restricting RVP, which they may indeed be doing
for some gasoline refiners, there is some speculation that ASTM
specifications could be changed in the future. Revisions have
been made in the past as vehicles have been designed to handle
more volatile fuels. The projected widespread use of fuel
injection systems continues this trend, so a relaxation of the
current ASTM RVP limits is not inconceivable. Third, the
current average RVP in Class C areas is roughly 11.0 psi, which
is approaching the maximum ASTM specification of 11.5 psi.
Based on these observations, then, this study assumes that, by
1988, gasoline RVP will on average rise to equal the ASTM
limits for each state.
in. Alcohol Blends
As a result of the oil crisis in the early 1970's, efforts
were begun in the United States to reduce the dependence on
imported oil. One idea introduced was the use of alcohols in
gasoline to extend the supply of gasoline. The Environmental
Protection Agency (EPA), under Section 211(f) of the Clean Air
Act, must approve any new unleaded fuels which are not
substantially similar to fuels already permitted in the
market. There is currently one ethanol/gasoline blend and two
methanol (and cosolvent)/gasoline blends allowed by waivers. A
list is shown in Table 2-9. Currently in the United States,
ethanol blends comprise around 7 percent of the total gasoline
market whereas no methanol blends are currently being marketed
primarily because of low oil prices.[20]
The addition of a polar alcohol affects the properties of
the non-polar gasoline to which it is added. The primary
effect is on the distillation curve and, thus, the parameters
associated with volatility. Two of these parameters, as
discussed earlier for gasoline are RVP, and %i60-
Although the exact distillation effects of alcohol
addition to gasoline vary for every base gasoline, some general
effects are common. Theoretically, the addition of a pure
compound to gasoline causes the percent evaporated at a
specified temperature, compared to the straight gasoline, to
decrease at temperatures below the boiling point of the
compound added. At that point, that compound is distilled off
and the percent evaporated at higher temperatures (compared to
straight gasoline) is increased by the presence of the
compound. However, both methanol and ethanol form azeotropes
with many of the components in gasoline. An azeotrope is a
mixture of a particular composition of two or more components
which, in the case of alcohols and gasoline, has a constant
boiling point lower than that of either individual compound.
The combination of these two effects causes the percent of fuel
evaporated at certain temperatures to be greater for an alcohol
-------
2-28
Table 2-9
Current Clean Air Act Section 211(f) Waivers
Name
Gas Plus Inc.
"Gasohol"
2. ARCO "Oxinol"
DuPont
(original)
(reconsidered)
Date Granted
12/16/78
(w/o decision)
11/7/81
1/14/85
10/22/86
Limitations
-up to 10% (vol.) anhydrous
EtOH
-no volatility limits
-up to 4.75% (vol.) MeOH
-up to 4.75 (vol.) TBA
-ratio of MeOH: TBA cannot
exceed 1
-ASTM gasoline volatility
1imits
-up to 5% (vol.) MeOH
-at least 2.5% (vol.)
cosolvent (EtOH, propanol,
butanols)
-corrosion inhibitor
-ASTM gasoline volatility
limits per reconsideration
of original waiver.
EtOH = ethanol.
MeOH = methanol.
TBA = tertiary butyl alcohol.
-------
2-29
blend than for straight gasoline, and to an extent much greater
than the percentage of the alcohol alone. This effect on
distillation occurs primarily in the boiling temperature range
from the lowest-boiling azeotrope to the highest-boiling
gasoline component forming an azeotrope.
The effect on the distillation curve is also dependent on
the alcohol added. The increase in %iSo for ethanol blends
is between ten to 15 percent. The increase in %ieo for
methanol/cosolvent blends is five to ten percent. Typical
distillation curves for ethanol blends and methanol blends are
shown in Figures 2-4 and 2-5, respectively.
As discussed earlier, RVP is a measurement of the
volatility of a fuel at 100°F. Since the vapor pressures of
methanol and ethanol are lower than gasoline at 100°F, the
addition of these alcohols to a gasoline could theoretically
cause the RVP of the alcohol blend to be lower than that of the
straight gasoline. Once again, however, the presence of highly
volatile azeotropes offsets this effect and causes an increase
in the RVP. The precise increase in RVP varies for every base
gasoline, but the increases from each alcohol are generally
similar. For the addition of methanol with a cosolvent
(ethanol, propanols, and/or butanols) to gasoline at five-ten
percent of total volume, the increase in RVP two to 2.5 psi.
The increase in RVP for the addition of two-ten percent (by
volume) ethanol to gasoline is 0.5 - 1 psi.[14]
The effects of alcohol addition on RVP, and %i6o can be
counteracted by adjusting the contents of the gasoline to which
the alcohol is added. By removing lighter hydrocarbons (such
as butanes and/or pentanes), the RVP or %16o of the final
blend can be controlled to levels of the original straight
gasoline.
However, even with control of alcohol blend volatility,
there can be an increase in in-use volatility due to
intermittent use of blends and gasoline. A phenomenon called
commingling can occur when an alcohol blend is added to a tank
partially filled with straight HC gasoline, wherein the RVP of
the mixture can be significantly higher than the RVP of the
original gasoline or blend. Commingling is depicted in Figure
2-6, where the increase in RVP (over the straight HC gasoline
level) is shown as a function of blend ratio and type of
blend. For example, in the top plot, if a 1:1 MeOH/isopropyl
alcohol blend (9.6 percent by volume) is added to fill a
60-percent full tank of straight HC gasoline, the RVP of the
new mixture (i.e., full tank) will be roughly 1.0 psi higher
than that of the original straight HC gasoline.
-------
2-30
Figure 2-4
Typical Methanol Blend Distillation Curves
tar caai [vimtiM
-------
2-31
Figure 2-5
Typical Ethanol Blend Distillation Curves
-------
2-32
Figure 2-6
COMINGLINQ VAPOR PRESSURE EFFECTS M«oh/ipa
BUENOS WITH ALL HYDROCARBON GASOLINE
•UNO »*TIQ.%
COMINGLING VAPOR PRESSURE EFFECTS
EtOH ANO M«OH/EtOH BLENOS WITH
ALL HYDROCARBON GASOLINE
2.0
I 3
1
| • 0
4
\
ioi
-a.5
'00 K> $0 70 iO 50 *0 JO JO '0 0
6A90UMC 0 <0 20 )0 «0 >0 «0 ^0 M M <00
KCNO RATIO, %
Source;rrco Petroleum Products Company, March 12, 1985
(in letter to Craig Harvey, EPA)
-------
2-33
The increase in volatility due to commingling has been
shown to be a function of the percentage of stations selling
blends, the amount of fuel remaining in the fuel tank when
refilling, and the habits of the buyer (i.e., whether fuel is
bought at random or loyally).
iv. Ambient Temperature Conditions
In addition to RVP, daily ambient temperature conditions
can also impact the level of evaporative emissions. Diurnal
losses are dependent upon not only the ambient temperature
excursion (daily maximum minus daily minimum), but also on the
absolute magnitude of these temperatures. Hot-soak losses are
also dependent on ambient temperature. For example, a vehicle
undergoing a 30°F diurnal change at an average temperature of
90°F would be expected to have significantly greater diurnal
and hot-soak losses than a vehicle experiencing only a 15°F
diurnal difference at an average temperature of 75°F. FTP
temperature specifications are 60° to 84°F for the diurnal test
and 81°F for the hot-soak test. These temperature conditions
are not completely representative of in-use summer temperature
conditions. Based on daily July 1985 weather data from the
National Oceanic and Atmospheric Administration, diurnal
temperature swings tend to be less than 24°F but start at
temperatures higher than 60°F. The temperature of the hot-soak
will depend upon the time of the day at which it occurs.
In an effort to quantify the magnitude of such temperature
(and RVP) differences, an EPA-sponsored test program was
carried out to measure diurnal and hot-soak losses at various
ambient temperature conditions. This test program conducted
for EPA at the Automotive Testing Laboratories (ATL) included a
matrix of three fuel RVPs, three diurnal starting temperatures,
three diurnal temperature excursions, and three average
hot-soak temperatures.[23] The standard EPA test procedure
(i.e., diurnal temperatures between 60°F and 84°F, and average
hot-soak temperature of roughly 82°F) was represented in the
full test matrix.
Diurnal emissions data from this ATL program are analyzed
later in this chapter. There, the diurnal averages are
compared to theoretical emissions indexes calculated for each
ATL test condition via the diurnal emissions model developed
for this analysis. Hot-soak emission data will not be analyzed
because of inconsistencies between hot-soak emission
measurements at EPA and ATL.
v. Fuel Weathering
The volatilities reported in various fuel surveys (e.g.,
NIPER, MVMA) represent those levels measured at the gasoline
-------
2-34
pump. However, as the gasoline in the vehicle's fuel system
responds to daily diurnal temperature changes and engine heat,
some of the lighter hydrocarbons are lost. Thus, the
volatility of the fuel gradually decreases. This phenomenon of
"weathering" is an important consideration in assessing
evaporative emissions that actually occur in the field. The
analysis of weathering effects on in-use diurnal emissions is
presented later in this chapter. The effects of weathering on
hot-soak emissions will not be examined since neither ambient
temperature effects nor temperature and RVP variability effects
have been included in the in-use hot-soak model. It is
expected that including these effects would tend to offset much
of the predicted decrease in hot-soak emissions due to fuel
weathering as occurs with the diurnal model.
vi. Driving Patterns
The way in which a vehicle is driven (i.e., the number of
trips per day, the length of the trips, and the time at which
the trip is taken) affects the amount of HC emissions from a
motor vehicle. As is explained in the following sections, the
effects of temperature, fuel volatility, fuel weathering, and
in-use fill level have been taken into account in this
analysis. However, the effects of the driving pattern
distribution on emissions are not taken into account. The
complicating factor is that an unlimited number of driving
patterns exist. Driving patterns vary anywhere from vehicles
which do not drive for several days in a row, whose canister
can become fully saturated, to vehicles which have many trips
per day, and possibly experience an interrupted diurnal but
many hot-soaks. Other vehicles regularly experience long trips
during which the canister is completely purged, resulting in
very low emissions.
The revised MOBILE3 model used for this analysis does vary
the number of trips per day and number of miles travelled per
day by vehicle age. The average fleet values are 3.05
hot-soaks per day, 1 diurnal per day, and 26 miles per day.
(The trip length in the June 1984 MOBILE3 report was 31.1 miles
per day.) However, due to the complexity of modeling in-use
driving patterns and the lack of data from which to derive the
effect of driving patterns on both exhaust and evaporative HC
emissions, the effect of different driving patterns has not
been included in this analysis.
Failing to include the effects of che driving pattern
distribution on in-use evaporative emissions may not have as
large an effect on the results of this study as might be
expected. For example, in the long-term, when certification
fuel RVP has been made equal to in-use fuel RVP (and thus the
excess emissions, which occur when operating a vehicle on a
-------
2-35
fuel more volatile than it was certified on, are nearly
eliminated), the incremental evaporative emission reductions
due to lowering in-use fuel volatility from current levels,
arise primarily from malmaintained/defective vehicles, tampered
vehicles and the gasoline distribution system.
With respect to the first two emission sources, the
evaporative control systems on these vehicles are inoperative
and result in the release of uncontrolled evaporative emissions
directly into the atmosphere. Canister purge is not likely a
factor in determining these emissions, and it is canister purge
that is most affected by the variety of driving patterns.
Thus, the total amount of hot-soak emissions from these
vehicles would depend on the total number of trips taken (i.e.,
the total number of hot-soaks experienced), but the emissions
per hot-soak would be independent of the number of trips per
day and the amount of driving. Since MOBILE3 represents the
average nuniber of trips per vehicle very well, it should
represent total hot-soak emissions from these vehicles quite
well, despite the lack of driving pattern distribution data.
On the other hand, the total amount of diurnal emissions
from these vehicles could be affected by driving patterns if a
trip interrupts the daily diurnal. During a trip, however, the
fuel in the gas tank heats up to temperatures above the ambient
temperature. This results in "running losses" which, on a
properly operating vehicle, are vented through the evaporative
control system to the engine (or directly to the engine) and
combusted. In malmaintained/defective vehicles and tampered
vehicles, these running losses would essentially be vented to
the atmosphere. The interruption of the diurnal by driving
could actually increase emissions over that of an uninterrupted
diurnal. Driving at other times during the day would actually
create artificial diurnals which results in more emissions to
the atmosphere; these running losses are not considered by our
evaporative emissions model. Thus, our model may actually
underestimate the incremental emission control associated with
additional long-term in-use RVP control.
The absence of driving patterns still may have an effect
on the emissions of properly operating vehicles. However, as
will be seen in Chapter 6, the initial step of matching in-use
and certification fuel RVP at 11.5 psi in Class C areas is so
cost effective that a 50 percent decrease in the estimated
evaporative emission reduction would still leave the step very
cost effective (i.e., less than $100 per ton).
b. Results of In-Use Motor Vehicle Testing
This section presents test data from EPA's in-use motor
vehicle emission factor (EF) program m an attempt to quantify
-------
2-36
the effect of the factors mentioned in the previous section
(i.e., fuel volatility, temperature and evaporative control
system design) on emissions from current vehicles. The section
begins with a brief description of the current evaporative test
procedure used in certification and changes made for the in-use
EF program (i.e., the addition of commercial fuels and the
switch in fuel sequence). Next, general evaporative emission
results from the EF program are presented for various RVP
levels with a comparison between the revised estimates and the
MOBILE3 figures published in June 1984. In the following
section, these revised hot-soak and diurnal emission levels are
broken down into several basic components of motor vehicle
evaporative emissions, based on an analysis of the vehicle test
fleet. By attributing certain portions of current total
evaporative losses to different sources (i.e., excess RVP,
malmaintenance/defect, improper design of purge system, etc.),
it is possible to estimate the effect that changes in in-use
RVP or certification fuel and test procedure will have on each
component and, thus, on total evaporative emissions. (The
effects of the various control strategies will be outlined, in
detail, for various model years in Chapter 3).
i. Test Procedure
The standard evaporative test procedure used in the
certification of new vehicles is outlined in detail in Part 86
of Title 40 of the Code of Federal Regulations (40 CFR 86).
Briefly, the vehicle is first drained of its fuel and refueled
with Indolene (with an average RVP of 9.0 psi) to the
40-percent full level. The vehicle is then preconditioned
using the Urban Dynamometer Driving Schedule (also referred to
as the "LA-4" cycle), which lasts approximately 23 minutes.
Under special circumstances (e.g., if the vehicle was
transported via carrier instead of being driven to the test
site), manufacturers can request up to two LA-4 cycles to
assure adequate purging of the evaporative canister. Following
preconditioning, the vehicle is stored (or "soaked") for a
period of 12-36 hours before the SHED (Sealed Housing
Evaporative Determination) test is conducted.
Just prior to beginning the SHED test for diurnal losses,
the vehicle is drained and refueled with chilled Indolene to
the 40-percent fill level.* Beginning with a fuel and tank
temperature of 60°F, the fuel is heated to 84°F over a period
of one hour, at which time the final HC concentration in the
SHED enclosure is recorded as the total diurnal loss.
The fuel is chilled to about 50°F to counteract the warm
fuel tank (70-75°F), which is already warmer than the
lower end of the temperature excursion (60°F).
-------
2-37
Following the diurnal test, a cold-start LA-4, followed by
the first half of a hot-start LA-4 is performed, during which
exhaust emissions are measured. At the completion of these
tests, the engine is shut off and the vehicle is pushed into
the SHED enclosure for the hot-soak test. The test vehicle
remains there for one hour at an average ambient temperature of
81°F* and the increase in HC concentration is recorded and
converted into a total hot-soak mass emission.
The estimation of evaporative emissions as a function of
fuel volatility used to evaluate the control options examined
in this study was based on data generated as part of EPA's
ongoing in-use emission factor (EF) test program. This program
involves the testing of in-use (privately-owned) passenger
cars, selected at random from State of Michigan vehicle
registration files. Prior to November 1983, in-use
(privately-owned) vehicles were evaluated for hot-soak and
diurnal losses only while operating on certification test fuel
(Indolene) with a 9.0-psi RVP. Since then, however, the effect
of fuel volatility on emissions has been examined with the
addition of two commercial fuels with nominal RVPs of 11.5 psi
(added in November 1983) and 10.5 psi (added in August 1984).
The test procedure used in the EF program has basically
followed certification practices except for certain differences
in vehicle preconditioning and, of course, the addition of
commercial fuels. The in-use EF test sequences are summarized
in Table 2-10. Between November 1983 and July 1984, vehicles
were preconditioned over a shortened LA-4 cycle which lasted
only ten minutes instead of the entire 23 minutes. Commercial
fuel with an 11.5-psi RVP was added to the test sequence
following all evaporative and exhaust emission tests conducted
on Indolene.
In July 1984, the shortened prep cycle was dropped and the
entire LA-4 cycle (used in certification) was reinstated.
Also, at this time, it was concluded that the evaporative tests
on ll.5-psi commercial fuel may have been unrepresentative
because they were run following a battery of tests on 9-psi
Indolene over which the evaporative canister was repeatedly
purged. Therefore, the commercial test was begun with an
essentially "unloaded" canister, which would probably bias
results toward lower emissions than those experienced in the
field. Further, because vehicles had been operated on 11.5-psi
commercial fuel prior to arriving at EPA, it follows that the
most accurate measurement of in-use emissions would be obtained
by testing commercial fuel first. Therefore, the test sequence
The hot-soak temperature range specified in the CFR is
68-86 °F.
-------
2-38
Table 2
Comparison of In-Use
Nov. 83 - July 84
1. Park indoors and/or
outdoors
2. Shortened dynamometer
(10-min.)
3. Indolene evaporative
tests
4. Exhaust emission tests
(HFET and short tests)
5. Commercial (11.5 psi)
evaporative tests
-10
Test Sequences
Post - July 84
1. Park indoors; loosen gas
cap
2. LA-4 dynamometer prep
(23-min.)
3. Commercial (11.5 psi)
evaporative tests
4. Mixture (10.5 psi)
evaporative tests*
5. Indolene evaporative
tests
6. Exhaust emissions tests
(HFET and short tests)
10.5 RVP added in August 1984
-------
2 -39
was changed and the 11.5-psi fuel was tested first, followed by
Indolene.
In August 1984, testing of the mid-range commercial fuel
with an RVP of roughly 10.5 psi was added to the sequence just
after the 11.5-psi fuel and before any Indolene testing. Other
minor changes (initiated in July 1984) involve the storage of
the vehicle prior to any evaporative tests. To avoid premature
saturation of the canister prior to testing, the gas cap is
loosened to allow vapors to bypass the control system and the
vehicle is stored inside at a fairly constant temperature.
n. General Test Results
Vehicle samples used in EPA's in-use EF testing both prior
to and after the July 1984 test procedure changes are broken
down by manufacturer in Table 2-11.* Carbureted and
fuel-injected samples are outlined separately, and sample
distributions are compared to 1984 market shares.
Evaporative emissions from these test vehicles are
summarized in Table 2-12. Results are shown for both Indolene
and commercial (11.5-psi) fuels, and are separated out by
vehicle type and test procedure. The MOBILE3 emission rates
published in June 1984 are from a subset of the November
1983-July 1984 tests and are listed separately for comparison.
The estimates used for this analysis are listed as July 1984 -
March 1986 results.
As indicated from the results shown in Table 2-12, the
change in the test procedure has had mixed effects on emissions
with Indolene and higher emissions with commercial fuel in all
cases. (This is as to be expected from the previous discussion
on the reasons behind the changes in fuel test sequence.) For
fuel-injected vehicles, higher emissions also may be partially
due to the higher average mileage of the vehicles being tested;
the vehicles tested after July 1984 have an average mileage
almost twice that of those tested previous to that time.
Results of the July 84 - March 1986 testing are shown
graphically in Figure 2-7, which plots revised hot-soak and
diurnal emissions versus fuel RVP.
Both sets of vehicles tested show average emissions
exceeding the 2-gram standard for total evaporative emissions,
even while operating on Indolene: carbureted vehicles
The post-July 1984 vehicle sample includes vehicles tested
only through March 1986, the point at which test results
were "frozen" for this analysis. Vehicles tested after
the cut-off date, are being tested only with two fuels of
9.0 psi and 11.5 psi RVP.
-------
2-40
Table 2-11
In-Use EF Vehicle Sample Distribution
Fuel-Injected
Carbureted
1984
1984
Nov. 83-
July 84-
Market
Nov. 83
- July 84-
Market
Manufacturer
July 84
March 86
Share
July 84
March 86
Share
GM
31(33%)
45(29%)
46%
57(74%)
75(46%)
42%
Ford
38(41%)
40(24%)
13%
9(12%)
38(23%)
25%
Other Domestic*
12(13%)
32(21%)
11%
—
25(15%)
11%
Toyota
2 ( 2%)
14( 9%)
4%
—
13( 8%)
10%
Nissan
3( 3%)
15(10%)
5%
3( 4%)
8 ( 5%)
9%
Other Imports**
7( 8%)
9 ( 5%)
21%
8(10%)
4 ( 3%)
3%
Total
93(100%)
155(100%)
100%
77(100%
) 163(100%)
100%
* AMC, Chrysler, VWA.
Renault,
Fuji .
** Honda, VWG,
Mitsubishi,
Toyo-Kogyo,
Audi,
-------
2-41
Table 2-12
Comparison of Evaporative EF Test Data from
Non-Tampered Vehicles (g/test)
Indolene (
Test Period Technology N
Nov 83 -
July 84 Carb. 93
Fuel-Inj. 77
Published Carb. 53
MOBILE3* Fuel-Inj. 62
July 84 -
March 86** Carb. 155
Fuel-Inj. 163
= 9.0 psi)
Mileage Diurnal Hot-Soak
45000 4.16 2.19
20000 1.64 0.96
60000 4.22 2.74
20000 2.21 1.12
58000 2.60 2.57
35000 1.67 0.88
Test Period Technology
Commercial Fuel (RVP = 11.5 psi)
N
Mileage Diurnal Hot-Soak
Nov -
July 84
Published
MOBILE3*
July 84-
March 86**
Carb.
Fuel-Inj
Carb.
Fuel-Inj
Carb.
Fuel-Inj
93
77
53
62
155
163
45000
20000
60000
20000
58000
35000
8.64
2.33
9.31
3.13***
10.33
8.51
3.29
1.28
3.98
1.55
4 .26
2.70
* The MOBILE3 results are from a subset of the Nov. 83 -
July 84 data pool.
** Revised MOBILE3 estimates, used in this study.
*** This is the average of actual test results. However, due
to uncertainties associated with the low mileage, the
value for carbureted vehicles was also used for
fuel-injected vehicles in MOBILE3.
-------
Figure 2-7
Non —Tampered Evap. Emissions vs. RVP
fuel RVP (psl)
-------
2-43
averaging 5.17 grams/test and fuel-injected vehicles averaging
2.55 grams/test. When tested on ll.5-psi commercial fuel,
these evaporative emissions are much larger: 14.59 grams/test
for carbureted vehicles and 11.21 grams/test for fuel-injected
vehicles. The possible causes of the excess evaporative
emissions are the subject of discussion in the following
section.
iii. Components of Excess Motor Vehicle Evaporative
Emissions
For purposes of analysis, it is useful to segregate the
evaporative emission excess described above according to its
probable causes. Using EF test data, these four basic
categories were chosen as: 1) effect of insufficient design
capacity/purge, 2) effect of malmaintenance and defects, 3)
effect of excess RVP, and 4) effect of evaporative system
tampering. The causes and magnitude of each of these effects
(for current vehicles) are detailed below. More detailed
information on the methodology used to separate these four
sources is provided in Appendix 2-A.
Insufficient Design Capacity/Purge
Even with properly functioning evaporative emission
control systems, many current vehicles fail to meet the 2-gram
standard on Indolene. This is primarily seen in carbureted
vehicles which average 3.12 grams/test when there are no
apparent malfunctions.* Conceivably, this failure could be due
to an insufficient capacity to store or purge vapors inherent
in the system design. It may also be valid to attribute some
of these excess emissions to the temporary effects of high RVP
fuels on charcoal working capacity.
The plausibility of insufficient design capacity and purge
is evident from the limitations of the evaporative emission
test procedure and certification process. To be certified, a
vehicle must meet the standard after applying an additive
deterioration factor to low mileage emission test data. Also,
MSAPC Advisory Circular No. 50A states that vehicles should be
able to pass the evaporative emission test when starting the
test sequence with a saturated canister. (Many EF vehicles are
probably received in this condition.) This means that purge
systems should be designed to completely purge a loaded
canister with the LA-4 prep cycle. However, there is no
requirement for this in the certification test procedure.
Thus, it is possible that a certification vehicle, as currently
As shown in Appendix 2-A, Table 2-A-2, for "problem-free"
vehicles (to be explained in the next few paragraphs).
-------
2-44
designed, would fail if it began the test with a saturated
canister.
The derivation of the magnitude of the insufficient design
capacity/purge effect is detailed in Appendix 2-A. The general
concept involved was to compare the average emission levels of
the "problem-free" vehicles, as defined in Table 2-13, with the
standard level of 2 grams/test for total evaporative
emissions. The difference between the two levels was assumed
to be due to the emission control system design. This effect
is not seen in fuel-injected vehicles, but averages 1.12
grams/test for carbureted light-duty vehicles. The
insufficient design capacity/purge effect is noted here because
it presumably would be eliminated for new vehicles by revising
the evaporative emission test procedure to require that a
vehicle begin the test with a saturated canister.
Malmaintenance and Defects
Non-tampered vehicles with the maintenance problems and
hardware defects listed in Table 2-13 (tampering is considered
separately below) will generally have higher evaporative
emissions than well-maintained vehicles. These problems can
lead to either a partial increase in emissions (e.g. from a
dirty canister filter) or to completely uncontrolled emissions
(e.g., from an inoperative canister purge solenoid or valve).
On average, excess emissions would not be expected to be as
high as the uncontrolled emission baseline because some purging
could still occur. This is different from the case of
tampering where complete system disablement is generally the
result.
The magnitude of the effect of malmaintenance and defects
for current vehicles operating on Indolene is estimated by
considering the difference between the non-tampered EF vehicle
sample average and the problem-free vehicle sample average
tested on Indolene. Inherent in this calculation is the
assumption that the EF sample has a representative
malmaintenance and defect rate. Table 2-14 compares the
malmaintenance and defect rate for various vehicle data bases.
For carbureted vehicles, the malmaintenance and defect rate
from EF cars tested since July 1984 (35 percent) is within the
range and essentially equal to the average of all other
available data samples. For fuel-injected vehicles, the newer
EF sample has a rate slightly higher than the older EF sample
(e.g., 12 percent in the more recent sample versus five percent
in the older sample). However, this is probably due to the
fact that the fuel-injected vehicles in the old EF sample
averaged fewer total miles on their odometers than those in the
current sample (20,000 vs. 35,000). Overall, then, the
malmaintenance and defect rates in the July 1984 - March 1986
EF sample appear representative.
-------
2-45
Table 2-13
Conditions Excluding Vehicles
From Problem-Free Sample
Fuel System
A. Carburetor Assembly
1. Loose on Manifold
2. Leaks Fuel
3. Exceptionally Dirty
B. Fuel Injection Components
1. Injectors Leaking
II. Evaporative System
A. Canister
1. Saturated with Fuel
2. Broken
3. Missing*
B. Canister Filter
1. Dirty
2. Saturated with Fuel
3. Missing
Canister Purge Solenoid/Valve
1. Leaks Vacuum
Sticking
Inoperative
Missing*
Disconnected*
Hoses, Lines, Wires
1. Vacuum Line Plugged
2. Vacuum Line Disconnected*
3. Vacuum Line Damaged
4. Vacuum Line Misrouted*
5. Vent Line Damaged
6. Vent Line Disconnected*
E. Other
1. EFE TVS Stuck Open/Closed
2. Non-OEM Gas Cap
3. Bowl Vent Control Valve Always Open/Closed
4. VCV Vacuum Control Valve Inoperative
5. Gas Cap Leaks
6. Sending Unit Gasket Leaking
7. Fuel Tank Rollover Valve Leaking
8. Air Cleaner Assembly Gasket Broken/Missing
9. EFE Control Switch Missing*
10. Gas Cap Missing*
Considered to be tampering.
-------
2-46
Table 2-14
Malmaintenance and Defect Rate Comparison
Carbureted
Fuel
Iniected
Sample
Sample
Size
Defect
Rate
Sample
Size
Defect
Rate
EF (new)*
155
35%
163
12%
EF (old)**
93
44%
77
5%
SwRI[2 4]
27
26%
—
—
API (NIPER)[25]
19
11%
32
13%
API (ATL)[26]
28
25%
10
0%
Average* * *
—
35%
—
10%
* Vehicles tested on three fuels w/LA-4 prep, commercial
fuel first; July 1984-March 1986.
** Vehicles tested on two fuels w/10-minute prep, Indolene
fuel first; November 1983-July 1984.
*** Sample-size weighted.
-------
2-47
While the above methodology appears satisfactory for
estimating the magnitude of the malmaintenance/defect effect
for vehicles operating on 9-psi Indolene, it is expected that
the effect would increase with the level of in-use RVP. The
same would be true for a tampered vehicle. Basically, if
excess emissions are being released from the canister or
elsewhere, increased vapor loadings that result from higher
in-use RVPs will lead to increased emissions. While
redesigning vehicles for higher RVP fuel should lower emissions
when the control system is fully operable, the effect of
malmaintenance or a defect would be expected to be independent
of the RVP for which the vehicle was designed and dependent
only on the RVP of the fuel actually used. Estimation of the
magnitude of this effect at various RVPs is described in detail
in Appendix 2-A.
While the insufficient design/purge and excess RVP effects
(discussed below) are assumed to be totally eliminated in new
vehicles via changes to certification fuel and test procedure,
the malmaintenance/defect effect can only be controlled through
reduction of in-use fuel volatility or possibly through an
effective evaporative system inspection and maintenance program
(See Appendix 6-B).
Excess RVP Effect
The excess RVP effect is defined as the emissions impact
of operating vehicles on a fuel of higher volatility than that
for which their evaporative control systems have been
designed. Canisters and purge systems on current vehicles are
designed to meet the 2-gram evaporative standard when operated
on 9.0 psi Indolene over the standard certification test
procedure. However, as discussed previously in this chapter,
the RVP of current in-use gasoline in many of the ozone
non-attainment areas of the country is significantly higher
than 9.0 psi. Because of the higher volatilities, more
evaporative HCs are emitted than the canister can accommodate,
which results in canister saturation and "breakthrough" of HC
vapors to the atmosphere.
The magnitude of this excess RVP effect can be calculated
by first subtracting total average emissions of the
problem-free sample on Indolene from those on commercial fuel.
The excess RVP effect is assumed to be at least partially
eliminated for new vehicles if certification fuel RVP is raised
to a level equal to or greater than in-use fuel RVP. (Ambient
temperature effects could create a situation where the excess
RVP diurnal effect is not completely eliminated.) However, for
vehicles certified prior to any change in certification fuel
RVP (i.e., 1990 in this analysis), the excess RVP effect will
remain and be dependent upon in-use RVP.
-------
2-48
Tampering Effect
Intentional system disablement also contributes toward
excess evaporative emissions. EPA's in-use EF sample is not
thought to have a representative number of tampered vehicles,
since those who tamper with their emission controls may be
reluctant to lend their vehicles to EPA for testing. The EF
sample is also relatively small due to the high cost of
emission testing. For these reasons, tampering rates to be
used in emissions modeling are developed from EPA tampering
surveys involving thousands of vehicle inspections. Those
conditions considered as tampering are primarily disconnected,
misrouted or missing hoses, missing canisters and missing fuel
caps (as indicated previously in Table 2-13).
Because emissions from tampered vehicles are not developed
from the EF sample, the MOBILE3 program accommodates them
separately. Tampering incidence rates are developed from
survey data and excess emissions are determined from emission
tests on completely disabled systems. Since the original
MOBILE3 estimates were published in June 1984, additional
tampering data has become available that has allowed
improvement over the June 1984 estimates. These revisions to
the MOBILE3 tampering estimates are discussed below.
The original M0BILE3 LDV tampering incidence rates
published in June 1984 were based on a linear regression of
tampering frequency versus mileage using the results of EPA's
1982 Tampering Survey. [27] For LDTs and HDVs, the rate of
increase of tampering with mileage for LDVs was applied to the
average LDT-sample tampering frequency and vehicle mileage,
because the LDT sample was too small to derive a change in
tampering over time and no HDVs were surveyed. The zero-mile
tampering rates, however, were developed from the LDT sample;
these LDT zero-mile rates were also used for HDVs.
As mentioned above, updated EPA survey results from 1983,
1984 and 1985 have since become available, so these data were
added to the 1982 tampering data used to develop the tampering
estimates used for this analysis.[28,29,30] The resulting
information was sufficient to develop separate tampering rate
estimates for both LDVs and LDTs; the HDV rates were still
assumed to be the same as those for LDTs. Additional revisions
to the tampering estimates include the designation of vehicles
with misrouted hoses and missing fuel caps as tampered
vehicles.* Since these conditions were not considered as
As misrouted hoses and missing fuel caps could be
unintentional, there remains some question as to whether
they should be considered "tampering" or "malmaintenance."
However, as these conditions were not felt to be properly
represented in the EF sample, they are currently regarded
as tampering.
-------
2-49
tampering in the June 1984 version of MOBILE3, previously
estimated effects of tampering may have been somewhat
understated. Plots of the revised tampering rates versus
mileage for LDVs and LDTs are shown, respectively, in Figures
2-8 and 2-9.
One possible form of tampering/malmaintenance that still
remains to be investigated is the use of replacement gas caps
not meeting the same specifications as the original gas cap
(referred to as non-OEM (original equipment manufacturer) gas
caps). Such gas caps may not seal properly and could result in
either partially or completely uncontrolled emissions. The
extent of their use and their effect on emissions is currently
being investigated.
In addition to revising tampering incidence rates, the
emission rates for tampered vehicles have also been modified by
supplementing the pre-June 1984 data with more recent data and
by incorporating emission excesses associated with missing fuel
caps. With certain exceptions, it is assumed that all types of
tampering result in completely uncontrolled emissions. A case
in which this assumption may not be strictly valid is for the
disconnection of a carburetor bowl vent line at the carburetor
end (i.e., not at the canister). In this case, only hot-soak
emissions would be uncontrolled. However, because the
tampering survey does not make this distinction, a better
estimate is not available at this time.
The uncontrolled evaporative emission rates used to
quantify the tampering effects are based on SHED testing of
vehicles with removed canisters and/or fuel caps. These are
summarized in Table 2-15, which also shows the increase with
in-use fuel RVP. Several assumptions that were made in
deriving these emission rates are described below.
First, with respect to diurnal losses, the effect of fuel
cap removal is assumed to be the same as canister removal for
both carbureted and fuel-injected vehicles, since diurnal
emissions result entirely from the fuel tank. Also, diurnal
emissions from fuel-injected vehicles with either missing gas
caps or canisters is assumed to be the same as those from
uncontrolled carbureted vehicles (for which more data exist).
Again, diurnal emissions occur entirely from the fuel tank
where the two technologies do not differ. Also, the fuel tank
volumes of the two vehicle types do not generally differ.
Second, with respect to hot-soak emissions from
fuel-injected vehicles, fuel cap removal is assumed to result
in completely uncontrolled hot-soak emissions. A properly
assembled fuel injector should emit little, if any, during a
hot soak, leaving the fuel tank as the primary source of
-------
9
8
7
6
5
4
3
2
1
0
Figure 2-8
LDV Tampering Rates
MOBILES ond New Surveys
Y-EVAP&CAP
Mlleoga/(1 000 miles)
/+¦ SURVEY—EVAP
/
to
I
Ln
o
-------
16
15
1 4
13
12
1 1
10
9
8
7
6
5
4
3
2
1
0
Figure 2-9
LDT Tampering Rates
MOBILE3 and New Surveys
20
—r~
40
60
-r~
80
100
V-EVAP&CAP
Mileage (1000 miles)
SURVEY—EVAP
/
MOBILES
-------
Table 2-15
Uncontrolled Evaporative Emissions (q/test) from Tampered Vehicles vs. RVP*
Canister
Disconnects
Gas Cap
Remova1
Vehicle
Model
Fuel
—9.0
psi—
-11.5
psi**-
—9.0
psi—
-11.5
psi**-
Type
Year
System
H.S.
Dnl.
H.S.
Dnl.
H.S.
Dnl.
H.S.
Dnl.
LDV
pre-71
All
14.67
26.08
22.45
47.99
14.67
26.08
22.45
47.99
and
71
All
14.67
26.08
22.45
47.99
10.91
26.08
16.15
47.99
LDTi
72-77
All
14.67
20.90
22.45
35.45
10.91
20.90
8.98
35.45
78-80
All
13.29
16.32
18.50
25.11
2.32
16.32
3.79
25.11
81 +
Carb
10.36
15.40
17.47
27.61
2.46
15.40
4.27
27.61
Fin]
5.20
15.40
9.00
27.61
5.20
15.40
9.00
27.61
LDT2
pre-79
All
18.08
42.33
27.66
77.89
18.08
42.33
27.66
77.89
79+
Same as LDV,
LDT1
HDV
pre-85
All
18.08
42.33
27.66
77.89
18.08
42.33
27.66
77.89
85+
14.67
26.08
23.31
39.87
3.92
26.08
6.81
39.87
Figures presented are for low altitudes; high-altitude correction factors are as follows: 1)
LDV — pre-1977 = 1.3, 1977 = 1.0, 1978-81 = 2.59, 1982-83 =1.3, 1984+ = 1.0; 2) LDT, ,
LDT2 and HDV — all model years = 1.3.
Values for uncontrolled emissions between 9.0 and 11.5 psi can be calculated via linear
interpolation of the RVP for hot-soak emissions and the UDI for diurnal emissions (except for
gas cap removal carbureted hot-soak emissions which are the same as non-tampered carbureted
hot-soak emissions).
-------
2-53
emissions. Data on thirteen fuel-injected vehicles confirm
this. Hot-soak emissions using Indolene with the gas cap
removed and those with the canister removed were essentially
identical at just over 4 grams/test, where controlled emissions
were below one gram/test.[31 ] Emissions using commercial fuel
without a canister were only slightly higher than those without
a gas cap. Thus, this assumption appears to be valid for
fuel-injected vehicles. However, for carbureted vehicles, the
increase in hot-soak emissions due to fuel cap removal is
expected to be less than totally uncontrolled hot-soak
emissions because the carburetor bowl contributes to, and
probably is the major source of, hot-soak losses from these
vehicles. Since data are not available to precisely predict
the degree to which hot-soak emissions from carbureted vehicles
would increase with fuel cap removal, it will be assumed that
the carburetor bowl dominates and that hot-soak emissions do
not increase. Thus, the values presented in Table 2-15 for
hot-soak emissions from carbureted vehicles with missing fuel
caps are the same as those for non-tampered carbureted vehicles
(i.e., 2.57 and 4.26 g/test for 9.0-psi and 11.5-psi RVPs,
respectively, as shown in Table 2-12).
As explained in Appendix 2-A, the tampering offsets used
in this analysis were calculated by subtracting the average
non-tampered vehicle emissions from the uncontrolled emission
levels. These offsets were then incorporated into the EF runs
at the tampering incidence rates developed from the survey data
previously discussed. These offsets represent extreme
increases in evaporative emissions, reaching combined hot-soak
and diurnal levels of 20-30 grams/test for 11.5 RVP fuel.
(Tampering offsets are shown in Appendix 2-A for light-duty and
heavy-duty vehicles in Tables 2-A-6 and 2-A-8, respectively.)
Summary
Based on the above discussions, motor vehicle evaporative
emissions can be divided into several different categories.
The first consists of emissions from properly-designed and
operated vehicles assumed to emit at the standard; therefore,
none of the control strategies (to be detailed in the last
section of this chapter) will reduce this portion. However,
the four components of current excess motor vehicle emissions
will be addressed in the remainder of this study.
The first — insufficient design of the purge system —
could be addressed via changes to the certification test
procedure. The excess RVP effect could theoretically be
reduced or eliminated through the reduction of in-use fuel RVP
and/or the revision of certification fuel specifications. The
effects of the remaining two sources of excess evaporative
losses (i.e., malmaintenance/defects and tampering) probably
-------
2-54
cannot be totally eliminated, but could be significantly
reduced if in-use RVP wete controlled to lower levels or an
effective inspection and maintenance program for evaporative
systems could be developed and implemented.
The extent to which each of these five sources contribute
to total motor vehicle evaporative losses — and to total
non-methane hydrocarbon (NMHC) inventories — will be explained
in Chapter 3. There, future total NMHC inventories will be
broken down into stationary source emissions (separated into
bulk storage. Stage I, and other) and motor vehicle losses
(divided into exhaust HC, refueling, and evaporative HC losses).
c. Development of a Motor Vehicle Evaporative Emissions
Model
Evaporative emissions are affected by many variables, such
as type of fuel metering system, RVP, ambient temperature,
percent full of tank, and tank size, among others. However,
the great majority of evaporative emissions data are measured
at the federal test procedure (FTP) conditions which are not
completely representative of all in-use conditions,
specifically temperature, RVP, and the fill level of the tank,
which vary considerably throughout the country. To better
represent in-use evaporative emissions, a model should take
into account the geographical-specific nature of evaporative
emissions. The purpose of this section is to present the motor
vehicle evaporative emissions model developed for this analysis
and compare the model to currently available emissions data and
evaporative models. The first section will describe the
diurnal evaporative emissions model and attempts to verify the
model, and the second section will describe the hot-soak model.
i. Diurnal Model
Uncontrolled diurnal emissions originate from the fuel
tank and are primarily a function of the properties of the fuel
in the gas tank, the diurnal temperature swing, and the amount
of fuel in the tank. Controlled diurnal emissions measured in
a SHED test are the hydrocarbons which pass through the
evaporative control system and are not trapped by the charcoal
canister. In addition to the factors affecting uncontrolled
diurnal emissions, controlled diurnal emissions are a function
of the capacity of the canister, the type of charcoal in the
canister, the amount of hydrocarbons loaded on the canister
before the beginning of the test, and other vehicle-related
parameters. Uncontrolled diurnal emissions can be predicted
readily; however, the evaporative control systems vary widely
from manufacturer to manufacturer, and also model to model and
do not lend themselves to being modeled accurately-
-------
2-55
The diurnal emissions model developed for this analysis
first predicts the uncontrolled diurnal emission potential at
specified in-use or test conditions analytically, and then uses
empirically-derived relationships between uncontrolled and
controlled diurnal emissions to predict the latter at the test
conditions of interest. (Appendix 2-A presents the derivation
of these relationships.) The key assumptions of the model are
that controlled emissions increase monotonically with
uncontrolled emissions, and that controlled emissions can be
related to an uncontrolled emissions potential. Based on the
characteristics of the evaporative control system (charcoal
properties and purge air effects), the assumptions would be
expected to be valid.
The uncontrolled diurnal emissions potential is based on
an equation developed by D.T. jWade in 1967.[32] The [equation
relies on changes in actual fuel vapor pressure, the Ideal Gas
Law, and the processes occurring in a vehicle's fuel tank to
predict uncontrolled diurnal losses from a fuel tank as a
function of fuel properties, temperature conditions, and
percent full of tank.
Wade's equation can be used to predict absolute
uncontrolled diurnal losses (i.e., grams of HC emitted from the
fuel tank) for any given set of conditions; however, its use
here was to predict relative losses and. not absolute values.
More specifically,Wade's equation was used here to calculate a
relative index of theoretical uncontrolled diurnal emissions
which could be related to measurements of controlled diurnal
emissions under various conditions to produce a diurnal
emissions model. The equation used for uncontrolled diurnal
losses is as follows:
G = Weight hydrocarbon lost, g
W = Fuel density, lb/gal
M = Molecular weight of hydrocarbon vapor, lb/lb mole
at average liquid temperature
p = Vapor pressure of gasoline, psia, at liquid temper-
temperature corresponding to T
P = Total pressure, psia
V = Volume of vapor space, cu ft
T = Temperature, R
G = 227 (W) (
( 520 ) T Pi + p 2
(690 - 4M) L(Pt - Pl) (Pt - pz)
where:
-------
2-56
Subscripts:
t = Tank
1 = Initial state
2 = Final state
The uncontrolled diurnal index (UDI) was calculated as the
ratio of Gtest (using various temperatures, RVPs, and percent
fill levels) to GCert (based on the 60-84°F diurnal test with
a 40 percent full tank and the RVP of the fuel for which the
vehicle was certified). Exhibit 2-1 is a listing of the
computer program which calculates the UDI for different in-use
and certification RVPs, temperatures, and fill levels.
The equation for fuel density is based on a linear
regression of fuel density versus RVP from MVMA fuel surveys.
The equation for molecular weight and the method of predicting
vapor pressures are based on information from the Wade report.
The size of the vapor space assumes an average gasoline tank
size of 16 gallons with a vapor space of two gallons when the
tank is "full." The size of the vapor space at the beginning
of the diurnal test is assumed to be equal to the size of the
vapor space at the end of the diurnal test. The tank is
assumed to be at atmospheric pressure. Atmospheric pressure is
assumed to be 14.7 psi unless the city has been formally
designated as high altitude (greater than 4,000 feet above sea
level), in which case the atmospheric pressure is assumed to be
12.5 ps i.
Table 2-16 contains the UDIs calculated at various in-use
or test conditions for varying certification fuel RVPs. The
UDIs are shown for comparison purposes only. The inputs used
in the diurnal model for MOBILE3 are based on city-specific
RVPs, temperatures, and the certification fuel of the control
scenario (see Tables 3-1, 3-2, and 3-5 for a list of the
city-specific inputs). The UDI is used by MOBILE3 to calculate
the diurnal emission factors as described in Appendix 2-A.
In an attempt to verify the evaporative emissions model
with other available diurnal emissions information, the model
was compared to a more sophisticated GM model of uncontrolled
diurnal emissions. Automotive Testing Laboratories (ATL)
diurnal emissions data, and some EPA emission data at varying
tank fill levels.
The method used by GM to calculate uncontrolled diurnal
emissions is similar to the EPA model.[33] However, a more
complex procedure was used to predict the gasoline vapor
pressure at different temperatures based on the chemical
composition of the gasoline. The GM model is compared to the
-------
B8000
FUNCTION
CAL1ND(RVP1.RVP2)
8B001
C
88002
C
CALIND
uses
passed in fuel RVP levels, minimum and maximum fuel tank
88003
C
temperatures
and fleet average percent of fuel tank filled to estimate
88004
C
the "uncontrolled diurnal index". Subsequently, the diurnal evaporative
88005
C
emission factor will be calculated as a function of this "index".
88006
C
88007
C
Ca11ed
by LOCAL
88008
C
88009
C
Input on cal
1 .
88010
C
8801 1
C
parameter
1 1 S t :
88012
C
RVP1
- RVP
to be evaluated
88013
C
RVP2
- RVP
to be normalized, e9., certification fuel RVP
88014
C
88015
C
common blocks.
8801 6
C
/CITCIN/ KREJN
B801 7
C
/CITPAR/ TEMMIN.TEMMAX
B8018
C
/RVPNAT/ PFUL
88019
C
B8020
c
Out put
on return:
BB02 1
c
68022
c
funct i on:
CALIND
88023
c
88024
c
Local array
subscripts :
88025
c
88026
c
T ( 2 )
-
T ( LT )
88027
c
VPPR(2)
VPPR ( LT )
88028
c
88029
c
Local v
ariable / array dictionary-
88030
c
88031
c
Name
Type
Desc r1pt ion
B8032
c
88033
c
ARPR
R
Atmospheric pressure, if KREJN = 1. then ARPR = 14.696
88034
c
PFTP
R
Percent of fuel tank filled under FTP (always 40%)
88035
c
DENSTV
R
Fuel density, a function of RVP
88036
c
VPSP
R
vapor Space in cubic ft, a function of PFUL
88037
c
T(LT)
R
Incremental fuel temperature in F
88038
c
TT
R
Maximum temperature
88039
c
VP100
R
Vapor pressure at 100F, a function of RVP
88040
c
C
R
Constant, used to calculate A100
88041
c
X
R
Coefficient used to calculate A100, a function of VP100
88042
c
A 1 00
R
Coefficient used to calculate A
88043
c
A
R
Coefficient used to calculate VPPR(I)
88044
c
WTM
R
Molecular weight, a function of RVP and fuel temp
88045
c
VPPR(I)
R
Vapor pressure at T (low/high temperatures)
88046
c
TP 1
R
First coefficient for uncontrolled diurnal equation
88047
c
TP2
R
Second coefficient for uncontrolled diurnal equation
88048
c
TP3
R
Third coefficient for uncontrolled diurnal equation
88049
c
G
R
Weight HC loss in grams under T(I), RVP1, and PFUL
88050
c
GFTP
R
Weight HC loss in grams under FTP (60-84F, 40% Full, RVP2)
88051
c
88052
c
Not es.
88053
c
88054
c
CALIND
is G/GFTP, each of which is a function of temperature increment
88055
c
(from TEMMIN
l to TEMMAX), RVP fuel and PFUL (percent tank filled)
88056
c
88057
c
CALIND
Mas added for In-House Version 09.
-------
88056 C
88059 C
88060 COMMON /CITCIN/ CIGA SI ,CIGAS2,CIETH1 ,CIETH2,ICERSW,KREJN
88061 COMMON /CITPAR/ CITNAM(4).TEMMIN,TEMMAX,FRETH.FRMETH.FRGAS
88062 COMMON /RVPNAT/ IUSESV.ICERSV.RVPCER,RVPLIM(2),PFUL
88063 C
88064 DIMENSION T(2).VPPR(2)
88065 C
88066 DATA PI/3. 14159/.PFTP/40.0 /
B8067 C
88068 ARPR=14 696
88069 IF(KREJN.NE 1) ARPR=12.5
88070 CINDEX=0.0
88071 GFTP=0.0
88072 C
88073 C First J pass computes denominator GFTP. Second pass repeatedly computes
88074 C numerator G, divides it Oy GFTP ana adds 1t to accumulator CINDEX.
B8075 C until T(2) exceeds TT.
88076 C
B8077 DO 50 J=1.2
B8078 RVP=RVP1
88079 IF(J.EQ.l) RVP=RVP2
B80B0 C
88081 C Calculate fuel density for given RVP.
88082 C
B8083 DENSTV=6 4-0 01977»RVP
88084 C
88085 C Calculate vapor space under given PFUL.
88086 C A,
B8087 IF(J.EQ.l) PF=PFTP 00
88088 IF(J EQ 2) PF = PFUL
88089 VPSP=2.4062-0 02139»PF
88090 C
88091 C Calculate vapor pressure at 100 (VP100) for given RVP.
88092 C
88093 VP100=1.0223»RVP
88094 • +(0.0357»RVP)/(1 0-0 0368»RVP)
88095 C
88096 C Calculate A100 according to VPI00
88097 C
88098 IF(VP 100 LT 14.18) GOTO 10
88099 C =80 861
88100 X=0 11»COS((4.O'VPl00-9 0)*PI/14 0)
88101 « +5.4•ALOG(VP 100)
88102 GOTO 20
88103 C
88104 10 C=66.561
88105 X=0. 12»C0S((VP 100-6.0 ) »PI/4 0)
88106 • -0.21»SIN(2.0*PI/7.5»(VP100-4 0))
88107 C
8B108 20 A 100 = C- 12 822»VP100
88109 » »1,3291»VP100»»2
88110 • -0.07991»VP100»*3
88111 • +1 .901 7E~03•VP 1 00**4-X
88112 C
88113 T(1)=TEMMIN
88114 IF(J EQ 1) T( 1 ) = 60 0
88115 TT=TEMMAX
-------
88116 IF(J.EQ 1) TT = 84.0
88117 T(2)=T(1 ) + 1 .0
B8UB C
88119 C Iteration starts hero.
88 120 C
88121 30 CONTINUE
88122 C
88123 C Calculate molecular weight
88124 C
88125 WTM=69 69-1.274»RVP
88126 • + 0.059»(T( I ) + T(2))/2.0
88127 C
88128 C Calculate vapor pressures
88129 C
88 130 DO 40 1 = 1 , 2
88131 A=A100+(100.0-T( I ) )
88132 « •((262.0/(A100/6.0+560 0))-0.01328)
88133 C
88134 VPPR(I)=14 696
88135 • ~0.53059*A
88136 » +7,6961E-03*A»»2
88137 • -5.4907E-05»A*»3
88138 ~ +1 7044E-07»A*«4
88139 40 CONTINUE
88140 C
88141 C Apply uncontrolled diurnal equation.
88142 C
88143 TP1=VPSP*118040 O'DENSTV/(690.0-4.0«WTM)
88144 TP2=VPPR(1)/(ARPR-VPPR(1))
88145 ~ +VPPR(2)/(ARPR-VPPR(2))
88146 TP3=(ARPR-VPPR(1))/(T(1)+460 0)
88147 • -(ARPR-VPPR(2))/(T(2)+460 0)
88148 G=TP1 • TP2 • TP3
88149 C
88150 C Next step depends on loop 50 counter J: 1 = calc GFTP, 2 = ca
88151 C
88152 IF(J EQ.l) GFTP=GFTP+G
88153 IF(J EQ 2) CINDEX=CINDEX+G/GFTP
88 154 C
88155 T( I )=T( 1 )+1 0
88156 T(2)=T(2 ) + 1 .0
88157 IF(T(2).LE.TT) GOTO 30
88158 50 CONTINUE
88 159 C
88160 CAL1ND=CINDEX
88161 C
88162 RETURN
88 163 END
NJ
I
ui
vo
CINOEX.
-------
2-60
Table 2-16
UDIs Under Various Test Conditions
Certification Test Tank T, T2 UDI
RVP (psi) RVP (psi) % Full (°F) (°F)
9.0 9.0 40 60 84 1.0000
9.5 1.1454
10.0 1.3094
10.5 1.4961
11.0 1.7104
11.5 1.9581
10.0 10.0 40 60 84 1.0000
10.5 1.1426
11.0 1.3062
11.5 1.4954
10.5 10.5 40 60 84 1.0000
11.0 1.1432
11.5 1.3088
11.5 11.5 40 60 84 1.0000
-------
2-61
EPA model by predicting a GM Uncontrolled Diurnal Index at
various temperature and RVP conditions. The GM model was run
at several in-use and test conditions and the GM Uncontrolled
Diurnal Index is compared to the UDI in Table 2-17. [34] The
percentage difference between the two indices at the different
conditions is from only -five percent to +three percent. This
comparison shows EPA's simpler method of predicting vapor
pressure, and subsequently the UDI, does not lead to errors of
more than five percent compared to the more complex GM model.
However, had the diurnal emission data been correlated versus
the GM diurnal emission potential instead of the UDI, the
effect on emissions would be eliminated.
GM conducted laboratory experiments to measure evaporative
vapor losses in an attempt to verify their model. The
percentage difference between the predicted and measured losses
were less than three percent. Most of the tests were conducted
with alcohol blends which are even more complicated to model
than straight gasolines.
The ATL emissions data were obtained as part of an
EPA-sponsored test program measuring diurnal and hot-soak
losses at various temperatures and RVPs.[23] The complete test
matrix consisted of the following:
Parameter Test Points
Gasoline RVP 9.0, 10.4, 11.8 psi
Diurnal Starting Temperature 60°, 68°, 75°F
Diurnal Temperature Change +15°, +20°, +24°F
Hot-Soak Temperature 70°, 82°, 95°F
Emissions data from three carbureted and three fuel-injected
vehicles which were available at the time of the analysis were
averaged for the comparison. Table 2-18 contains the average
diurnal emissions at the various diurnal temperatures and
RVPs. Table 2-19 lists the UDIs associated with the diurnal
test conditions.
The average diurnal emissions were plotted versus the UDI
and are shown in Figures 2-10 and 2-11 for the carbureted and
fuel-injected vehicles, respectively. The emissions at each
RVP at diurnal temperatures between 60-84°F are highlighted
with solid squares. Two conclusions can be drawn from these
figures; one, controlled diurnal emissions are a function of
UDI, and two, the three emission points tested at 60-84°F (FTP
temperature conditions) predict the trend in controlled
emissions as well as the entire data base for UDIs below 4.0,
the upper limit of the majority of in-use conditions. These
two assumptions are the basis for the diurnal emissions model
based on the three EF test data points.
-------
2-62
Table 2-17
Comparison of GM Uncontrolled Diurnal Index to UPI
Conditions
RVP
T,
t2
GM Uncontrolled
or City
(psi)
(°F)
(°F)
Diurnal Index
UDI*
FTP
9.0
60
84
1.000
(-)
1.000
FTP
10.4
60
84
1. 503
(+3%)* *
1.457
FTP
11.7
60
84
2 . 078
( + 1%)
2. 068
New York
11 . 5
72
92
2.696
(-3%)
2.783
Chicago
11.9
77
95
3 .516
(-5%)
3.714
Philadelphia
11.5
62
87
2 .304
(-2%)
2.344
Detroit
12 . 0
66
87
2 . 446
(-1%)
2.472
Boston
11. 5
73
97
4 . 046
(-5%)
4 . 248
Wash. , D.C.
11 . 5
62
85
1 . 991
(-1%)
2 . 016
Miami
11.3
81
96
2 . 784
(-3%)
2.865
Houston
10 . 5
76
93
1 . 958
(+1%)
1 . 945
Pittsburgh
12 . 1
64
87
2.655
(-1%)
2.672
Connecticut
11 . 5
72
90
2.266
(-2%)
2.321
* Based on
a 40%
full tank.
** Percentage difference
between
UDI and
GM Uncontrolled
Diurnal
Index.
-------
2-63
Table 2-18
Average ATL Diurnal Emissions[23]
Fuel Metering No. of RVP Starting Diurnal Emission (g/test)
System Vehicles (psi) Temp (°F) +15°F +20°F +24°F
CARB 3 9.0 60 0.71 1.05 1.45
68 0.96 1.57 2.70
75 1.82 3.54 6.54
10 . 4
60
68
75
23
72
79
2
4 .
9
10
29
08
3
9
80
05
16. 72
11.8
60
68
75
12
17
99
4 . 24
10 . 85
24 . 14
8. 14
19 .31
42.38
FI
9 . 0
60
68
75
0 . 45
0 .41
0 . 53
0.68
0.80
1 .45
1. 01
1. 61
3.69
10 . 4
60
68
75
61
02
70
1
3
7
33
32
00
2. 65
7. 59
16.35
11. 8
60
68
75
0,
2
7 ,
97
03
68
2.74
8 .09
21 . 79
6.21
17.99
41.26
-------
2-64
Table 2-19
UDI at ATL Diurnal Test Conditions*
RVP
Starting Temp.
UDI
(psi)
(°F)
+ 15°F
+20 °F
+24 °F
9.0
60
0 . 500
0 .753
1 . 000
68
0 . 706
1 . 072
1.438
75
0 . 973
1 . 495
2 . 032
10.4
60
0 . 715
1 . 085
1 . 457
68
1 . 034
1 . 595
2 . 175
75
1 . 475
2.325
3.250
00
60
1 . 012
1 . 559
2 . 125
68
1 . 521
2 . 403
3 .372
75
2 . 289
3.776
5.609
*
Based on a 40% full
tank.
-------
ij
Figure 2-10
ATL CARBURET? 0 DIURNAL EMISSION
,•5 UDI
UDI
-------
Figure 2-11
ATL FUEL-INJECTED DIURNAL EMISSIONS
vs -JDI
UDI
-------
2-67
EPA in Ann Arbor performed evaporative emissions tests on
vehicles with varying tank fill levels. Sixteen vehicles were
tested over the FTP on commercial fuel at tank fill levels of
40, 20, 60, and 80 percent, in that order. The vehicle
information and emissions are listed in Table 2-20. The
diurnal and hot-soak emissions are presented graphically in
Figures 2-12 and 2-13, respectively.
From the top half of Figure 2-12 for diurnal emissions,
there is no clear trend in emissions versus fill level.
However, as many test programs have shown, the order of testing
a vehicle while varying a parameter (as for different RVPs) has
a direct impact on the level of emissions. Testing the
vehicles at a fill level out of order (i.e., testing at 40
percent full, first, then 20, 60 and 80 percent) appears to
have had such an effect on diurnal emissions. When only the
20, 60, and 80 percent fill level diurnal emissions data is
examined, the expected trend of a decrease in diurnal emissions
with an increase in fill level can be seen -etewer-, as in the
bottom half of Figure 2-12. c\gaAy
There appears to be no effect of fill level on hot-soak
emissions as shown in Figure 2-13. No different trends exist
when examining carbureted and fuel-injected vehicles separately.
ii. Hot-Soak Model
Uncontrolled hot-soak emissions originate from the
carburetor or fuel-injection system, and the fuel tank to a
degree due to recirculation of heated gas from the fuel pump.
Uncontrolled hot-soak emissions are primarily a function of
RVP, distillation characteristics, carburetor size, engine
temperatures, and amount of recirculated gasoline. Unlike
uncontrolled diurnal emissions, which are relatively
straightforward to model, no reasonably accurate model was
found or could be developed for uncontrolled hot-soak
emissions. A similar approach to the diurnal model to
calculate an uncontrolled hot-soak emission potential, based on
the same ATL temperature data, engine temperature, and
distillation properties of the fuel, and correlate it to
controlled hot-soak emissions was attempted. Due to the lack
of information on distillation properties, gasoline
recirculation data, and reliable temperature effects data, the
attempt was not successful. Therefore in-use hot-soak
emissions are modeled as a function of in-use RVP,
certification RVP, and ASTM limit RVP. The adjusted RVP level
used in MOBILE3 to calculate in-use hot-soak emissions is as
follows:
Adjusted = (In-Use x 11.5 psi x 9.0 psi
Hot-Soak RVP) (ASTM (Certification
RVP limit) RVP)
-------
Table 2-20
Evaporative Emisions from Vehicles Tested at
Varying Tank Fill Levels
Fuel
Vehicle
Diurnal Emissions (g/test)
Hot-Soak Emissions (g/test)
System
40%
20%
60%
80%
40%
20%
60%
80%
1984
Mercury Marquis
FI
0.27
0.20
0.28
0.54
0.32
0.39
0.35
0.23
1982
Pontiac J2000
FI
4.87
3.69
0.71
0.28
0.50
0.45
0.43
0.46
1982
Renault Fuego
FI
1.65
1.62
0.92
0.48
1.44
1.66
1.19
1.08
1982
Cadillac DeVille
FI
25.78
22.54
15.62
9.62
1.34
0.88
0.91
1.01
1985
Buick Century
FI
3.45
4. 64
0.98
1.22
6.22
26.09
1.43
2.56
1983
Toyota Celica
FI
12.61
9.90
7.63
1.17
0.17
0.12
0.12
0.11
1982
Lincoln Mark VI
FI
15.29
7.75
11.19
9.93
1.09
0.74
0.94
0.43
1985
Chevrolet Cavalier
FI
1.29
5.28
0.90
0.49
0.33
1.06
0.51
0.42
1983
Ford Ranger
Carb
5.83
11.90
14.24
7.93
1.62
1.55
1.22
1.60
1982
Chrysler Diplomat
Carb
7.85
10.35
8.65
2.63
1.89
4.40
2.79
4.08
1982
Chevrolet Impala
Carb
22.67
38.13
16.69
9.55
5.66
5.05
5.19
5.31
1982
AMC Concord
Carb
19.89
12.68
13.62
6.50
1.10
1.30
1.76
1.21
1983
Chrysler Aries
Carb
7.44
12.13
5.73
2.19
1.01
1.22
1.24
1.09
1984
Chevy S-10
Carb
5.67
4.21
3.42
1.09
0.36
0.30
0.28
0.29
1985
Nissan Sentra
Carb
0.62
0.97
0.76
0.38
1.84
1.82
2.11
1.79
1985
Honda Civic
Carb
1.85
5.79
0.54
1.78
0.91
0.90
1.00
0.89
-------
2-69
Figure 2-12
o
UJ
X
«/>
(/>
2
O
Q
UJ
X
(/)
in
2
O
EVAPORATIVE DIURNAL EMISSIONS
AT VARYING TANK FILL LEVELS
40 60
PERCENT FILL
PERCENT FILL
VEHICLE:
&
84 MERC
X
82 PONT
~
82 RENA
a
82 CADI
B
85 BUIC
M
83 TOYO
~
82 LINC
©
85 CHEV
o
83 FORD
+
82 CHRY
0
82 CHEV
a
82 AMC
7
83 CHRY
a
84 CHEV
85 NISS
£k
85 HOND
VEHICLE:
&
84 MERC
X
82 j»0NT _
~
82 RENA
a
82 CADI
0
85 BUIC
A
83 TOYO
~
82 LINC
9
85CHEV
O
83F0RD_
+
82 CHRY
O
82 CHEV
a
82 AMC
7
83 CHRY
a
84 CHEV
85 NISS
A
85 HON0
-------
2-70
Figure 2-13
EVAPORATIVE HOT SOAK EMISSIONS
AT VARYING TANK FILL LEVELS
PERCENT FILL
VEHICLE:
a
84 MERC
X
62 PONT
~
82 RENA
a
82 CADI
B
85 BUIC
A
83 TOYO
~
82 LINC
®
85_CHEV
O
83FORD_
+
82 ChRY
0
82 CHEV
EB
82 AMC
7
83 CHRY
a
84 CHEV
D
85 NISS
A
85 HOND
-------
2-71
The ASTM limit term is an attempt to account for temperature
differences between ASTM classes. The assumption is low RVPs
and higher ambient temperatures typical of ASTM Classes A or B
areas produce the same emissions potential as the high RVPs and
lower ambient temperatures of ASTM Classes C, D or E areas.
The certification RVP term accounts for the effect of a
different certification RVP on emissions. The assumption is
that a change in certification fuel RVP to a higher RVP level,
would decrease controlled hot-soak emissions because a larger
canister would be required to meet a 2-gram standard.
Similar attempts to verify the hot-soak model were not
possible because of a lack of hot-soak models and reliable
emissions data showing the effect of the relevant parameters on
hot-soak emissions.
d. Fuel Weathering and Average In-Use Fuel Tank Level
As gasoline in vehicle fuel tank systems is heated by
diurnal temperature increases, many of the more volatile
hydrocarbons in the gasoline mixture are lost, thus decreasing
the volatility of the fuel. This phenomenon, known as
weathering, is an important consideration in assessing
evaporative emissions that actually occur in the field. This
section will fdescribe how the diurnal evaporative emissions
model was adjusted to account for the effects of fuel
weathering and in-use tank fill level.
During the summer of 1985, EPA conducted a study which
examined the effects of weathering on the RVP of gasoline as
the vehicle's fuel tank is emptied by use. [35] Three
carbureted test vehicles fueled with commercial fuel were
driven for three trips per day (one 12 mile trip and two 8 mile
trips) on Mondays through Thursdays, and were parked outside on
weekends. Results of this testing showed that fuel RVP can
drop by as much as 0.9 psi by the time a vehicles fuel tank
goes from full to 40 percent full. Test results also show that
the RVP of the fuel drops faster as the fuel tank level
decreases. This is to be expected, since diurnal evaporative
emissions are higher due to greater vapor space and since the
lesser amounts of liquid fuel are heated to higher temperatures
during vehicle operation, driving off more of the lighter
hydrocarbons.
Another factor, related to weathering, which must be
incorporated into estimations of evaporative emissions is the
variations in in-use fuel tank levels. Since, the amount of
"uncontrolled diurnal emissions is directly proportional to the
amount of vapor space in the fuel tank (Wade's equation for
uncontrolled diurnal emissions), a fuel tank that is 20 percent
full will produce greater emissions than the same size fuel
tank at 80 percent full, all other conditions being equal.
-------
2-72
However, since emissions also decrease with decreasing fuel RVP
the emissions of a 20 percent full tank in which the fuel has
weathered significantly may be less than those of an 80 percent
full tank with unweathered fuel.
Hence we see that in order to model vehicle emissions, we
need to know present tank level, the dispensed fuel RVP, and
the amount of weathering that has been experienced. Further
complicating matters is the problem of determining the amount
of RVP loss experienced during the refueling of the vehicle and
accounting for the mixing of low RVP fuel already in the tank
before refueling with the dispensed fuel. As both fuel usage
and weathering occur in concert their effects on emissions are
best dealt with analytically together.
In order to model evaporative emissions while
incorporating weathering and fuel tank level effects, ideally
one would conduct a survey of vehicles in the private sector
and measure tank RVP, fuel tank level, and local dispensed RVP
levels. From this distribution of vehicle and fuel volatility
situations one could develop a reasonably accurate model of
fleet emissions.
Since a survey of this nature was not available for use,
distributions of vehicle fuel tank situations based on four
refueling surveys and weathering estimates from the above
mentioned EPA weathering study were generated. Though this
limited amount of refueling data could be influenced by
uncontrolled parameters ( i.e., gas prices, credit card usage,
fuel station location, self-serve vs. full-serve orientation,
etc.), no other reliable refueling data was available for use
at this time.
Three of the four refueling surveys used in this analysis
were part of an efficiency evaluation survey of Stage II
refueling vapor recovery systems. Two of these surveys were
performed by Scott Environmental Technology, Inc.; the other by
Exxon Research and Engineering Co. From the three surveys,
the results of only those vehicles on which fuel tank size
information could be found were used. The survey performed by
Exxon was conducted in April of 1976 at a service station in
Linden, New Jersey.[36] Two hundred seventy six of the
refueling events (attendant serve) contained in this survey
were used, 181 of which (66 percent) were complete fillups.
The second survey, prepared for CARB by Scott Environmental
Technology Inc., was conducted in 1979 at a Union 76 Service
Station, out of which 81 attendant serve refueling events were
used, 74 of which (91 percent) were fillups. [37] The third
survey was performed by Scott Environmental Technology, Inc.,
for Cambridge Engineering, Inc. in 1981 at a World Oil Company
-------
2-73
service station in Los Angeles, from which 94 self serve
refueling events were used, 18 of which (19 percent) were
fillups.[38]
Each of these three surveys contained information on
vehicle type, amount of fuel dispensed, and whether or not each
vehicle was filled completely. These three surveys were
somewhat incomplete however, in that the fuel tank level of the
vehicles before refueling was not measured. The fourth survey,
conducted by GM, did have information on fuel tank, levels
before refueling, and was therefore used as the preferential
^database, as will be discussed below . /~By using the combined
data from the three incomplete surveys, and vehicle tank sizes
, from the Automotive News Market Data Book Issue, a distribution
of complete and partial fills was generated showing fill
fractions and the number (and fraction) of vehicles in each
particular situation, as presented in Table 2-21.
The contents of Table 2-21 show a "snapshot" distribution
of vehicle refueling events at the service stations. In order
to model emissions, however, one needs a "snapshot"
distribution of the refueling characteristics of all vehicles
in society. These are not the same. For instance, assuming
all vehicles consume the same fraction of their fuel tank
capacity each day, a vehicle which refuels from empty to 50
percent full will need to refuel twice as often as a vehicle
which refuels from empty to 100 percent full. Therefore, if an
equal number of 50 percent fillers and 100 percent fillers
existed in society, a service station refueling survey would
show twice as many 50 percent refueling events as 100 percent
refueling events. Similarly, one would observe four times as
many 25 percent refueling events. In order to account for the
higher frequency of appearance of those who refuel in small
amounts, the number of vehicles at each percent of tank filled
range was multiplied (see Table 2-21) by the average fuel tank
fill fraction to obtain a weighted number of vehicles in each
refueling situation. The results of this computation are given
in Table 2-22.
Since no measurements were taken of the vehicle fuel tank
levels before and after partial fills, it was necessary to
determine them analytically from the amount of fuel dispensed.
The weighted number of vehicles in each partial fill category
(see Table 2-22) were divided equally among likely combinations
of initial and final tank levels, under the following three
assumptions: 1) vehicle fill levels were divided to the nearest
10 percent, 2) no vehicles were 0 percent full before
refueling, and 3) none of the vehicles were greater than 40
percent full before refueling, since Table 2-22 shows that very
few weighted, in-use vehicles undergoing complete fillups were
greater than 40 percent full before refueling. The resulting
distribution for partial fills is shown in Table 2-23. For
-------
2-74
Table 2-21
Results of the Three Incomplete Refueling Surveys
Complete Fillups
% Full
Before Refueling
70
60
50
40
30
20
10
(65-75)
(55-65)
(45-55)
(35-45)
(25-35)
(15-25)
( 5-15)
Amount Dispensed
(°o of Tank)
30
40
50
60
70
80
90
(25-35)
(35-45)
(4 5-55)
(55-65)
(65-75)
(75-85)
(85-95)
S**
S***
Total
% of
_E*
Carb
CAM
tl Vehicles
Vehicles
2
3
0
5
1.11
18
6
1
25
5.54
28
6
1
35
7.76
32
19
0
51
11.31
31
14
6
51
11.31
41
17
5
63
13.97
29
_9
_5
43
9.53
181
74
18
273
60. 53
Partial Fills
Amount
Dispensed
s**
s***
Total
<% of Tank)
E*
Carb
Cam
If Vehicles
% of Vehicles
90
(85-95)
0
0
0
0
0.00
80
(75-85)
1
0
1
2
0.44
70
(65-75)
6
0
10
16
3.55
60
(55-65)
10
0
10
20
4.43
50
(45-55)
16
3
19
38
8.43
40
(35-45)
32
1
19
52
11.53
30
(25-35)
17
1
11
29
6.43
20
(15-25)
11
2
6
19
4.21
10
( 5-15)
2
_0
_0
2
0.44
95
7
76
178
39.46
* E =Exxon Survey
(ft Vehicles)
** S =Scott Envr. Tech.
Carb Survey (for CARB)
(ft Vehicles)
*** S =Scott Envr. Tech.
Cam Survey (for Cambridge)
(ft Vehicles)
-------
2-75
Table 2-22
Determination of a Distribution of In-Use
Refueling Patterns from Service Station Refueling Patterns
(Three Incomplete Surveys)
Complete Fillups
Full Before
Refueling
% of
70
60
50
40
30
20
10
(65-75)
(55-65)
(45-55)
(35-45)
(25-35)
(15-25)
(5-15)
tt Vehicles
(Srv. Station)
5
25
35
51
51
63
43
273
Vehicles
l
5
7
11
11
13
9
11
54
76
31
31
97
53
of Vehicles Vehicles
(Srv. Station) (In-use)
1
10
17
30
35
50
38
5
0
5
6
7
4
7
60 . 53
184 . 4
(In-use)
0 . 57
3.82
6.69
11 . 69
13 . 64
19 . 26
14 ¦ 79
70 . 46
Partial Fills
Amount Dispensed
(% of Tank) # Vehicles
% Vehicles
of Vehicles
(Srv. Station) (Srv. Station) (In-use)
% of
Vehicles
(In-use)
90
80
70
60
50
40
30
20
10
(85-95)
(75-85)
(65-75)
(55-65)
(45-55)
(35-45)
(25-35)
(15-25)
(5-15)
0
2
16
20
38
52
29
19
2
178
0.00
0 . 44
3 . 55
4.43
8.43
11. 53
6.43
4.21
0.44
39.46
0
1
11
12
19
20 .8
8.7
3.8
0.2
77.30
0 . 00
0 .61
4.28
4
7
7
3
1
0
59
26
95
32
45
08
29.54
-------
2-76
Table 2-23
Determination
of Initial and
Final Tank
Levels
for
Partial
Fills (Three Incomplete Surveys)
Partial Fills
Tank
Tank
Amount
Level After
Level Before
Dispensed
# of
Refueling
Refueling
tt of
(% of Tank)
Vehicles
(% Full)
(% Full)
Vehicles
(In-use)
(In-use)
80
l. 6
90
10
1.60
70
11.2
90
20
5.60
80
10
5.60
60
12.0
90
30
4 . 00
80
20
4 . 00
70
10
4.00
50
19 . 0
90
40
4 . 75
80
30
4 . 75
70
20
4 . 75
60
10
4.75
40
20.8
80
40
5.20
70
30
5.20
60
20
5.20
50
10
5.20
30
8.7
70
40
2. 18
60
30
2. 18
50
20
2. 18
40
10
2. 18
20
3.8
60
40
0.95
50
30
0.95
40
20
0.95
30
10
0.95
10
0.2
50
40
0.05
40
30
0 . 05
30
20
0.05
20
10
0.05
-------
2-77
instance, the 11.2 vehicles that were filled 70 percent of
capacity, were divided equally (5.6 vehicles each) between
vehicles being filled from 20 percent full initially to 90
percent full and from ten percent full initially to 80 percent
full. This process was repeated for each of the partial fill
amounts shown in Table 2-22.
In order to model fleet emissions, it was necessary to
make an estimate of the vehicle fuel consumption rate. All of
the vehicles from the surveys were normalized to a fuel tank
size of 16 gallons (the sales-weighted average of 1984
automobiles). Using the MOBILE3 Fuel Consumption Model
estimates for 1988, it was determined that light-duty gasoline
vehicles consume 1.24 gallons per day, or about eight percent
of a 16 gallon tank. For simplicity's sake, and since the
amount of weathering was determined in terms of tank percent
full instead of time anyway, we assumed that fuel consumption
occurred at roughly ten percent of tank per day.
Vehicles were assumed to have consumed this ten percent of
their tank before experiencing their diurnal each day, thus,
vehicle tanks were assumed to never have experienced diurnals
at their fullest level. Rather vehicles ranging from 100
percent full to 70 percent full, for instance, would experience
one-third of their diurnals at 90 percent, one-third at 80
percent, and one-third at 70 percent full, but none at 100
percent full. Calculations were done in this way because it
was assumed that all cars would experience some amount of
driving between refueling and the first diurnal. It also
partially offsets the previous assumption that no cars run down
below ten percent full prior to refueling. Using these
assumptions, a distribution of vehicle situations showing
percent full after refueling and present percent full was
generated and is shown in Table 2-24.
Recently, the results of a gas and service station survey
conducted in the Detroit area by General Motors Corporation
became available for use.[39] The survey contained information
on fuel tank levels both before and after refueling for 1,184
customers, 60.6 percent of which filled their vehicles
completely. The results of the survey were then categorized in
the same way as were the other three refueling surveys, and are
presented in Table 2-25. One advantage of the GM data over the
results of the other three surveys is the fact that the initial
tank levels were recorded for all partial fills, which
eliminates some of the uncertainty inherent in the other three
surveys' results. Since the GM survey did report the initial
tank levels and also because the data was much more recent than
the other surveys (1986 as opposed to late 1970's and early
1980*s), the GM survey results were evaulated independently of
the results of the other three surveys. Another reason why the
-------
T«bl* 2-24
Distribution of U*hlcl* Situations
R+fu+1
Fr»s*iit r*nk
P»rc»nt«g« of t'«hicl*3
it Ful 1 >
too
lOO
o.ooo
too
90
10.432
1O0
80
10.432
100
70
10.-432
100
60
lO. 24 1
100
50
9. 285
1O0
40
7 . 548
100
30
£ . 399
lOO
20
4.O50
lOO
io
J .»>43
JOO
o
O.OOO
90
90
O.OOO
90
80
J .ooo
90
70
] .000
90
60
1 . ooo
90
50
I .ooo
90
40
J - OOO
90
30
0.&37
90
20
C». 982
90
10
0.076
90
O
O.OOO
80
80
0.000
80
70
I . 420
80
60
1 . 420
80
50
1 . 420
80
40
1 . 420
80
30
O. 323
80
20
O . *560
80
IO
O. 506
80
O
ci. OOO
70
70
cf.oo'o"
70
60
1 . 592
70
50
1 . 392
70
40
1 . 592
7C»
30
1. 115
70
20
O. 18
70
10
O. 255
70
0
O.OOO
60
60
O.OOO ~
60
50
1. 519
60
40
3 . 519
60
30
3 . 137
60
20
O.I360
60
10
0. 363
60
O
O.OOO
50
50
O. OOO
50
40
0.-375
50
30
0.-356
50
20
O. r74
50
10
C . *497
50
0
O.OOO
40
40"
O.OOO
40
30
0.-478
40
20
O . *4S9
40
10
0..278
_4Q_ .
0
O.OOO
30
30
___ _ _ 'O.OOO
30
20
Ci. .20 1
30
10
O. 182
30
0
O . OOO
20
20
- - - o.ooo
20
10
0.019
20
0
O.OOO
10
10
o.ooo—
10
0
O.OOO
NJ
I
oo
-------
T*bl* 2-25
Results of Oft ft»fu»linq 5uri<»u
T*nk L»u»l
T*nk
NuHbtr of
P*rc»i
nt*g+ of
U*ight*d
f t»r
utl1ng
Btfor» R*fu»llng
f#hlcl~*
T *r>k
Filled
Nur«b#r of
[Percent Full 3
CP»rc»nt Full J
in Surutg
Cl ft
100
<">95)
0
<<5)
66
100
68.0
100
<>9S)
10
<5-14>
239
90
215. 1
100
<>9S)
20
<.15-24)
68
80
54.4
100
<>95)
30
<2S-34)
199
70
139.3
100
< >95)
AO
<35-44)
49
60
29.4
lOO
<>95)
SO
<45-54)
68
50
34.0
100
<>95)
60
<55-64)
7
40
2.8
100
<>95)
70
<65-74)
8
30
2.4
100
<>95)
80
<75-84)
12
20
2.4
lOO
<>3S)
90
C85-94)
2
10
0.2
100
<>9S)
lOO
<>9 4)
0
0
O.O
90
<"86-95,>
0
— —
6
90
5 • 4
90
*86-95)
10
<5-14)
28
80
22.4
90
*86-95)
20
<15-24)
7
70
4.9
90
c86-95">
30
<25-34**
21
60
12.6
90
<86-95)
AO
<35-44)
2
50
1.0
90
<.86-95)
50
<45—54)
3
40
1.2
90
C86-9S)
60
<55-64)
0
30
O.O
90
<86-95)
70
<65-74)
0
20
0.0
90
*86-9S)
80
<75-84)
0
10
0.0
90
~gtr
< 86-95)
<'76~-*85)
90
0
<"85-94)
<.<5)
0
4
0
80
0.0
3.2
80
C76-8S)
10
<5-14)
22
70
15.4
eo
<76-85'>
20
<15-24)
5
60
3 . 0
eo
<"76-85)
30
<25-34)
8
50
4.0
90
< 76-05)
<40
<35-44)
3
40
1 . 2
80
C76-051
50
<45-54)
1
30
0 . 3
80
<76-85)
60
<55-64)
1
20
0.2
80
<76-95"'
70
<65-74)
O
10
0 . 0
80
<76-95)
80
<75-84)
0
0
O.O
70 <66—75)
O
«S>
11
70
7 . 7
70
< 66-75>
10
<5-14)
28
60
16.8
70
<66-75)
20
<15-24)
7
50
3.5
70
<66-75)
30
<25-34)
26
40
10.4
70
<.66-75)
AO
<35-44)
3
30
0.9
70
<66-75)
SO
<45—54">
1
20
0.2
70
< 66-75)
GO
<55-64)
O
10
0.0
70
<66-7S)
70
<65—74)
O
o
O.O
*6"0
<.S6-6S)
0
~«S>
7
- - —— •
60
4.2
60
<56-65)
10
<5-14)
17
50
8.5
60
<56-6S^>
20
<15-24)
3
40
1.2
60
<56-65)
30
<25-34)
io
30
3 . O
60
<56-65)
AO
<35-44)
1
20
0.2
60
<56-65)
SO
<45-54)
0
10
0.0
60
~50"
11
SO- ~
5.5
50
<46-55)
10
<5-14)
49
40
19.6
50
<46-55)
20
<15-24)
11
30
3.3
50
< 46-55)
30
<25-34)
18
20
3 . 6
50
<46-S5->
AO
<35-44)
0
10
0.0
?o
<46-55)
50
<45-54)
0
0
0.0
40
C36—15^
o"
<<5>
15
40
6.0
40
<36-45)
10
<5-14)
32
30
9*6
AO
<36-45)
20
<1S-24)
5
20
1.0
AO
<36-45)
30
<25-34)
11
10
1.1
AO
<36-4S)
AO
<35-44)
0
0
0.0
90
<26-35)
0
«5>
11
30
3 3
30
<26-35)
10
<5-14>
21
20
4.2
30
<.26-35">
20
O
18
--
0
20
- -0.0.
3.6
20
<16-25)
10
<5—14>
24
10
2 • 4
2Q C16*25)
20
<15-24'*
1
0
0.0
0.8'
10
< 6-15>
0
«5)
8
' 10
10
<6-15)
10
0
<<5> " * '
1
o
o\o
-------
2-80
results of this survey were not evaluated together with the
results of the other three surveys was because a number of the
vehicles reported a completely empty tank before refueling in
the GM survey, whereas the lowest tank level assumed in the
three incomplete surveys was ten percent full.
For these reasons, two additional distributions of vehicle
situations (as in Table 2-24) were then generated from the GM
refueling survey results. One distribution was generated
assuming that ten percent of the fuel in the tank was consumed
before the first diurnal (a "worst-case" scenario), while the
other was generated assuming that ten percent of the fuel in
the tank was consumed after the first diurnal (a "best case"
scenario). Both distributions are shown in Table 2-26.
In order to account for the effect of weathering in the
model, weathering data was taken from the results of the study
performed by EPA during the summer of 1985 in which three
carbureted vehicles underwent two weeks of outdoor testing.
The three vehicles were initially filled with commercial grade
fuel after which the RVP of the fuel in the tank was measured.
The vehicles were then driven three trips per day (one 12 mile
trip in the morning and two 8 mile trips in the afternoon) on
Mondays through Thursdays, and parked outside Fridays through
Sundays. Figure 2-14 shows the RVP drop of the three test
vehicles as a function of fuel tank level.
Although weathering studies which showed results similar
to our own were available from General Motors Corporation and
Southwest Research Institute, they were not used in this
analysis.[40,41] The General Motors study was not used
primarily because of the unrepresentatively high fuel tank
temperatures (110-120°F) which occurred during testing (the
vehicles were operated on a dynamometer with relatively short
time intervals between driving cycles of about one hour), which
would tend to increase the amount of weathering experienced.
The study by Southwest Research was not used because the
vehicles tested underwent unusually long cycles (50 mile trips
daily) and because the vehicles were parked in a controlled
environment (76° F) on days when they were not driven.
To estimate weathering in our model, we used the average
RVP drop of the three EPA test vehicles between two given tank
levels using Figure 2-14. These values are shown for the EPA
test results, as well as for the GM and SWRI results in Table
2-27. For instance, the amount of weathering between 50
percent full and 40 percent full is 0.17 psi. The weathering
was always assumed to be the same between given fuel tank
levels, regardless of the initial tank RVP or the initial fill
level. This assumption is based on the results of the Vehicle
Tank Fuel Weathering test program by General Motors
-------
*f t.
: Fu
100
100
100
100
3O0
3 OO
lOO
100
100
lOO
JOO
¦90
SO
90
90
90
90
90
90
90
90
80
80
80
eo
eo
eo
80
eo
80
TO
to
TO
TO
TO
TO
TO
TO
6o
60
60
SO
60
60
60
SO
50
50
SO
SO
SO
40
10
40
40
40
30
30
30
30
20
20
20
10
10
Tabl* 2-26
Distribution of L^hicl* Situations
Pr#i#nt T*r»|. Leuel P*rc*nt*9* of U*hicl*f
CP#rc*nt Full 'j COM Refueling Suru»g--1D^ of Tinl'
Cemumd Br-foi-* 1st Diurnjl.'
100
O.OOO
90
9-683
80
9.656
70
9.494
60
9.387
£0
9.293
40
8.378
30
7-719
20
5-043
10
4.129
O
0.914
90
O.OOO
60
0.9O1
70
0.90 1
60
O . 90 1
SO
O .90 1
40
0.861
30
0.834
20
0.55 1
10
0.4S7
0
0.081
80
O.OOO
70
0.S32
60
0.592
50
0.578
40
0 .565
30
0 .524
20
0.417
10
0.350
0
0.O54
70
O.OOO
60
1*022
50
1.022
40
1.009
30
0.968
20
0.619
10
0.524
0
0. 148
60
O.OOO
50
0.S11
40
0.S11
30
0.498
20
0.363
io
0.323
o
0.O94
50
O.OOO
40
1 . 197
30
1. 197
20
0.955
10
0.807
o
0. 148
40
O.OOO
30
0.847
20
0.699
10
0.632
0
0.202
30
0.000
20
0.457
10
0.430
0
0. 148
20
0.000
10
0.565
0
0.242
10
O.OOO
0
0. 108
Ftrcvrit'ig* of U*hicl«s
fOH R«fu»lin9 Suru*y—No Fuel
fcnsuN»d Before 1st Diurnal)
"nr—= —
9.6S6
9.494
9. 367
9. 293
8. 378
7. 7 19
5.043
4. 129
0.914
O.OOO
0. 90 1
0.90 1
0.901
0.901
0.861
0.834
O.SSl
0.457
0.061
O.OOO
0.592
0.S32
0.576
0.565
0.524
0.417
0.3SO
0.054
O.OOO
1.022
1.022
1.009
0.966
0.6 19
0.S24
0. 148
O.OOO
0.511
O.S 11
0.498
0.363
0.323
0.094
0.000
1. 197
1. 197
0.955
0.607
0. 146
0.000
0.847
0.699
0.632
0.202
0.000
0.457
0.430
0. 148
O.OOO
0.565
0.242
0.000
o. loe
o.ooo
-------
FIGURE 2-14 KVP DROP as a function of TANK LEVEL (EPA WEATHERING DATA)
s~\
H
Ti
G.
(j
*
Q RANADA
Fuel Tonic Level (J4 Full)
+ RELIANT
SKTIARK
-------
2-83
Table 2-27
RVP Drop as a Function of Tank Level
Initial Fill
Final Fill
EPA
GM
SWRI
Level
Level
Data
Data
Data
(% Full)
(% Full)
RVP Drop
RVP Drop
RVP Drop
100
90
. 07
.31
. 13
90
80
. 09
. 19
. 13
80
70
. 10
. 16
. 04
70
60
. 13
. 17
. 05
60
50
. 14
. 18
. 07
50
40
. 17
.28
. 06
40
30
. 22
.34
. 08
30
20
.30
.33
. 10
20
10
. 44
.29
. 17
10
0
. 50
-------
2-84
Corporation.[39] This study observed the amount of weathering
experienced by four fuel injected vehicles driven approximately
30 miles per day in the Environmental Test Chamber at the
General Motors Proving Ground. Two test fuels were used one
with RVP of 11.8 psi and the other with RVP of 10.4. It was
found that the average amount of weathering experienced from
the as-dispensed level to the 40 percent level was 1.08 psi RVP
for the 11.8 RVP fuel and 1.41 psi for the 10.4 RVP fuel.
Although an opposite relationship would be expected (i.e., the
more volatile fuel should weather more), the results for these
two fuels seem to support the simplifying assumption with
respect to weathering and fill tank level used in our model.
Once the distribution of vehicle situations and the
amounts of weathering were determined, uncontrolled diurnal
emissions were calculated for various cities using the
uncontrolled diurnal emissions equation presented earlier in
this chapter. Calculations were performed using 30-year
average high and low July temperatures which were taken from
the "Statistical Abstract of the United States, National Data
Book and Guide to Sources," and using ozone non-attainment
design value day temperatures.[42 ] Dispensed fuel RVP was
assumed to be at the ASTM limit for each city. All
calculations were done assuming a 16 gallon tank with 2 gallon
head space, and assuming that the vapor space volume remained
constant during each diurnal.
Next, to estimate controlled emissions, the /Tjpi was
calculated for each vehicle situation. As explained earlier,
this [is defined as the ratio of uncontrolled emissions at the
^iven conditions to the uncontrolled emissions for the same
}size fuel tank at certification test conditions (i.e., 40
percent full tank, 9.0 RVP fuel, 60 °F to 84 °F temperature
rise). The UDI has been correlated with non-tampered
controlled emissions for a number of vehicles, yielding the
following equations:
1) Controlled Diurnal Emissions = 1.8283 x (UDI)Z
(fuel-injected vehicles)
2) Controlled Diurnal Emissions = 2.2777 x (UDI)2
(carbureted vehicles)
These equations were determined by curve-fitting the EF
test data versus the UDI. This is a simpler form of the
equations derived in Appendix 2-A, however, this should have
^minimal effect on pthe adjusted fuel tank level which is
back-calculated from these equations as explained in the
,following paragraphs.
-------
2-85
Assuming that about about 88 percent of all vehicles are
fuel-injected, while 12 percent are carbureted*, these
equations can be combined to give:
3) Controlled Diurnal Emissions = 1.8822 x (UDI)J
(all vehicles)
Sample calculations of controlled diurnal emissions for
each vehicle situation (as per the three incomplete refueling
surveys) using the ASTM fuel RVP and average July temperatures
for the city of Detroit are shown in Table 2-28. Average fleet
controlled emissions were calculated by multiplying the
controlled diurnal emissions for each vehicle situation with
the appropriate weighting factors (from Tables 2-24 and 2-26).
\ The average controlled emission value was then used to
1 back calculate average uncontrolled emissions using the UDI /
correlation (eqn. 3). The uncontrolled diurnal emission
equation was then used along with city specific average July
;temperatures (or ozone design value day temperatures) and
dispensed fuel RVP, to give an effective average vapor space
volume, which was converted to a fuel tank level. This
effective average fuel tank level can be used to incorporate
the effects of weathering and variations in fuel tank levels
into calculations of average fleet emissions for each city
using city specific temperatures and dispensed fuel RVP.
Results of the calculations of the effective fuel tank level
for nine of the ten largest non-California urban areas** are
shown in Table 2-29.
Instead of finding a large variance between cities in this
representative fuel tank level, it was found that a fuel tank
level of approximately 60 percent full was fairly well
representative for each city modeled. The representative fuel
tank levels for each city were then population weighted and
averaged. Using ozone design value day temperatures and the GM
refueling survey data, it appears that the representative fuel
tank level lies somewhere between 61.01 and 64.70 percent full
The accuracy of this assumption is not critical, since the
results of the model are determined by taking the ratio of
controlled emissions at different situations. Hence the
value of the constant in eqn. 3 is not important, since it
drops out of the calculations.
Calculations were not performed for Miami since ozone
design temperatures for this city were taken from a winter
day with a small diurnal temperature rise. All other
ozone design temperatures occurred during the summer
months.
-------
r*bl* 2-29
Diurn«l £hisfIons
cC*lcul»tlon> Pgr P>
L #c*l
Pr#i»nt
Control1«d
Ueighting Factor
F»fu*l
r«r»k L*ia*l
Prti»nt
E*isslons
<3 Incoriplftt
Full D
CX Full}
RUP
UDI
Suru»gf Contained
lOO
100
11.50
0.299
0. 17
0.oooo
100
90
11.43
0.526
0.52
0.1043
100
80
11.34
0.741
1.03
0. 1043
lOO
70
11.24
O.940
1.66
0.J043
1D0
60
11.11
1 . 118
2. 35
0. 1024
lOO
50
10.97
1.277
3.07
0.0929
100
40
10.80
1.410
3. 74
0.0795
lOO
30
10.56
1.505
4.26
0.0600
lOO
20
10.26
1.S47
4.50
0.0405
lOO
10
9.84
1.511
4. 29
0.0164
lOO
0
9.34
1.438
3.89
0.0000
90
90
11.50
0.S37
0.54
0.OOOO
90
80
11.41
0.756
1.08
O.OIOO
90
70
11.31
0.960
1. 73
O.OIOO
90
60
11. 16
1. 141
2.45
O.OIOO
90
50
11.04
1.304
3.20
O.OIOO
90
40
1O.07
1.439
3.90
O.OIOO
90
30
10.65
1.S36
4. 44
0.0064
90
20
10.35
1.578
4.69
0.0038
90 -
10
9.91
1.541
4. 47
O.OOQ8
90
0
9.41
1.467
4.05
0.OOOO
eo
80
11.50
0.776
1. 13
0.G000
60
70
11.40
0.966
1.83
0.0142
eo
60
11.27
1. 172
2.58
0.0142
eo
50
11. 13
1 .339
3.37
0.O142
eo
40
10.96
1.477
4.11
O.O142
eo
30
10.74
1.576
4.68
0.0092
eo
20
10.44
1.620
4. 94
0.OOS6
eo
10
10.00
1.581
4.71
0.O031
eo
0
9.50
1.505
4. 26
0.OOOO
70
70
11.50
l.OlS
1.94
0.OOOO
70
60
11.37
1.207
2. 74
0.O139
70
50
11.23
1. 379
3. 58
0.O139
70
40
11.06
1.521
4. 36
0.0139
70
30
10.84
1.623
4. 96
0.0112
70
20
10.54
1.668
5.23
0.0062
70
10
10. 10
1.627
4.99
0.0026
70
0
9.60
1.548
4.51
0.OOOO
eo
60
11.50
1.254
2.96
0.0000
60
50
11.36
1.433
3.86
0.0132
60
40
11. 19
1.581
4. 70
O.0132
60
30
10.97
1.686
5. 35
0.O114
60
20
10.67
1.732
5.65
0.0086
60
10
10.23
1.689
5.37
0.0036
60
O
9.73
1.607
4. 86
O.OOOO
£0
SO
11 .SO
1 .493
4. 20
O.OOOO
50
40
11.33
1.647
5.11
0.0098
SO
30
11. 11
1.757
5.81
0.0096
SO
20
10.81
1.804
6. 13
0.0077
50
io
10.37
1.7S9
5.82
0.OOSO
50
0
9.87
1.672
5.26
O.OOOO
40
40
11. SO
1.732
5. 64
O.OOOO
40
30
11.26
1.847
6.42
0.O048
40
20
10.98
1.896
6. 77
0.0046
40
10
10.54
1 .048
6.43
0.0028
40
0
10.04
1.756
5.80
0.OOOO
SO
30
11.50
1.971
7.31
O.OOOO
30
20
11.20
2.023
7.70
0.0020
30
10
10.76
1.970
7.31
o.ooie
30
O
10.26
1.870
6.58
O.OOOO
20
20
1 l.SO
2.210
9. 19
0.OOOO
20
10
11.06
2. 1S1
8.71
0.0002
20
0
10.56
2.040
7.83
0.0000 —-
10
10
11.SO
2.448
11.28
O.OOOO
1 0
0
11.00
2.319
10. 13
O.OOOO
U*ight+d
C n ission?
U*lght*d
Fl*»t Enifilont ¦
o.ooo
O. 054
O. 109
O. 174
0.241
0. 285
0.297
0. 2S6
0. ie2
0.071
0. 000
O.OOO
0.011
O. 0 17
0.025
0.032
O.039
O.028
0.018
0.003
0.000
0.000
0.026
0. 037
0.048
0.056
0.043
0.029
0. 0 14
0.000
O.OOO
0.038
0.050
0.061
0.055
0.032
0.013
0.000
O.OOO
O.051
0.062
0.061
0.04-3
0.020
0.000
0.000
0.050
O.056
0.047
0.029
O.OOO
O.OOO
0.031
O • 031
0.018
0.000
O.OOO
0.015
0.013
0.000
O.OOO
0.002
O.OOO
O.OOO
O.OOO
2.877
to
1
00
-------
2-87
Table 2-29
Representative Fuel Tank Level
Cor Emissions Modeling
A. Average July Temperatures
Representative Tank Level
% Full
GM Survey
(10% Consumed
GM Survey
(10% Consumed
ASTM
Low
High
3 Incomplete
Before
After
City
RVP
Temp.(°F)
Temp.(°
F)
Surveys
1st Diurnal)
1st Diurnal)
New York
11.5
68.0
85.2
61.12
59.12
63.35
Chicago
11.5
60.7
83.1
60.73
58.65
63.03
Phil.
11.5
66.7
86.8
61.44
59.49
63.62
Detroit
11.5
63.4
83.1
60.74
58.65
63.03
Boston
11.5
65.1
81.4
60.49
58. 36
62.82
Wash. DC
11.5
69. 1
88.2
61.76
59.88
63.90
Houston
10.0
72 . 8
93.8
61.99
60.17
64.09
Pitts.
11.5
61.3
82.5
60.64
58. 53
62.95
Gr. CT
11.5
61.2
84.1
60.91
58.86
63.17
Population
Weighted Avg.
61.06
59.05
63.31
Avg.
: 61.18
B. Ozone
Design
Value Temperatures
Representative Tank
Level
% Full
GM Survey
GM Survey
(10% Consumed
(10% Consumed
ASTM
Low
High
3 Incomplete
Before
After
City
RVP
Temp.(0 F)
Temp.(°
F)
Surveys
1st Diurnal)
1st Diurnal)
New York
11.5
72
92
62.86
61.19
64.83
Chicago
11.5
77
95
64.05
62.59
65.83
Phil.
11.5
62
87
61. 50
59.56
63.67
Detroit
11.5
66
87
61.48
59.55
63.66
Boston
11.5
73
97
65.21
63.93
66.81
Wash. DC
11.5
62
85
61.07
59.06
63.31
Houston
10.0
76
93
61.85
60.01
63.98
Pitts.
11.5
64
87
61.49
59.55
63.66
Gt. CT
11.5
72
90
62.23
60.45
64. 30
Population Weighted Avg.
62.71
61.01
64.70
Avg. = 62.86
-------
2-88
(depending on if the initial ten percent of the fuel in the
tank is consumed before or after the first diurnal). Using the
three incomplete surveys gave a value of 62.71 percent full.
It should be stressed that these three values^obtained from the
different refueling surveys lie quite closely\together. Since
the GM survey data was most complete a final value of 62.86
percent full was determined, the average of the two extremes.
Similarly, using average July temperatures, an average value of
61.18 percent full was determined. By using these fuel tank
levels with city specific dispensed fuel RVP and temperatures,
one can make a reasonable adjustment for the effects of
weathering and in-use refueling patterns in estimations of
fleet emissions, assuming that the results of the refueling
surveys are representative.
Because of the scarcity of data on these subjects, and
discrepancies between existing data sets, the results of this
model are somewhat uncertain. For example, the different
refueling surveys used in this study are in some disagreement
over the fraction of vehicles completely refueled. The Exxon
and GM surveys show between 60 and 70 percent complete fills,
while the two surveys by Scott Environmental Technology, Inc.
show 91 percent and 16 percent complete fills, respectively.
Since it is difficult to say which survey best represents
in-use refueling patterns, and since there is a great deal of
discrepancy between the two Scott surveys, it was beneficial to
evaluate the effect that each of the extremes had on the
results of our model. Therefore, two additional distributions
of vehicle situations were generated, one based on the survey
with the lowest fraction (16 percent) of fillups, and one based
on the survey with the highest fraction (91 percent) of fillups.
Calculations were performed which examined the sensitivity
of the model's results to these refueling survey differences.
Using Detroit temperatures and fuel RVP, the representative
fuel tank level calculated ranged from 54.8 percent full for
the survey with 16 percent fillups to 65.1 percent full for the
survey with 91 percent fillups. Thus even these vastly
different survey results affect the representative fuel tank
level by only about ten percent.
Another possible source of error (in the evaluation of the
three incomplete surveys only) is the assumption that the fuel
tank level before partial fills only ranges between 10 percent
and 40 percent full. It may be possible that a number of fuel
tanks range from 50 percent full to 80 percent full before
partial fills. On the other hand, it may be possible that
those who don't fill their tanks completely always let their
tanks run down to ten percent full before they refuel.
Although neither of these scenarios is likely the actual case,
the actual case should lie somewhere between them. Therefore,
two more vehicle distributions were generated using the sum of
-------
2-89
all three incomplete surveys, one based on the assumption that
it is equally likely for a partial fill to start at anywhere
from ten percent full to 80 percent full, and one assuming that
all partial fills begin with a ten percent full tank. Once
again the results of the model using these two vehicle
distributions for the city of Detroit showed little difference
(less than seven percent) in the fuel tank level determined.
Other factors which may contribute to the uncertainty of
the results of the model include the uncertainty over how much
fuel, on the average, is consumed before the first diurnal is
experienced. The scarcity of both on-road weathering data and
of evaporative emissions data as a function of various
temperatures and fill levels also adds uncertainty to the
results. As more testing is performed in these areas, however,
these uncertainties may disappear.
In summary, it seems reasonably accurate to state that the
effect of weathering and the variability in in-use fuel tank
levels can be incorporated into average fleet evaporative
emissions modeling by using city specific fuel RVP and
temperatures and assuming a fuel tank level of 62.86 percent
full with ozone design value day temperatures and 61.18 percent
full with average July temperatures. Variance of this
representative tank level from city to city was negligible.
Since our diurnal test procedure is intended to be 'worst-case'
in nature, a fuel tank level of 40 percent full (which would
give higher emissions than a 60 percent full tank) is used in
certifying new vehicles. A more detailed discussion into how
the effects of weathering, city-specific temperatures, and
in-use fuel tank levels relate to the FTP is reserved for
Appendix 6-A.
e. RVP and Temperature Variability
The evaporative emissions model used for this analysis to
predict in-use emissions is partly a function of RVP and
temperature. The values chosen for RVP and temperature inputs
are important and need to reflect the conditions for which the
model is being used. Design value day temperatures and average
RVP are used for environmental analyses (i.e., inventories,
future ozone-compliance), whereas average July temperatures and
average July RVP are used for economic analyses (i.e.,
evaporative recovery credit). The basis for these assumptions
are explained in greater detail in Section II of Chapter 3.
The use of average in-use RVPs and average July
temperatures as inputs to the evaporative emissions model tends
to underestimate average evaporative emissions due to the
non-linearity of evaporative emissions with RVP and
-------
2-90
temperature. For example, the average of the evaporative
emissions (sum of one hot-soak, and one diurnal) from a vehicle
tested over the FTP on a 9.0 psi and an 11.8 psi fuel is
greater than the evaporative emissions on a 10.4 psi fuel. The
following section will describe how the evaporative emissions
model is adjusted through the adjustment of the fuel tank level
to account for in-use RVP variability in the case of the
environmental analyses, and both in-use RVP variability and
daily temperature variability in the case of the economic
analyses.
i. RVP Variability
The city-specific RVP used in the evaporative emissions
model is 1) the average RVP of the non-alcohol containing
unleaded fuels based on the Summer 1985 MVMA National Gasoline
Survey for the specific city or nearest available city or 2)
the ASTM limit for that city, whichever is higher. However, as
the survey results show, RVP variability between service
stations can be substantial. A typical range between measured
low and high RVPs is 1.5 psi based on the MVMA survey.
To account for this in-use RVP variability effect on
in-use evaporative emissions, the evaporative emission model
was run for both the environmental and economic analyses for
eight cities. Only the diurnal model was run since hot-soak
emissions are more a linear function of RVP than diurnal
emissions and thus are not affected to as great an extent by
RVP variability.
The first run, used to account for the RVP variability
effect in the environmental analyses, was based on the ozone
design value day temperatures and corresponding fuel tank level
of 62.86 percent as determined previously in the section on
fuel weathering and fuel tank level variability. UDIs were
calculated at each survey RVP, and the non-tampered emissions
were projected assuming the 2010 fuel-injected/carbureted
vehicle mix of 0.888/0.112. Table 2-30 presents the analysis
for Dallas with ozone design value day temperatures. From the
average of the non-tampered diurnal emissions, the UDI which
yields this average diurnal emission value was determined.
Using the average RVP and ozone design value day temperatures,
and the back-calculated UDI from the average emissions, the RVP
variability adjusted fuel tank level was determined as shown in
Table 2-30. [in other words, an adjusted in-use fuel tank level
was calculated which, in combination with the ozone design
value day temperatures and average RVP, yields the true
emissions average for environmental analyses.
The second run, used to account for the RVP variability in
the effect in the economic analyses, was based on the 30-year
-------
n ^
\0 a \C'r N
d 'c-v
w; \p^
2-91
Table 2-30
RVP Variability Effect on In-Use Diurnal
Emissions in Dallas
(Ozone Design Value Day Temperatures)
Non-Tampered
n Diurnal*
RVP
Emissions
(psi)
UDI**
(a/test)
10.1
1.6116
5.073
10.2
1 . 6646
5. 467
„10". 8
2. 0311
8. 689
10.4
1.7769
6.363
J..0r6
1.8987
7.426
"10. 1
1.6116
5. 073
10.0
1.5607
4 . 711
J*.-8
1.4641
4 . 070
JL0 . 8
2.0311
8.689
J.0:0
1.5607
4 . 711
10. 4
1.7769
6.363
10. 1
1 .6116
5.073
Average
emissions =
5.976 g/test
UDI
which yields this
average emission
result is 1.7296
RVP Adjusted V = 1.7296
8.4 gallon 1.7605***
RVP Adjusted V = 8.253 gallons
RVP Adjusted = 16 - 8.253 + 2 X 100 = 60.9%
Fuel Tank Level 16
* Weighted for 2010 vehicle mix, 88.8 percent FI, 11.2
percent carbureted.
** Calculated at 62.86 percent full tank, T, = 76°F, T2 =
96°F.
*** UDI at average survey RVP, and ozone design value day
temperatures.
-------
2-92
average July minimum and maximum temperatures and corresponding
fuel tank level of 61.18 percent also determined in the
previous section on fuel weathering and fuel tank level
variability. Again, UDIs were calculated at each survey RVP,
and the non-tampered emissions were projected assuming the 2010
fuel-injected/carbureted vehicle mix of 0.888/0.112. From the
average of the non-tampered diurnal emissions, the UDI which
yields this average diurnal emission value was determined.
Using the average RVP and 30-year average July temperatures,
and the back-calculated UDI from the average emissions, the RVP
variability adjusted fuel tank level was determined. The July
average day adjusted fuel tank level still needs to be adjusted
for temperature variability below.
Table 2-31 contains a summary of the RVP variability fuel
tank level calculations for all eight cities for both the
environmental analyses (based on ozone design value day
temperatures) and the economic analyses (based on 30-year
average July temperatures). The ozone design value day
adjusted fuel tank level of 61 percent is used in MOBILE3 for
the city-specific environmental analyses. The July average day
adjusted fuel tank level of 60 percent is adjusted for
temperature variability as follows.
ii. Temperature Variability
The city-specific temperatures used for the economic
analyses {evaporative recovery credit runs) are 30-year average
daily July minimum and maximum temperatures for the specific
city or nearest available city. However, as with average RVP,
using the average daily temperatures instead of actual daily
temperatures tends to underestimate in-use diurnal emissions
because of the non-linearity of diurnal emissions with
temperature. Typical differences between the daily July
temperatures and the 30-year average July daily temperatures
are five-ten degrees when comparing minimum temperatures as
well as maximum temperatures.
To account for the daily temperature variability effect on
in-use evaporative emissions, the diurnal evaporative emissions
model was run for five cities at daily July 1985 minimum and
maximum temperature conditions and also at the average minimum
and maximum temperatures for that month.[43] UDIs were
calculated for each day, and the non-tampered emissions were
projected assuming the 2010 fuel-injected/carbureted vehicle
mix of 0.888/0.112. Table 2-32 presents the temperature
variability analysis for Boston. From the average of the daily
non-tampered emissions, the UDI which yields this average
diurnal emission value was determined. Using the average of
the July 1985 daily temperatures, average RVP, and the back
calculated UDI from the daily average emissions, the
-------
2-93
Table 2-31
In-Use Fuel Tank Level
After In-Use RVP Variability is Figured In
Environmental Analyses
(Design Value Day Temperatures)
Adjusted Fuel
City Tank Level (%)
Atlanta 61.3
Dallas 60.9
Denver 60.9
Detroit 61.2
Kansas City 62.2
Miami 60.9
New York 62.1
Phoenix 62.5
Average* 61.5%
Economic Analyses
(July Average Temperatures)
Adjusted Fuel
City Tank Level (%)
Atlanta 58.9
Dallas 59.1
Denver 59.2
Detroit 59.0
Kansas City 60.6
Miami 59.7
New York 60.4
Phoenix 60.9
Average 59.7%
Final fuel tank level used in city specific MOBILE3
environmental analysis runs.
-------
2-94
Table 2-32
Daily July Temperature Variability Effect
on In-Use Diurnal Emissions in Boston
(July Average RVP)
Non-Tampered
Date
T,(°F)
T2(°F)
UDI 1
Diurnal Emissions2 (q/test)
1
54
69
0 .5197
0.950
2
59
77
0 .8499
1.391
3
62
81
1 . 0696
2.071
4
66
85
1.3158
3.201
5
64
84
1.2865
3.047
6
68
85
1.2289
2.758
7
69
86
1 .2978
3 . 105
8
63
81
1 . 0341
1. 940
9
62
80
0 .9836
1. 768
10
68
86
1.3444
3 .358
11
69
85
1.1823
2 . 540
12
64
79
0.8299
1.344
13
62
82
1.1605
2.443
14
66
86
1.4313
3.865
15
71
85
1.0821
2.119
16
73
81
0.5738
0.974
17
66
76
0.5314
0.954
18
61
79
0.9363
1 . 621
19
64
88
1.7651
6.265
20
72
92
2.0256
8 . 635
21
70
83
0.9227
1.582
22
65
87
1.5950
4 .953
23
63
79
0.8669
1 .432
24
61
83
1.2907
3.069
25
61
86
1.6170
5. 112
26
72
80
0.5444
0.959
27
69
79
0.6171
1 . 007
28
63
73
0.4602
0.944
29
62
84
1.3590
3.440
30
68
88
1.5989
4 .981
31
62
69
0.2853
0.944
Avg.
65
82
2.670 - UDI which
yields this average
emission result is
1.2104
Temperature Adjusted V = 1.2104
8.445 gallon 1.04933
Temperature Adjusted V = 9.741 gallons
Temperature Adjusted = 16 - 9.741 + 2 X 100 = 51.6%
Fuel Tank Level 16
1 Calculated at 59.72 percent full tank, and RVP =11.5 psi.
2 Weighted for 2010 vehicle mix, 88.8 percent Fuel-Injected,
11 2 percent Carbureted
3 UDI at averaae survey 9VP, arrt average July 1985 temperatures.
-------
2-95
temperature variability adjusted fuel tank level was determined
as shown in Table 2-32. In other words, an adjusted in-use
fuel tank level was calculated which, in combination with the
average July minimum and maximum temperatures and average RVP,
yields the true emissions average for economic analyses.
Table 2-33 contains a summary of the temperature
variability fuel tank level calculations for all five cities.
The average temperature adjusted fuel tank level of 53 percent
is used in M0BILE3 for the city-specific economic analyses.
f. Emission Factors Under Various Control Scenarios
The evaporative emission factors used for this analysis
are city-specific. Each city included has specific RVP,
temperature, and alcohol blend fraction inputs which are
presented in Chapter 3. This section is intended to show the
effects of volatility control on evaporative emissions in three
example cities. The three cities chosen, Houston, New York,
and Boston, represent urban areas whose ozone design value day
temperatures and in-use or ASTM RVP conditions place them in
the low, average, and high categories, respectively, for
potential evaporative emissions, compared to all the areas
analyzed.
Table 2-34 contains the RVP and temperature information
necessary for these three cities to calculate the evaporative
emissions potential. The UDI and adjusted RVP used in the
diurnal and hot-soak equations of the evaporative emissions
model are also listed in Table 2-34. Both current conditions,
and control of in-use RVP to 9.0 psi in ASTM Class C areas
(with a proportional reduction of RVP in other ASTM classes)
are presented.
Based on the UDI and adjusted RVP, Table 2-35 presents the
predicted in-use evaportive emissions and reductions by fuel
metering type and the hot-soak and diurnal breakdown. The
rates listed are for 1981+ light-duty vehicles only, with
tampering included (based on an average tampering rate at
40,000 miles). The emissions are calculated from the equations
listed in Appendix 2-A.
Certain trends exist for the absolute diurnal and hot-soak
emissions. Diurnal emissions increase with increasing UDI,
which as already explained, is a function of RVP, the diurnal
temperatures and tank fill level. Hot-soak emissions, by
definition of the model, are equal for any city at its ASTM
limit. In other words, hot-soak emissions in Houston (ASTM
Class B in July) on a 10.0 psi RVP fuel would be assumed to be
equal to hot-soak emissions in Boston (ASTM Class C in July) on
a 11.5 psi RVP fuel to account for temperature differences
between ASTM class areas.
-------
2-96
Table 2-33
In-Use Fuel Tank Level After RVP and Temperature
Variability Are Figured In for Economic Analyses
(July Average RVPs and Temperatures)
Adjusted Fuel
City Tank Level (%)
Atlanta 51.6
Boston 51.6
Detroit 51.2
Salt Lake City 52.8
San Antonio 58.1
Average* 53.1%
Final fuel tank level used in city specific MOBILE3
economic analysis runs.
-------
2-97
Table 2-34
City-Specific Evaporative Emissions Model Inputs
City
RVP(psi)
T,(°F)
T2(°F)
UDI
Adi. RVP
Houston
10 . 5
76
93
1 .3818
12. 1
New York
11 . 5
72
92
1 .9766
11.5
Boston
11 . 5
73
97
3 .0175
11 . 5
Houston
7.8
76
93
0.5982
9.0
New York
9 . 0
72
92
0.9177
9.0
Boston
9 . 0
73
97
1.3030
9.0
-------
2-98
Table 2-35
City-Specific Emission Factors (q/test)
In-Use
1981+ LDV Evaporative Emission Factor
CARB FI
City
RVP
DI
HS
DI
HS
Houston
10 . 5
5.38
5.34
4 . 05
3.44
New York
11. 5
9 . 86
4 . 57
8. 77
2.87
Boston
11 . 5
20 . 24
4 . 57
22.66
2 . 87
Houston
7 . 8
2 . 70
2.64
1 . 94
1.09
New York
9 . 0
2 . 99
2.64
2 . 20
1. 09
Boston
9 . 0
4 . 86
2.64
3 . 61
1 . 09
Evaporative Emission Factor Reductions* (q/test)
CARB
City
DI
HS
Houston
New York
Boston
2.68(-50%)
6.87(-70%)
15.38(-76%)
2.80(-52%)
1.93(-42%)
1.93(-42%)
FI
DI
2.11(-52%)
6.57(-75%)
19.05(-84%)
HS
2 .35(-68%)
1.78(-62%)
1.78(-62%)
Both the absolute reductions and the percent reductions
are for reducing in-use RVP from current or ASTM RVP
levels to 9.0 psi in Class C Areas (proportional RVP
reductions in non-class C areas.)
-------
2-99
Patterns also exist for both the absolute evaporative
emission reductions, and percent reductions. The reduction in
diurnal emissions, both absolute and percent comparisons, is
greater for cities with higher RVP fuels given the same
proportional reduction from ASTM RVP. The reduction in
hot-soak emissions is equal for all cities when RVP is
proportionately reduced from ASTM levels. For these three
cities, the predicted hot-soak emission reduction for Houston
is higher than for New York or Boston, because the RVP in
Houston was above ASTM, and therefore the reduction is also
greater.
Although actual reductions will vary from city to city,
the estimates shown in the bottom half of Table 2-35 offer a
good range for this reduction in volatility for all of the
cities in this analysis. Emission reductions from other
vehicle classes and pre-1981 light-duty vehicles will be
different also. Chapter 3 will present the nationwide emission
inventories which are a population weighted average of the
product of the separate city emission factors and the city's
vehicle miles travelled.
g. Control of %i6o of Fuels
As already stated, volatility controls of certification
and in-use fuels are options for reducing HC emissions from
motor vehicles. In previous sections, the effects of
controlling the RVP of certification and in-use fuels on HC
emission factors were discussed. Another important volatility
parameter of certification and in-use fuels which could be
controlled is a mid-range volatility characteristic, such as
the percent evaporated at 160°F (%i6o). The %l6o value has
gathered attention because experimental work has shown the
maximum temperature in the carburetor after engine shut-off to
be around 160°F under standard test conditions.
The purpose of this section is to develop the emission
effects of controlling the %i6o value of in-use fuels. The
following section will discuss the approach taken with respect
to control of the %I6o of certification fuel. Based on the
emissions effects derived in this section and the costs
associated with the control of the %iS0 point derived in
Chapter 5, the cost effectiveness of controlling the %iSo of
in-use fuels will be presented later in Chapter 6.
i. Certification Fuel Control
Control of the llt0 of certification fuel is the
vehicle-related solution to the emission effect of the %iso-
The goal is to match the distillation properties of
certification fuel to in-use fuel distillation properties at a
-------
2-100
given RVP level. Currently, no specifications for %iso of
the certification fuel exist, however, target ranges for other
distillation points are specified which limit the %iso to
between 20 and 30 percent approximately.
The distillation properties of current in-use fuels and EF
test fuels are similar when compared at similar RVPs. Because
of this similarity in properties, the EF results already
account for the effect of the %i6<> on emissions along with
the effect of RVP. Also, the vehicle related costs of
certification fuel %iso control (i.e., a larger canister) are
already included in the vehicle cost estimates which will be
developed in Chapter 4. Therefore, the cost effectiveness
results which will be presented in Chapter 6 include the
certification fuel %iSo control effect. This means that
specifying a range for the %i6o point of the certification
fuel (or a cap based on current in-use fuel properties) can be
a cost effective approach to controlling evaporative emissions
depending upon the certification RVP level chosen.
ii. In-Use Fuel Control
Control of the %i6o of in-use fuels is the fuel-related
solution to the emission effect of the %i6<>. Hot-soak
emissions of carbureted vehicles, and not the hot-soak
emissions of fuel-injected vehicles are affected by control of
the %iso point. This is due to the differences between the
two fuel metering systems. In carbureted vehicles, fuel is
stored in the carburetor bowl before distribution to the
cylinders. After the engine is turned off, the fuel remaining
in the bowl is heated due to the dissipation of engine heat,
and fuel in the carburetor bowl essentially undergoes a simple
distillation. These evaporated hydrocarbons are the primary
source of carbureted hot-soak emissions. In fuel-injected
vehicles, the fuel is injected to the engine through a
pressurized and sealed system and is not susceptible to
distillation.
Diurnal emissions of fuel-injected and carbureted vehicles
are not affected to any extent, since diurnal vapor generation
is dependent on the RVP of the fuel rather than the %i6o
point. (RVP is a measurement of the volatility at 100°F which
is more representative of the typical maximum diurnal tank
temperature under standard test conditions.) It should be
noted that the %iso point can influence the load on the
canister at the start of the diurnal test and thus has the
potential to effect the measured emissions during the diurnal.
However, this effect has been found to be weak compared to the
RVP effect.[44]
-------
2-101
For tampered vehicles (i.e., uncontrolled emissions), it
was assumed only the hot-soak emissions from carbureted
vehicles would be affected. It is possible that evaporative
emissions from some tampered fuel-injected vehicles could be
affected, for example, vehicles with broken or missing
gaskets. However, the number of vehicles affected would be
very small and the effect on fleet emissions likewise.
Therefore, only the effect of the %iso point on carbureted
hot-soak losses will be incorporated into the predicted
emission reductions for tampered vehicles.
The experimental emissions data was obtained from a
program carried out by Automotive Testing Laboratories (ATL)
under contract to the EPA.[45] Vehicles were tested on four
fuels, a pair of 9.2 psi fuels and a pair of 10.9 psi fuels,
one fuel of each pair having a %l6o point of approximately 25
percent and the other fuel with a %16o point of approximately
40 percent. Both evaporative and exhaust emissions were
measured on all vehicles. Table 2-36 contains the paired
t-test statistical analysis of the ATL hot-soak emissions
versus fuel %i6o at constant RVP by fuel metering system.
The significance (a) verifies a relationship between %i6o
and carbureted hot-soak emissions, and lack of a relationship
between %i6o and fuel-injected hot-soak emissions (at a
confidence level of 95 percent).
Hot-Soak results of twenty-two carbureted vehicles (all
model year 1981) tested on the 10.9 psi fuels were used in this
analysis and are listed in Table 2-37. The 10.9 psi RVP fuel
emissions results were analyzed instead of the 9.2 psi RVP fuel
emissions results because a greater effect of %i6o on
evaporative emissions would be expected on the higher
volatility fuel. Therefore the cost effectiveness results,
which will be presented in Chapter 6, are as low as could be
expected. Control of RVP to the 9.0 to 10.0 psi range would
increase the $/ton figures by 10 to 20 percent.
Two different approaches were taken in extrapolating the
effect of the %16o on hot-soak emissions to in-use emissions,
an additive approach and a multiplicative approach. The
additive approach uses the actual increase in ATL hot-soak
emissions and assumes the increase would be the same for EF
hot-soak emissions. The multiplicative approach applies the
percentage change in ATL hot-soak emissions from a base test
fuel with %iso equal to 35 percent, to non-tampered EF
hot-soak emissions at 11.7 psi to estimate the increase in EF
hot-soak emissions. The multiplicative approach is one
possible way to address the discrepancy between hot-soak
emissions measured at ATL and hot-soak emissions measured under
the same test procedure at the EPA laboratory in Ann Arbor.[44]
-------
Table 2-36
Paired T-Test Analysis of ATL Hot-Soak Emissions
Fuel
Metering
System
CARB
RVP
N (psi)
22 9.2
Nominal Mean
_%i6o (g/test)
25.0
40.0
0.98
1.29
Mean Standard
Difference Deviation T-stat
-0.30455 0.76553 -1.8660
Significance
(a)
0.038
10.9
25.0
40.0
1.25
1.59
-0.33818 0.88190 -1.7986
0.043
FI
7 9.2
25.0
40.0
0.37
0.39
-0.01429 0.11717 -0.3226
0.379
10.9
25.0
40.0
0.45
0.48
-0.03571 0.18519 -0.5102
0.314
NJ
I
O
to
-------
2-103
Table 2-37
ATL Carbureted Hot-Soak Emissions on 10.9 psi RVP Fuel
(g/test)
ATL vehicle # %,6 o = 22¦6 %i 6 0 = 39.0
201 0.50 1.31
202 0.90 1.93
203 0.53 0.59
204 0.83 1.67
205 1.17 1.08
206 0.82 0.90
207 1.07 1.23
208 0.95 135
209 0.96 102
210 151 2.62
211 118 0.85
212 1.14 1.02
213 0.72 0.86
214 3.84 7.51
215 1.22 1.46
216 1.00 1.20
217 0.61 0.80
218 3.79 2.95
219 1.25 1.02
230 1.74 1.36
239 0.78 1.33
240 1.01 0.89
Average
1.25 g/test
l.59 g/test
-------
2-104
For both the additive and multiplicative approaches, two
estimates of hot-soak emissions versus fuel %i(0 values were
derived. For the first additive estimate, a straight line was
drawn through the two average ATL data points. However, based
on the general design properties of evaporative control
systems, this approximation would likely underestimate the
actual emission reductions. Therefore for the second estimate,
a curve was drawn through the two average ATL data points which
would better approximate the emission reductions, though
possibly on the high side. The two curves result m a range
for the predicted emissions reduction in the additive approach.
The straight line would be the result if the measured
hot-soak losses (i.e., controlled emissions) were a linear
function of uncontrolled emissions, which in turn should be
proportional to the %i6o point. However, since the
incremental effectiveness of the charcoal canister decreases as
uncontrolled emissions increase, hot-soak emissions would most
likely be an upward sloping non-linear function of the %i«0
point. In other words, for fuels with relatively low %i6o
points, the charcoal would be able to handle the additional HC
vapor loading and the hot-soak emissions would be lower than
predicted by the straight-line method. For fuels with
relatively high %i6o points, the charcoal would become
saturated with HC vapors and the hot-soak emissions would be
greater than predicted by the straight-line method. As this
function is unknown, a curve was drawn to represent a
reasonable upper limit of the degree of curvature.
To draw a worst-case curve, a similar relationship was
assumed to hold between diurnal emissions and fuel RVP as holds
between hot-soak emissions and the %i6o point. This should
be a reasonable upper limit for two reasons. One, RVP is the
most appropriate fuel parameter for diurnal emissions as %iSo
is for hot-soak emissions. Two, in-use diurnal emissions
increase more between 9 and 11.5 psi than hot-soak emissions
increase between %i6o values of 25 and 40 percent, indicating
a stronger movement from an unsaturated to a saturated
condition. Thus, the diurnal emissions of carbureted vehicles
versus fuel RVP from the EF data base were used to determine a
third intermediate point for the hot-soak emissions versus
% i s o curve.
Table 2-38 presents the numbers used to determine the
intermediate point of the hot-soak curve. The fraction of the
increase in diurnal emissions in going from 9.0 psi to 10.4 psi
as compared to going from 9.0 psi to 11.7 psi (35 percent of
the total diurnal increase) was applied to the increase in
average hot-soak emissions measured on the two different ATL
fuels. The intermediate hot-soak result was graphed at the
% 16 o point of 31.1 percent, which is equidistant between 22.6
-------
2-105
Table 2-38
Values Used to Curve Fit Average ATL
Hot-Soak Emissions as a Function of %l6o
Non-Tampered EF Carbureted
RVP (psi) Diurnal Emissions 1 (g/test)
9.0 2.60
52% < > 35%
10.4 5.29
48% < > 65%
11.7 10.33
ATL Carbureted
%!6 o Hot-Soak Emissions (g/test)
22.6 1.25'
52% < > 35%
31.I3 1.371
48% < > 65%
39.0 1.592
Actual EF data averages as listed in Table 2-12.
Average ATL emissions.
Predicted ATL emissions based on the increase in
carbureted EF diurnal emissions as a function of RVP for
the intermediate RVP level.
-------
2-106
percent and 39 percent as 10.4 psi is between 9.0 psi and 11.7
psi. Figure 2-15 presents the resulting additive functions of
ATL hot-soak emissions versus the %i6o point and can be used
to predict the decrease in hot-soak emissions of non-tampered
carbureted EF vehicles from reducing the %iSo point of any
in-use gasoline or alcohol blend.
For the multiplicative approach, the percent change in
hot-soak emissions was determined from the two estimates of
Figure 2-15 from a base fuel with a %i6o value of 35
percent. The base fuel in this analysis is the high volatility
EF test fuel which had a %i6o value of around 35 percent.
Table 2-39 presents the ATL emissions predicted under both the
straight line and curve methods at various %i6o values from
Figure 2-15 and the percentage change from the base fuel.
These percentage changes were applied to the EF non-tampered
data (4.26 g/test on 11.7 psi fuel) and are also listed in
Table 2-39. Figure 2-16 presents the resulting multiplicative
functions of hot-soak emissions versus the %i6o point, and
can be used to predict the decrease in hot-soak emissions of
non-tampered carbureted EF vehicles from reducing the %i6o
point of any in-use gasoline or alcohol blend. A different
analysis of the effect of %iso on in-use emissions also
resulted in a similar curve lying between the straight-line and
worst-case estimates as derived above.[44]
As previously mentioned, control of the %iso point of
in-use fuels would also cause emission reductions from tampered
carbureted vehicles. Since emissions from tampered vehicles
are uncontrolled emissions, the reduction in hot-soak emissions
should be proportional to the reduction in the %t6o point of
the test fuel upon which the emission estimates were based.
For example, reducing the %iso point of gasohol from 35
percent to 25 percent should result in approximately a 29
percent decrease (10/35 = 0.29) in tampered carbureted vehicle
hot-soak emissions. Based on the uncontrolled hot-soak
emission rates listed in Table 2-A-6 of Appendix 2-A, the
hot-soak emission reduction in lowering the %i6o value by ten
percent (i.e. 45 to 35 or 35 to 25) from a tampered carbureted
vehicle tested on a high volatility (11.5 psi) EF test fuel
would be 4.99 g/test.
The effects of the %iSo on fleeetwide emissions will be
examined in Chapter 6 along with the determination of the
cost-effectiveness of controlling the %l6o of in-use fuels.
2. Exhaust Emissions
The following sections investigate the relation between
fuel volatility and exhaust emissions. The first section
establishes the relation, the second section discusses the
-------
Figure 2-15
HOT—SOAK EMISSIONS versus &160
(Addftiv» Method)
STRAIGHT LINE
CURVE
-------
2-108
Table 2-39
Application of Percentage Change in ATL
Hot-Soak Emissions to Predict EF Hot-Soak Emissions
(Multiplicative Method)
Straight Line:
<1 so
Average ATL
Hot-Soak
Emissions 1
Change in
ATL Hot-Soak
Emissions from Base
Predicted EF
Hot-Soak
Emissions
22. 6
35.0
39.0
1.25 g/test
1. 50
1.59
-16.7%
Base2
+6.0%
3.55 g/test
4 .26'
4.52
Curve;
•6160
Average ATL
Hot-Soak
Emissions 1
Change in
ATL Hot-Soak
Emissions from Base
Predicted EF
Hot-Soak
Emissions
22
30
35.
39,
45.
1.25
1.35
1.47
1.59
1.80
g/test
-15.0%
-8.2%
Base2
+8.2%
+22.4%
3.62 g/test
3.91
4 .263
4 . 61
5.21
From Figure 2-15.
The 11.7 psi EF test fuel had %i6o around 35 percent.
Average EF Carbureted hot-soak emissions on 11.7 psi fuel.
-------
Figure 2-16
HOT—SOAK EMISSIONS versus 1 6 0
(MwltipiiQcitiv* Metnod]
«
-------
2-110
effect of volatility controls on vehicle emissions, and the
third section investigates the possible reasons for the fuel
volatility/HC exhaust emission correlation.
a. Correlation Between RVP and Exhaust Emissions
EPA's Emission Factor (EF) program includes tests for
exhaust emissions as well as evaporative emissions. Prior to
November of 1983, when testing was only performed using
Indolene, the effect of RVP on in-use emissions was not known.
Between October 1983 and July 1984, RVP appeared to have little
effect on exhaust emissions.[46 ] However, since July 1984
^¦(when the test sequence was improved)^ a significant effect has
been seen, particularly that high RVP fuel increases exhaust
emissions of HC and CO. No significant increase in NOx
emissions with higher volatility fuels has been noted. The
lack of an effect prior to July 1984 is presumed to be due to
the extra purging of the in-use canister during the evaluation
on Indolene, in which the HFET and various short tests were
conducted prior to testing on high RVP, commercial fuel.
Insofar as metropolitan areas are in non-compliance with
the NAAQS for CO, the CO emission reductions occuring with a
reduction of RVP would produce a public health benefit.
However, violation of the NAAQS for CO is not nearly as
widespread as that for ozone. Also, the primary focus of this
rulemaking is the effect of RVP on nationwide ozone NAAQS
compliance. Thus, only the effect of RVP on HC exhaust
emissions will be further discussed in this section. It should
be remembered that some CO benefit exists with respect to RVP
control.
An analysis was made of exhaust emissions versus RVP,
based on a group of 322 EF cars, tested on three fuels between
August 1984 and February 1986. The three fuels were tested in
the following sequence: commercial fuel (11.7 psi RVP), a blend
of commercial fuel and Indolene (10.5 psi), and Indolene (9.0
psi). The prep cycle prior to the diurnal test was a full LA-4
in all cases.
Cars with "large" HC exhaust emissions were found to be
insensitive to fuel volatility. Apparently the problems
causing these large emissions were masking the volatility
effect. Therefore, the ten cars with HC emissions exceeding
4.0 g/mi (on Indolene) were excluded from the analysis. In
addition, the three tampered vehicles excluded in the
evaporative emission analysis were also excluded, along with
one vehicle that had questionable test results (an increase in
HC exhaust emissions from 0.42 to 6.48 grams/mile for Indolene
and commercial fuel, respectively). This left 308 cars in the
data base for the analysis. The model years for these cars
were 1981 through 1984.
% lC>/&3 g, OllBA , LOETE. run Or\ I nclo 1 £r\£,
"Z.A on Y^V\£_r - TZ.VJP .foeJ . \r\do tS ^ ^-UP/ | r\Jio £
mcludf,s FTP, HFtTl ( a,Ut- "Espf-ed *s. Ci'P
9^^), - & uri o~P"
lH-U$r£ C(fr\d i~n\\ OiKisA I A o)
-------
2-111
Table 2-40 shows the average HC exhaust emission as found
in this analysis. The groups shown include carbureted (closed
loop, open-loop, all carbureted), fuel-injected, and all 308
vehicles. In every case, the average HC exhaust emission
increases with fuel volatility. However, since average values
do not show how individual vehicles react to fuel volatility,
the number of vehicles with larger HC exhaust emissions using
commercial fuel than from Indolene are shown in Table 2-41. In
general, 80 percent of the vehicles show larger exhaust
emissions with commercial fuel than with Indolene. A Wilcoxen
test showed a 0.00 percent probability that this difference was
due to random error.
To guantify how the HC exhaust emissions from individual
vehicles react to volatility, a Paired t statistic was
calculated to find the probability, a. Vehicles are
considered to show a significant exhaust difference with
respect to fuel volatility when a is less than 0.05 (95
percent confidence that the exhaust differences are not due to
random error). As shown by Table 2-42, all of the technology
classes (fuel metering systems) have an a of 0.000.
Therefore, the Paired t-Test indicates an extremely significant
relation between fuel volatility and HC exhaust emissions.
As a further check on the exhaust emission/volatility
correlation, the data base was further broken down into smaller
groups: method of aspiration; type of fuel injection - TBI/PFI;
season - summer/winter; and manufacturer - GM/Ford/Chrysler.
Vehicles from other manufacturers were included in the overall
database, but were too few in number for a Paired t-Test. An
examination of the GM, Ford, and Chrysler subcategories show
that for each group, fuel volatility causes a significant
difference in HC exhaust emissions. Within these groups, only
the Ford and Chrysler TBI systems indicate that RVP is not
significant. Since the Chrysler TBI subgroup has only 10
vehicles, this insignificance may be caused by random test
error. However, the Ford TBI subgroup was sufficiently large
enough (23 vehicles) to provide confidence in the Paired
t-statistic. No conclusive explanation has been found for
.v their sinoignificanco- to fuel volatility. Insufficient canister
^ capacity at high RVP fuels would result in the evaporated HC's
breaking through the canister rather than being captured and
later purged, however no relation was found between exhaust
emissions and evaporative emissions.
Fuel volatility is increased during the colder
temperatures of the winter season to overcome vehicle starting
problems. As a result, vehicles received during the winter for
testing may have a different initial canister loading than cars
received during the summer. Potentially, this seasonal
difference could confound the HC exhaust emission analysis, if
-------
2-112
Table 2-40
Average HC Exhaust Emissions (q/mi)*
RVP (psi) Carbureted Fuel-Injected All Vehicles
All Closed-Loop Open-Loop
9.0 0.76 0.82 0.68 0.43 0.59
10.5 0.80 0.84 0.74 0.46 0.62
11.7 0.89 0.95 0.82 0.51 0.69
* Data from 308 LDVs tested between August 1984 and February 1986 as
part of EPA's Emission Factor program.
-------
2-113
Table 2-41
Number of Vehicles*
RVP (psi)
Carbureted
Fuel-lniected
All Vehicles
All
Closed-Loop
Open-Loop
11.7>9.0**
120
69
51
119
239
Total
147
85
62
161
308
Percent* *
82%
81%
82%
74%
78%
Data from 308 LDVs tested between August 1984 and February
1986 as part of EPA's Emission Factor program.
11.7>9.0 = Number (percent) of vehicles where the HC
exhaust emissions are larger for commercial fuel (11.7 psi
RVP) than for Indolene (9.0 psi RVP).
-------
2-114
Table 2-42
Paired t-Test Results*
Category # of cars a** for HC
All vehicles
308
0
. 000
FI
161
0
. 000
TBI
92
0
. 000
PFI
69
0
. 000
Carbureted
147
0
. 000
Open-loop
62
0
. 000
Closed-loop
87
0
. 000
Winter
187
0
. 000
Summer
121
0
. 000
Turbo
Nat'1 Asp
289 -31 '
o o
. 001
. 000
Ford
75
o,
. 000
PFI
15
o.
,000
TBI
23
0 .
,306
Carb.
37
0 ,
.002
GM
116
0 .
, 000
PFI
26
0 .
, 001
TBI
49
0.
, 004
Carb.
41
0 ,
, 014
Chrysler
37
0 .
, 012
TBI
10
0 .
159
Carb.
27
0.
020
Data from 308 LDVs tested between August 1984 and
February 1986 as part of EPAs Emissions Factor program,
a =The probability that the differences seen are due
to random error and not due to differences between the
test conditions (i.e., commercial fuel versus
Indolene). Values below 0.05 indicate significant
differences.
-------
2-115
the canisters were not adequately purged before the FTP tests.
Therefore, to check the possibility of seasonal effects,
separate Paired t-Tests were made for cars tested during the
summer versus the winter. As shown in Table 2-42, no evidence
of seasonal variation was found.
The analysis of the 308 cars in the Emission Factor data
base indicate that for a majority of 1981-1984 vehicles,
exhaust emissions increase with increased fuel volatility. An
analysis of the exhaust emission benefits possible by in-use
RVP reduction will now be made.
b. Effect of RVP Control
Exhaust hydrocarbon emissions usually increase with an
increase in fuel RVP. At present, there is no limit on how
high in-use RVP can go. By setting an in-use RVP limit, a
reduction in HC exhaust emissions can be obtained. This "cap"
could provide a temporary reduction in HC emissions until
vehicles capable of handling higher RVP fuels become common
throughout the fleet, or the cap could be used to permanently
limit RVP and therefore reduce exhaust HC emissions without
vehicle modifications.
To calculate the exhaust benefit, a linear regression of
HC exhaust emissions versus actual RVPs was performed for 1981
and later model year cars. Vehicles in this age group
accumulate roughly 60 percent of the total vehicle miles
traveled (VMT) by LDVs, per year. The regression utilized the
322* cars from the EF data base, all of which were tested on
three different RVP fuels. Due to the similarity in regression
terms between the various technologies (carbureted vs.
fuel-injected), all classes were grouped together. Therefore,
using this 322 car data set, the slope was found to be 0.0387
g/mi/RVP with a constant of 0.593 g/mi. Using the 292 cars
with HC emissions less than 1.50 g/mi (on Indolene), the slope
was found to be 0.0478 g/mi/RVP with a constant of 0.0216.
To calculate the average change in HC exhaust emission
with RVP, an exhaust emission correction factor (CF) was
developed which normalized the data to 11.5 RVP. Therefore, at
11.5 RVP the CF is 1.0, and below 11.5 RVP the CF is less than
1.0. No upper RVP limit was imposed on the correction factor,
therefore CF values greater than 1.0 are possible. However,
since engines are already designed to handle 9.0 RVP fuel, the
CF is reset to the 9.0 level for RVPs less than 9.0 psi.
To obtain the average fleet response to RVP, the
regression incorporated the "high emitters" and tampered
vehicles previously excluded.
-------
2-116
Table 2-43 shows the HC exhaust emission benefits for RVPs
between 8.0 and 11.5, for both data sets. Lowering in-use RVP
from 11.5 to 9.0 psi results in a nine percent reduction in
exhaust hydrocarbons when all the vehicles are analyzed.
Considering only those vehicles with HC exhaust emissions less
than 1.50 g/mi (on Indolene), the HC exhaust reduction is 21
percent. For this analysis, the effect of volatility control
on exhaust emissions is based on the data set including all 322
vehicles.
Modifying the engine/evaporative system is another
potential method for reducing HC exhaust emissions on future
vehicles. An analysis of why the exhaust emissions are
sensitive to fuel volatility is made below m order to evaluate
potential modifications.
c. Potential Causes for Increased Exhaust Emissions
For a majority of existing vehicles, the change in HC
exhaust emissions can be directly correlated to fuel RVP.
These changes in HC exhaust emissions can be attributed to
changes in HC combustion dynamics and/or to changes in catalyst
efficiency - i.e. oxygen shifts causing shifts in the exhaust
air/fuel ratio. While in-cylinder combustion differences
between various RVP fuels may cause this exhaust difference,
most manufacturers claim that the quantity of unmetered fuel*
is the predominant factor.
Fuel volatility will alter the amount of unmetered fuel
introduced to the engine. An increase in unmetered fuel may
increase the charge density, which may then increase HC
emissions by hindering combustion. In this scenario, the
amount and timing of unmetered fuel becomes critical. In an
alternate theory, unmetered fuel affects HC emissions by
consuming excess oxygen and thereby decreasing the HC
conversion efficiency in the catalytic converter. With time,
the oxygen sensor should reduce the amount of metered fuel,
allowing the exhaust air/fuel ratio to return to
stoichiometry. However, this correction may be too slow or
inadequate, allowing a net increase in HC emissions.
No conclusive and confirmed tests have been performed to
show which reason(s) is/are responsible for the exhaust
effect. However, various studies performed by automotive
manufacturers and API, and EPA vehicle test programs can be
Metered fuel is the fuel introduced into the engine in
specified quantities. Unmetered fuel is the unmeasured
and loosely controlled fuel introduced through the
canister purge and fuel tank vent lines.
-------
Exhaust
2-117
Table 2-
Emissions
43
versus RVP*
Entire Data Set (322 cars):
RVP
8.0
8.5
9.0
9 . 5
10.0
10.5
11.0
11 . 5
HC (g/mi)
. 903
. 922
.941
.961
.980
. 999
1 . 02
1 . 04
CF
. 907
. 907
.907
.925
.944
.963
.981
1 .00
% Red.
9.3
9.3
9.3
7 . 5
5 . 6
3.7
1.9
0.0
Normal Emitters**
(292
cars):
RVP
8 . 0
8 . 5
9 .0
9 . 5
10 . 0
10 . 5
11.0
11.5
HC (g/mi)
. 404
.428
.452
.476
. 500
. 524
. 547
. 571
CF
. 791
. 791
. 791
.833
.874
.916
.958
1 . 00
% Red.
20 . 9
20.9
20 .9
16.7
12.6
8.4
4.2
0.0
Values were obtained from a linear regression of the
Emission Factor data base for 1981 and later LDVs,
tested between August 1984 and February 1986.
Normal emitters are those vehicles with HC and CO
exhaust emissions below 1.5 g/mi and 2.0 g/mi,
respectively (Indolene fuel).
-------
2-118
used directly or indirectly to better understand the cause. A
majority of these studies indicate the exhaust effect is
related to unmetered fuel. The following is a summary of two
EPA programs.
The first program investigated the effects of fuel tank
temperature on emissions. Automobiles are presently being
designed with fuel circulation pumps that return unused fuel
back to the fuel tank. Due to engine heat, the returned fuel
is hot and may eventually raise the fuel tank temperature,
which increases the amount of evaporated fuel which must be
purged. In this study, a series of replicated FTPs, using two
cars (a 1986 Buick and a 1984 Plymouth), two fuels (9.0 and
11.7 RVP), and two dynamometer cooling arrangements (EPA
standard and "high" cooling) were used.[47] The "high" cooling
arrangement resulted in lower tank temperature rises than shown
by track tests performed by Exxon and EPA.[48,49]
With the Buick, the fuel tank temperatures and exhaust
emissions were fairly constant for all conditions. Therefore
this vehicle does not help to explain why exhaust emissions may
change with fuel volatility.
The Plymouth however, was found to be sensitive to fuel
volatility and fan arrangement. On 11.7 RVP fuel, a 55 percent
reduction (1.10 to 0.50 g/mi) in HC exhaust emissions was
obtained with a corresponding 19°F drop in the tank temperature
rise. In another case, a similar reduction in HC exhaust
emissions was obtained by plugging the purge line. Examining
the exhaust CO levels can also be helpful in verifying the
effect of unmetered fuel, since CO is very sensitive to the
air/fuel ratio. For both fuels, the CO emissions were
increased by a factor of four with the warmer fuel tank
temperature.
The Plymouth test results indicate that tank heating of
high RVP fuel will increase the amount of unmetered fuel,
causing an increase in HC exhaust emissions. The Buick test
results support the possibility of designing vehicles that can
adequately manage high RVP fuel, without increased HC exhaust
emissions.
In another program, the EPA tested nine cars on both 9.0
and 11.7 psi RVP fuel, with and without connections to the
charcoal canister (purge lines were left unplugged).[50] The
average HC and CO results, along with the number of vehicles
with increased emissions are shown in Table 2-44.
With Indolene, there is basically no difference in HC
exhaust emissions for the vehicles tested with and without
-------
2-119
Table 2-44
Summary from Tampered Vehicle Study*
Test
Category
Ave. HC
Exhaust
Emissions
(q/mile)
# of Cars
w/Increased
HC Exhaust
Emissions**
Ave. CO
Exhaust
Emissions
(q/mile)
# of Cars
w/Increased
CO Exhaust
Emissions**
9.0 RVP w/*** 0.36 5 4.59 6
9.0 RVP w/out**** 0.33 - 4.22
11.7 RVP w/*** 0.42 8 6.93 8
11.7 RVP w/out**** 0.37 7 4.53 4
An EPA study of nine vehicles tested on 9.0 and 11.7 RVP fuel,
with and without proper connections to the charcoal canister. All
disconnected vapor lines to the canister were left unplugged, as
would be found in a tampered situation.
Exhaust emissions compared to exhaust emissions with 9.0 RVP fuel
and no connections to the charcoal canister.
w/ = with proper connections to the charcoal canister.
w/out = without proper connections (no connections) to the
charcoal canister.
-------
2-120
functional evaporative systems. Since the vehicles were
designed to adequately purge Indolene vapors with a minimal
effect on exhaust emissions, the lack of change in exhaust
emissions is expected if the exhaust emission phenomena is
purge related. However, with 11.7 RVP fuel, the average
difference in HC exhaust emissions almost doubled. Since
"high" RVP fuel causes a correspondingly larger uncontrolled
evaporative emission than "low" RVP fuel, these tests may
indicate that the increased hydrocarbon purge concentration
(from the uncontrolled evaporative emissions) exceeded the
design values, causing the larger exhaust emissions. The CO
levels also support the rich combustion theory of increased
exhaust emissions with increased consumption of unmetered fuel.
In both of these EPA studies, the increased HC exhaust
emissions appear to be related to increases in unmetered fuel.
This exhaust effect can be remedied by lowering in-use RVP to
certification levels (as discussed in Chapter 3), or by
redesigning vehicles to adequately manage the increase in
unmetered fuel. Suggested modifications in purge control will
be discussed in Chapter 4. It should be noted that some
vehicles are presently insensitive to fuel volatility while
other vehicles may be sensitive but still within the 0.41 g/mi
standard. Therefore, these suggested modifications are not
expected to be required for all new vehicles.
In addition, some vehicles may be able to reduce their
fuel tank heating and subsequent canister loading by altering
the placement of their fuel tank and lines relative to the
exhaust pipe. The relocation will reduce fuel heating, which
in turn will reduce fuel vaporization. However, due to the
variability in vehicle design, along with the individual design
considerations required for each model, this topic will not be
developed further.
3. Refueling Emissions
Refueling emissions are the third category of vehicle HC
emissions along with evaporative and exhaust emissions that are
affected by fuel RVP. Refueling emissions are the emissions
generated when someone purchases gasoline at the gas station
and fills their gas tank.
About 90 percent of all refueling emissions consist of
vapors displaced from the vehicle fuel tank by the incoming
gasoline. The total mass of these emissions depends on the
volume of vapor displaced and its density, which in turn are
determined by the temperature of the fuel being dispensed and
of the fuel already in the tank, the tank size and geometry of
the tank, the volatility of the fuel, and a number of other
minor factors. (The determinants of refueling emissions are
-------
2-121
discussed in greater detail in "Refueling Emissions from
Uncontrolled Vehicles," an EPA technical report.[1])
Less significant sources of refueling emissions are
spillage and underground tank emptying losses. Spillage occurs
as a result of "splash back" from the fill pipe or the escape
of gasoline from the dispensing nozzle. Underground tank
emptying losses represent the escape of vapor from the vent of
the service station underground storage tank. This emptying
loss occurs because as fuel is pumped into the vehicle fuel
tank, ambient air is ingested into the service station tank
through its tank vent. This "fresh" air causes liquid fuel to
evaporate in the tank until an equilibrium concentration
between the vapor and liquid phases is reestablished. As this
occurs, the pressure within the tank becomes greater than the
atmospheric pressure. This increases the total volume of vapor
in the underground tank, with the excess volume being emitted
from the vent in the form of gasoline vapor emissions. The
spillage and emptying loss sources each account for about five
percent of the total emissions associated with the refueling
process.[51] In this analysis, underground tank emptying
losses are included in Stage I emissions, and only displacement
and spillage losses are included in refueling emissions. This
is consistent with the gasoline marketing study.[2]
The composition and amount of refueling vapors depends on
their source (e.g., fuel tank displacement or spillage), and
the volatility of the fuel. For example, the largest source of
refueling emissions, that portion resulting from vapor
displacement in the vehicle fuel tank, does not consist of
fully volatilized gasoline. All available information shows
that the "light-end" hydrocarbons generally evaporate more
readily than the higher molecular weight hydrocarbons, and so
refueling emissions are primarily light-end hydrocarbons.
The portion of refueling emissions that results from
spilling gasoline, on the other hand, reflects the composition
of the liquid fuel. This is due to the total evaporation of
liquid fuel that is spilled. However, as noted above, it is
estimated that no more than five percent of total refueling
emissions currently result from fuel spillage and evaporation.
Thus, the statements made with respect to the composition of
refueling emissions due to vapor displacement are essentially
valid for total refueling emissions as well.
The estimates for uncontrolled refueling emission factors
used in this analysis are based on the information derived in
the previously mentioned technical report, "Refueling Emissions
from Uncontrolled Vehicles."[1] The equation is as follows.
-------
2-122
Refueling = Displacement + Spillage
Spillage = 0.32 g/gallon
Displacement = -5.909 - 0.0949 (AT)
+ 0.0884 (Td) + 0.485 (RVP)
(in g/gallon)
where
Td = dispensed fuel temperature
AT = difference between tank temperature
and dispensed fuel temperature
RVP = Reid Vapor Pressure of dispensed fuel
Insufficient data were available to derive dispensed fuel
temperatures on a local basis for all 61 ozone non-attainment
areas, so city-specific refueling emission factors could not be
generated. A nationwide-average refueling emission factor was
used instead and was assumed to apply in all cities.
The refueling report identified inputs for this equation
based on a summertime (five-month) analysis of available
ambient and dispensed fuel temperatures and ASTM-limit RVPs
weighted by state and/or regional fuel consumption. The
results were a dispensed fuel temperature of 78.8°F, a
difference between tank and dispensed temperature of 9.4°F and
a summertime nationwide ASTM average RVP of 11.3 psi.
The 11.3 psi RVP level derived from the ASTM limits is
equivalent to the 11.5 Class C RVP control case, which also
represents control to the ASTM limits. The refueling RVPs for
all other scenarios depicted in Table 2-45 were simply derived
by multiplying the Class C RVP by the ratio of the base
nationwide refueling RVP (11.3 psi) over 11.5 psi. This
results in the same percentage reduction in RVP for both the
refueling and evaporative emissions portions of the analysis,
which reflects what is happening in-use. Table 2-45 shows the
nationwide fuel RVP used in the refueling emission factor
equation associated with each control scenario of interest and
the resulting refueling emission factor.
The g/gallon estimates of refueling are divided by the
vehicle fuel economy to determine the refueling rates on a g/mi
basis, as used in MOBILE3. The top half of Table 2-46 presents
the uncontrolled refueling emission factors for all vehicles by
model year on an 11.5 psi fuel.
-------
2-123
Table 2-45
Refueling Emission Rates at Various RVP Levels
(g/gallon)
Uncontrolled Nationwide Average
Class C RVP Refueling RVP Refueling Rate
11.0 (1983) 10.8 5.7
11.7 (1988 and 11.5 6.1
beyond)
Controlled
Class C RVP
11.5 11.3 6.0
11.0 10.8 5.7
10.5 10.3 5.5
10.0 9.8 5.2
9.5 9.3 5.0
9.0 8.8 4.8
8.5 8.4 4.6
8.0 7.9 4.3
-------
Year
pre
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
Mode
Year
1990
1991
1992
1993
1994
1995
2-124
Table 2-46
Nationwide LDV Uncontrolled Refueling
Emission Factors on 11.5 psi RVP Gasoline
(g/mi)
Emission
Model
Emission
Factor
Year
Factor
0.4803
1985
0.2629
0.4803
1986
0.2574
0.4959
1987
0.2542
0 . 4919
1988
0.2510
0.5000
1989
0.2480
0.5000
1990
0 .2440
0.4518
1991
0.2402
0.4094
1992
0.2355
0.3910
1993
0.2319
0.3588
1994
0.2293
0.3546
1995
0.2251
0.3065
1996
0.2210
0.2864
1997
0.2179
0.2760
1998
0.2148
0.2760
1999
0.2118
0.2669
2000 +
0.2096
Nationwide LDV Controlled Refueling Emissions
Factors on 11.5 psi RVP Gasoline
(g/mi)
Emission
Model
Emission
Factor
Year
Factor
0 .0192
1996
0.0174
0.0189
1997
0.0171
0.0185
1998
0.0169
0.0182
1999
0.0167
0.0180
2000+
0.0165
0 .0177
-------
2-125
Table 2-46 (continued)
Nationwide LDT1 & LDT2 Uncontrolled
Refueling Emission Factors on 11.5 psi RVP Gasoline
(g/mi)
Model
Emission
Model
Emission
Year
Factor
Year
Factor
pre 1969
0 . 5496
1985
0.3446
1970
0.5496
1986
0.3408
1971
0.5701
1987
0.3370
1972
0.5648
1988
0.3333
1973
0.5755
1989
0.3315
1974
0.5755
1990
0.3280
1975
0.5126
1991
0.3245
1976
0.4959
1992
0.3228
1977
0.4586
1993
0.3210
1978
0.4692
1994
0.3210
1979
0.4841
1995
0.3177
1980
0.3861
1996
0.3128
1981
0.3610
1997
0.3096
1982
0 .3547
1998
0.3035
1983
0.3486
1999
0.3005
1984
0.3486
2000 +
0.2961
Nationwide LDT1 &
LDT2 Controlled Refueling
Emission Factors
on 11.5 ps i RVP
Gasoline
(g/mi)
Model
Emission
Model
Emission
Year
Factor
Year
Factor
1990
0.0260
1996
0.0248
1991
0.0257
1997
0.0246
1992
0.0256
1998
0.0241
1993
0.0254
1999
0.0238
1994
0.0254
2000 +
0.0235
1995
0.0252
-------
2-126
Table 2-46 (continued)
Nationwide HDV Uncontrolled
Refueling Emission Factors on 11.5 psi RVP Gasoline
(g/mi)
Model
Emission
Year
Factor
pre 1969
0.9457
1970
0.9457
1971
1.0033
1972
1.0702
1973
1.0000
1974
0.9385
1975
0.8815
1976
0.8957
1977
0.8520
1978
0.7881
1979
0.7559
1980
0.6971
1981
0.6711
1982
0.6455
1983
0.6131
1984
0.6094
Model
Emission
Year
Factor
1985
0.6094
1986
0.6082
1987
0.6070
1988
0.6022
1989
0.5986
1990
0.5945
1991
0.5905
1992
0.5854
1993
0.5749
1994
0.5690
1995
0.5653
1996
0.5607
1997
0.5612
1998
0.5576
1999
0 . 5546
2000 +
0.5505
Nationwide HDV Controlled Refueling Emissions
Factors on 11.5 psi RVP Gasoline
(g/mi)
Model
Emission
Model
Emission
Year
Factor
Year
Factor
1990
0.0467
1996
0.0441
1991
0.0464
1997
0.0441
1992
0.0460
1998
0.0438
1993
0.0452
1999
0.0436
1994
0.0447
2000 +
0.0433
1995
0.0444
-------
2-127
As is discussed later in Chapter 3, the M0BILE3 emission
model is run under two different refueling scenarios. The
first scenario assumes no control of refueling emissions on any
vehicles. The second scenario assumes that on-board technology
is required by regulation on all 1990 and later model years
gasoline vehicles.
A typical on-board system would be expected to consist, at
least in part, of a charcoal canister and a fill-neck seal.
The fill-neck seal prevents refueling emissions from entering
the atmosphere, and the charcoal canister traps the refueling
emissions to be burned in the engine later. The on-board
system eliminates the great majority of both displacement and
spillage losses. The control efficiency of a properly
operating on-board system is assumed to be 97 percent.
Tampering, malmaintenance and defects, and system
deterioration lower the in-use fleetwide efficiency which
MOBILE3 calculates. The controlled refueling emission factors
are presented in the bottom half of Table 2-46 for all 1990 and
later vehicles by model year.
B. Alcohol Blend-Fueled Vehicles
The addition of ethanol and/or methanol to gasoline
affects both evaporative and exhaust emissions. When dealing
only with gasoline, the emission of interest with respect to
ozone formation potential is HC, but when alcohols are added to
the fuel it is more accurate to speak of organics emissions,
since that encompasses HC as well as other
hydrogen-carbon-oxygen compounds that can play an important
part in ozone formation. The quantitative effects of blend use
on evaporative and exhaust organics emissions and ozone
formation potential will be covered later in this section, but
a qualitative description of those effects will be presented
here.
The addition of either ethanol or methanol to gasoline
increases the RVP of the resultant blend above the RVP of the
alcohol or the gasoline alone. This is shown in Table 2-47,
which summarizes the volatility effects of splash blending
ethanol or methanol/cosolvent with gasoline. If nothing is
done to the composition of the gasoline to compensate for this
RVP increase, similar directional effects on diurnal, hot-soak,
and exhaust emissions will occur as described previously for
gasoline. In other words, evaporative and exhaust emissions
increase with increasing RVP whether the RVP increase is due to
butane addition or alcohol addition. Therefore, the
determination of the effect of RVP on emissions will rely on
the earlier analyses of evaporative and exhaust emissions from
gasoline-fueled vehicles as a function of vapor pressure.
-------
2-128
Table 2-47
Blend Volatility
Volatility Increases
with Splash Blending
RVP (psi) 0.5-1.0 2.0-2.5
10%
Ethanol
5%
Methanol
+Cosolvent
10 - 15%
5 - 10%
-------
2-129
A number of modifications to those analyses, however, are
necessary to more precisely account for the presence of
alcohol. One modification accounts for the difference in
molecular weights of ethanol or methanol compared to
evaporative hydrocarbons. This factor is added because the
vapor pressure of a fuel determines (or is determined by) the
number of moles (rather than the mass) of that fuel which will
be in the vapor phase at a given temperature.
Another factor accounts for the increase in RVP which
occurs with the commingling of gasoline-alcohol blends with
gasoline-only fuels. Due to the non-linear nature of the RVP
increase of alcohol addition, the commingling of a 9.0 psi
gasoline with a 9.0 psi blend will result in a fuel with a
greater RVP than 9.0 psi. This situation would occur in
vehicle fuel tanks whenever vehicle owners switch from a
gasoline to a blend or vice versa.
In addition to the above RVP-related factors which
influence both diurnal and hot-soak emissions, blends tend to
increase hot-soak losses from carbureted vehicles because of
the change in the distillation curve with the addition of
alcohol. Figure 2-17 shows how the presence of alcohol tends
to increase the fraction of fuel evaporated at a given
temperature in the 120 - 200°F range, which corresponds to fuel
temperatures in the float bowl of a carburetor after the engine
is shut off.
Exhaust emissions can be affected by the use of blends in
two ways. In addition to the RVP-related increase mentioned
above, the second factor, increased oxygen content, tends to
decrease exhaust organics from many vehicles by causing
enleanment of the fuel/air mixture.
Finally, besides the quantity of organics emitted from
blend-fueled vehicles, the composition of those organics
differs from gasoline-fueled vehicles, which affects the
over-all photochemical reactivity associated with the
emissions. For instance, evaporative emissions from methanol
blend-fueled vehicles contain a certain fraction of methanol,
which is less reactive than evaporative HC vapor. Each of
these factors will be taken up in detail below, first for
evaporative emissions and then for exhaust emissions.
1. Effects of Alcohol Blends on Evaporative Emissions
a. Diurnal Emission Mass
As mentioned in the introduction to this chapter, the two
major fuel factors that determine diurnal emissions are vapor
pressure of the fuel itself and, for blends, commingling of the
-------
2-130
Figure 2-17
Effects Of OXINOL™ 50
On Basefuel Volatility
D-86 Results
LL
o
®
480n
440-
400-
360-
320-
| 280-
a
E
© 240-
200-
160-
120-
80
0 10 20 30 40 50 60 70 80 90 100
% Vaporized
-------
2-131
blend with gasoline (which is just another factor affecting the
vapor pressure of the final fuel). The analysis of the effect
of fuel vapor pressure on diurnal emissions was presented for
gasoline-only fuel earlier in this chapter, and that same model
will be used here for blends. As shown there, this means that
a blend with the same RVP as a gasoline yields the same number
of molecules (moles) of vapor as the gasoline during a diurnal
temperature rise.*
Note that this is number of moles of vapor rather than
mass of vapor. When an adjustment is made to account for the
lower molecular weight of ethanol (46) or methanol (32)
relative to gasoline vapor (approximately 64), the diurnal
emissions on a total organics mass basis are somewhat less than
the gasoline. For ten percent ethanol blends the organics mass
is about 96.7 percent as much as a gasoline of equal vapor
pressure, while for methanol-cosolvent blends it is about 87
percent as much. These values assume a ten percent by mass
average ethanol concentration in the ethanol blend
vapor,[52,53,54,55,56] and a 15 percent methanol concentration
in the methanol blend vapor.[57,58,59,60,61] For methanol
blends, it is assumed that the cosolvent alcohol (TBA, ethanol,
or other) is present in the vapor phase at a concentration low
enough to be neglected.
The other factor, commingling, refers to the mixing of
gasoline/alcohol blends with non-alcohol gasolines in vehicle
fuel tanks whenever consumers switch from one fuel type to the
other when refueling their vehicles at a service station. As
shown in Figure 2-18, most of the RVP increase with alcohols
occurs with the first one percent addition of alcohol. This
information has been obtained in laboratory testing in which
different proportions of gasoline and gasoline/alcohol blends
are mixed to determine the resulting RVP and
distillation.[61,62,63,64]
If a gas tank that is 20 percent full (i.e., has a 20
percent heel) of a gasoline/alcohol blend is refueled with
gasoline of equal RVP but without alcohol, the RVP of the new
mixture will be greater than either of the starting fuels
because of the interaction of the alcohol with the new
gasoline. It is not clear what an average heel is, but with a
ten percent heel of methanol blend the RVP increase when
filling the tank with gasoline would be about 1.4 psi. With a
50 percent heel the RVP increase would be about 2.0 psi.
Another difference between alcohol blends and gasoline is
that the vapor pressure of alcohols decreases more
drastically with lowered temperature than does the vapor
pressure of gasoline. Therefore, for a blend and a
gasoline with equal RVP (which is measured at 100°F), the
blend would probably have a lower vapor pressure than the
gasoline at a lower temperature. This factor has been
neglected in this analysis since many of the cities
considered have maximum diurnal temperatures close to
100 °F.
-------
2-132
For ethanol blends the effect is assumed to be about one
third as great based on (a) the relative RVP boosts of ethanol
and methanol blends and (b) the similar non-linearity of RVP
vs. concentration of either blend, as shown in Figure 2-18.
(i.e., the curves have similar shapes.) Therefore, in any area
where blends and gasolines are available to consumers, some
increase in average RVP and evaporative emissions due to
commingling would be expected, but this effect is less for
ethanol blends than methanol blends.
Some modeling has been done by ARCO to calculate what the
increase in pool average RVP would be for various scenarios
covering different blend market shares, degrees of customer
loyalty to a particular retailer, and percentages of fuel
remaining in the tank when it is to be refilled (heel). These
results are shown in Figure 2-19 for the scenario with what
ARCO considers to be a representative degree of customer
loyalty. Since the RVP increases calculated with the ARCO
model are fleet averages, the appropriate RVP increases to
apply to blend-fueled vehicles are greater by a factor of
roughly one/(market share). The resulting increases in RVP
assuming brand loyalty, 20 percent full tank at refueling, and
various market shares are shown in Table 2-48. A 0.2 psi
commingling effect was chosen as being most representative of
the current in-use situation, since gasohol dominates current
alcohol blend use with local market shares between 10 and 40
percent.
One factor that has not been taken into account with
respect to commingling is the fact that the vapor alcohol
content with blends is not a linear function of the liquid
alcohol content. This non-linear effect is shown in Figure
2-20 for a methanol/cosolvent blend (dashed line) and also for
a methanol-only blend (solid line). If this were taken into
account in the methanol blend calculations, it would tend to
slightly decrease the organics mass associated with commingling
(due to the lower molecular weight of methanol relative to
gasoline vapor) and also slightly decrease the reactivity of
commingled blend vapor due to the higher methanol content.
Although this has not been directly factored into the
calculations, it has been at least partially accounted for by
assuming a methanol vapor content near the high end of the
likely range (15 percent).*
Although no comparable vapor composition data are
available for ethanol blends, it is reasonble to assume a more
linear relationship between the vapor and liquid ethanol
content than for methanol blends based on the less severe vapor
Note that the composition and concentration of the head
space vapors has been shown to be highly dependent on
head-space temperature.[51a]
-------
2-133
Figure 2-18
*1
RVP EFFECT WHEN NIXINC
lOi; OXjCNOL. FUEL WITH
HYDROCARBON-ONL^ FUEL.
RVP V9 Alcohol Content
&
a.
>
as
20 -r-
-1*
-1.
IT
.! •
•
1
•
.. ...
1— i—i—
ID
20 40 60 80
10% OXINOL FUEL, V%
100
• 1 * *
MCOMOl COMCtXT**nOM VOl«
+¦
+
100
SO 60 40 20
HYDROCARBON-ONLY FUEL, V%
-------
Figure 2-19
ALCOHOL FUEL POCTRATION
LCOfND: HCCL ~ > I 10 it WW 20 • » • 90 • ¦ ¦ 40 » • • M
-------
2-135
Table 2-48
Commingling
Increases in Average RVP With Blend Use (psi)
Blend-Fueled
Market Fleet Average Vehicle Average 1
Share Ethanol Methanol Ethanol2 Methanol
3% — 0.025 — 0.8
10% 0.023 0.07 0.23 0.7
40% 0.06 0.19 0.16 0.5
1 These numbers are simply back-calculated from the fleet
averages. With a given (e.g., 3%) market share there is
not necessarily just that (3%) of the fuel that is
affected by commingling. Some fuel will not get
commingled at all, and other fuel will get commingled more
than once as more fuel is added to the tank with each new
fill-up.
2 Ethanol values are based on methanol blend values
multiplied by 1/3 which is roughly the ratio of the RVP
boost of ethanol addition (0.5-1.0 psi) to the RVP boost
of methanol/cosolvent addition (2.0-2.5 psi).
-------
2-136
Figure 2-20
t€TW*CL CONCENTRATION IN CA60LIKE HEAD6PACE
METHANOL IN LIQUID. VOL *
Methanol alone
Methanol + cosolvent
-------
2-137
pressure effects of ethanol. Also, the impact of any
non-linearity would be less for ethanol than methanol since the
molecular weight of ethanol is closer to that of gasoline
vapor, and there are no reactivity benefits as there are with
methanol.
b. Hot-Soak Emission Mass
For purposes of analysis hot-soak emissions can be
considered to depend on two fuel factors. First is the RVP
effect, which can be calculated exactly as it was for
gasoline. The second factor affecting hot-soak emissions is
the higher mid-range volatility, or distillation effect of
blends if the blends have not had their distillation curves
matched to gasoline. As examined earlier in this chapter in
the mid-range volatility section, it is highly unlikely that
the distillation curves of alcohol blends will match that of
gasoline unless required by regulation to do so.
Hot-soak emissions usually contain a higher proportion of
heavier hydrocarbons than diurnal emissions. The first reason
for this is that higher temperatures are involved, which are
able to vaporize higher molecular weight compounds. The second
reason is that a charcoal canister is more effective in
retaining the heavier compounds than the lighter ones, and some
hot-soak losses, especially in carbureted vehicles, escape
without going through the canister. For example, during a
hot-soak gasoline may boil off at the carburetor in the
vicinity of the engine resulting in emissions that resemble
whole gasoline vapors.[51a]
The calculation of the effect of the higher RVP of an
alcohol blend (if present) on hot-soak emissions is done
exactly as with gasoline. The adjustment for the lower
molecular weight of alcohol, which was done for diurnal
emissions, is not applied. In the case of diurnal emissions,
matched vapor pressure at some temperature implies equal moles
of vapor generated at that temperature. For hot-soak
emissions, however, a matched distillation curve means that
equal volumes of fuel are evaporated at any given temperature
in the ASTM D-86 test. For hot-soak emissions from carbureted
vehicles, the temperature of interest is 160°F, since that is a
typical fuel temperature in a carburetor float bowl during a
hot-soak.
If the specific gravities (densities) of two fuels are the
same, then equal volume percent evaporated also implies equal
mass evaporated. The specific gravity of methanol is 0.796,
while that of gasoline normally ranges from about 0.70 to
0.75. However, the fraction of gasoline that evaporates under
hot-soak test conditions is composed of a higher proportion of
lighter hydrocarbons with specific gravities around 0.65. This
-------
2-138
means that for equal volumes of fuel evaporated at a given
temperature, a methanol blend would be expected to have a
slightly greater mass evaporated than an HC-only fuel.
Calculations that account for this factor indicate that it
would result in roughly a three percent greater projected mass
of hot-soak emissions from a methanol blend compared to a
gasoline with matched %i6o. For ethanol blends this effect
would be less, since the density of ethanol is closer to that
of gasoline and the vapor generated from ethanol blends tends
to have a smaller fraction of alcohol than that from methanol
blends.
This is a very simplified view of a complex issue, since
it does not take into account other variables such as the
possible influence of the charcoal canister. Therefore, for
purposes of this analysis it will be assumed that if the RVP
and %ieo of a blend are equal to that of a gasoline, the
resulting hot-soak emissions will also be equal on a total mass
basis (rather than a total moles basis as was the case for
diurnal emissions).
Table 2-49 shows the net RVP effect on combined gram/mile
diurnal and hot-soak organics emissions relative to gasoline
assuming gram/mile values are calculated using the M0BILE3
equation:
Gram/mile = (diurnal grams + (3.05 * hot-soak grams))/31.1
The weighting of carbureted to fuel injected vehicle
evaporative contributions for the 1990 and 2010 fleets are
based on both relative VMT and relative gram/mile evaporative
emissions. These weights depend on the RVP of the fuel, so a
range of 9.0 - 11.7 psi was used. In 1990, at 9.0 psi the
weighting is .68/.32 carbureted/F.I. total evaporative
organics, and at 11.7 psi the weighting is .56/. 44. In 2010,
at 9.0 psi the weighting is .22/.78 carbureted/F.I. total
evaporative organics, and at 11.7 psi the weighting is .14/.86.
The distillation effect on hot-soak emissions results from
the greater mid-range volatility of blends shown in Figure
2-17. This effect could be neglected if the distillation curve
of the blend were matched to gasoline, but due to the cost of
such matching it is not likely to be done by refiners unless
specifically required by regulation.
Hot-soak emissions from carbureted vehicles depend on the
distillation curve of the fuel in addition to the RVP.
Addition of methanol and/or ethanol to gasoline tends to
increase the volatility of the fuel in the 120°F - 200°F
temperature range as evidenced by the increased amount of fuel
evaporated at these temperatures in an ASTM D86 distillation
test. This effect was shown in Figure 2-17 with typical graphs
of temperature versus percent evaporated for a gasoline and a
-------
2-139
Table 2-49
Projected RVP Impact On Evaporative Organics Mass
From Blend-Fueled Vehicles*
10% Ethanol Blend
Carbureted
Fuel Injected
1990 Mix
2010 Mix
Relative to Gasoline
Matched RVP 1.0 psi Margin
-1.0-1.6%
-1.6-2.0%
-1.2-1.8%
-1.4-1.9%
+24-55%
+53-76%
+33-64%
+46-73%
Methanol/Cosolvent Blend
Carbureted -3.5-5.4%
Fuel Injected -5.4-6.7%
1990 Mix -4.1-6.0%
2010 Mix -4.9-6.5%
+21-47%
+47-65%
+29-55%
+41-63%
This includes the RVP effect on both diurnal and hot-soak
emissions. The minimums correspond to current
certification fuel RVP (9.0 psi) and the maximums to
current in-use RVP (averaging near 11.5 psi for ASTM
Class C).
-------
2-140
matched RVP methanol/cosolvent blend. Ethanol blends produce a
similar "knee" in the curve as methanol blends, but the knee is
offset toward higher temperatures due to the higher boiling
point of ethanol.
This temperature range corresponds to the temperatures
experienced by fuel in hot-soak situations during the SHED
hot-soak test. For instance, on carbureted vehicles, 160°F is
a typical temperature for fuel in the carburetor float bowl
during a hot-soak. On fuel injected vehicles, for which the
fuel tank is the main source of hot-soak losses, the fuel
temperature in the tank may range up to 120°F depending on how
the tank is situated relative to the exhaust system and how
much cooling it received during the prior driving.
As shown in Table 2-50 the percent evaporated at 160°F
(%,6o) in the ASTM D86 distillation test increases by roughly
12 percent for a matched RVP ten percent ethanol blend and five
percent for a matched RVP methanol/cosolvent blend relative to
gasoline.[62,65,66] (e.g., for ethanol, an increase from 25
percent evaporated to about 37 percent evaporated at 160°F.)
Since these values are based on experimental data, it can be
assumed that they account for whatever butane removal would be
needed to achieve equal RVP, as well as the effect of that
butane removal on %iso. The amount of change in l1S0 with
blends is dependent on the individual base gasoline composition
as well as the type and amount of alcohol used.
The effect of commingling on %i60 is negligible, since
the effect of alcohol on %iSo is a nearly linear function of
the amount of alcohol, whereas the RVP effect is very
non-linear.
The effect of a change in %i6o, as separate from any RVP
change, has been investigated by EPA by testing about thirty
2-gram vehicles with gasolines having an identical RVP (10.9
psi) but differing %i 6 0 (23 percent for one fuel and 39
percent for the other).[44] These data indicate an increase in
hot-soak emissions relative to gasoline for vehicles with
carburetors, but no significant change for those with fuel
injection. A more detailed description of this analysis was
presented earlier in this chapter. This analysis has been
applied to blends and the calculated effects of the increased
%ie o on hot-soak emissions are shown in Table 2-50.
In the case of a blend meeting a 1.0 psi higher RVP limit
than gasoline, the same calculations apply, except for two
adjustments. First, with a 1.0 psi margin for the blend there
is a greater difference in %i6o relative to gasoline than in
the matched RVP case. The first part ("Full Effect") of Table
2-51 shows the increases in %i6o and the corresponding
increases in evaporative emissions that would be expected with
blends that have an RVP l.o psi greater than gasolines.
-------
2-141
Table 2-50
Distillation Effect On Mass Hot-Soak Losses
(Relative to Gasoline of Equal RVP)
Delta Delta Hot-Soak1 Delta g/Mile2
Blend %16 o Carb. F. Inj. Carb. F. Inj.
Ethanol 12% +8.3-21% +0% +4.8-11.7% +0%
Methanol 5% +4.4- 8% +0% +2.6- 4.5% +0%
1 Based on analysis of ATL test data in Chapter 2. The low
number is based on an additive analysis, and the high
number is based on a multiplicative analysis.
2 The hot-soak percentage is weighted three times the
diurnal percentage, and is also weighted by MOBILE3
hot-soak versus diurnal g/test in calculating percent
changes in g/mile.
-------
2-142
Table 2-51
Distillation Effect On Mass Hot-Soak Losses
(Relative to Gasoline at 1.0 psi Lower RVP)
Blend
Delta
o
•6 1 6 0
Delta Hot-Soak1 Delta
Carb. F. Inj. Carb.
g/Mile*
F. Inj
Full Effect:
Ethanol 15%
Methanol 7%
+12.8-27%
+6.0-11.8®
+0%
+ 0%
+7.2-15%
+3.5-6.6%
+ 0%
+0%
Minus 4%ig o
Included in
RVP Effect:
Ethanol
Methanol
11%
3%
+8.9-19%
+2.6- 5%
+ 0%
+0%
+5.1-10.8% +0%
+1.5- 2.7% +0%
(splash methanol,
10%
2.3 psi above base)
+8.1-17% +0%
+4.7-9.3%
+0%
1 Based on analysis of ATL test data in Chapter 2. The low
number is based on an additive analysis, and the high
number is based on a multiplicative analysis.
2 The hot-soak percentage is weighted three times the
diurnal percentage, and is also weighted by MOBILE3
hot-soak versus diurnal g/test in calculating percent
changes in g/mile.
-------
2-143
The second adjustment arises from the fact that the high
RVP test fuels used in EPA's emission factor program not only
had a higher RVP than the low RVP fuel, but also had a greater
%lG0. This effect on hot soak emissions is automatically
included when the 1.0 psi RVP difference is accounted for.
Therefore, a portion of the 'Full Effect' increases in %lSo
should be removed.
According to the fuel specifications for the gasolines
used in the Emission Factor (EF) work (which was the basis for
the analysis of vapor pressure effects on evaporative
emissions) the 2.7 psi difference in RVP between the lowest and
highest RVP test fuels was accompanied by a difference in
%l60 of about ten percent. Therefore, a 1.0 psi increase in
RVP is assumed to correspond to a four percent increase in
%i60 in the gasoline. The subtraction of this 4%i6o is
shown in the second part of Table 2-51. For comparison the
values corresponding to a splash methanol blend are also shown,
although this is not an option that is being considered.
c. Reactivity of Evaporative Emissions
This issue relates to the fact that a portion of the
evaporative emissions from gasoline/alcohol blends consists of
alcohol, and various studies conducted by EPA, CARB, and others
indicate that, on a mole carbon basis, methanol is less
photochemically reactive than typical hydrocarbon vapors. In
other words, when exposed to sunlight, methanol's carbon atom
reacts to form ozone more slowly than typical hydrocarbons.
The reactivity of ethanol, however, is expected to be greater
than methanol and on the same order as hydrocarbons such as
butane or toluene based on its rate constant for reaction with
OH radicals.[67] The subject of hydrocarbon reactivity is
covered in more detail in Appendix 3-A.
The two factors involved in this issue are: (1) what
portion of the evaporative emissions consists of methanol, and
(2) how much less reactive is methanol relative to the
hydrocarbons it displaces. The first factor was covered above
in the section on blend RVP effects. As already mentioned,
based on available data for methanol/cosolvent blends a
methanol content of 15 percent of the total vapor mass (HC plus
alcohol) is assumed in both diurnal and hot-soak emissions.
For ethanol blends an ethanol content of ten percent is assumed.
EPA photochemical modeling and smog chamber data for
single day episodes indicate that methanol is about two
percent - 43 percent as reactive as non-methane hydrocarbons
on a carbon-for-carbon basis.[68,69] Therefore, the much lower
reactivity of the methanol in the vapor needs to be accounted
for in determining the effect on ozone forming potential of
methanol blends. Table 2-52 gives a comparison of reactivity
with gasoline-only evaporative vapor and with vapor containing
-------
2-144
Table 2-52
Reactivity of Blend Evaporative Orqanics
Component:
Gasoline
Ethanol
Methanol
Relative reactivity
per Carbon atom 1.00
Relative Carbon atoms
per gram of component 1.00
Blend vapor composition 100% HC
(mass %)
Relative reactivity per
gram of blend evaporative
organics 1.00
1.00
0. 62
90% HC
10% EtOH
0.962
0.02-0.43
0 . 44
851 HC
15% MeOH
0.851-0.878
-------
2-145
either ten percent ethanol or 15 percent methanol by mass. The
reactivity per gram of ethanol blend vapor is slightly lower
than gasoline (by four percent), because although an equal
reactivity per carbon atom is assumed, ethanol has fewer carbon
atoms per gram than gasoline. Methanol blend vapor is
estimated to be about 12 percent - 15 percent less reactive
than gasoline vapor.
A more recent multi-day study conducted at the Riverside
smog chamber by CARB indicates that methanol may react more
completely given a second day of irradiation, and if this is
validated in subsequent modeling, it could negate at least some
of the above reactivity benefits of methanol substitution in
multi-day ozone situations.[70] However, the extrapolation of
smog chamber results to real world situations is very complex
and more study is necessary before it is clear how this smog
chamber work affects our current understanding of methanol's
reactivity in the atmosphere.
2. Effect of Alcohol Blends on Exhaust Emissions
a. Exhaust Emission Mass
The use of gasoline-alcohol blends can affect the quantity
of exhaust organics in two ways. First, the presence of oxygen
in the fuel tends to alter the fuel/air mixture, and second the
RVP itself affects exhaust HC.
Regarding the first factor, neat methanol has an oxygen
content of 50 percent by weight while that of ethanol is 35
percent. When blended with gasoline per existing EPA waivers
the oxygen content of all blends is limited to 3.7 percent (3.5
percent for Oxinol). Oxygen content can be an important factor
in exhaust emissions from vehicles with open-loop
(non-feedback) control systems because the presence of oxygen
in the fuel makes the air/fuel mixture leaner than if the fuel
consisted only of hydrocarbons. If a carburetor is operating
rich on gasoline, with the excess HC and CO emissions that
result from rich operation, and then an oxygenated fuel is used
instead of gasoline, the mixture will become leaner and the
excess HC and CO emissions would be expected to decrease.
As summarized in Table 2-53, a large gasohol study showed
a substantial reduction in average exhaust HC for open-loop
vehicles using ethanol splash blends compared to base gasolines
of 9.0 and 10.0 psi RVP. [65] With closed-loop vehicles the
results were sufficiently mixed that no significant change in
average HC could be attributed to the addition of the ethanol.
-------
2-146
Table 2-53
Exhaust HC Effects of Blends
Study Fuels
EPA-MSED[65 ] a) Indolene
(ethanol) b) (a)+10% ethanol
c) 10 psi gasoline
d) (c)+10% ethanol
e) (d) volatility
matched with (c)
Blend Results
15-20% reduction in exhaust
HC with splash blends in
open-loop vehicles.
No significant change for
closed-loop vehicles.
DuPont [13] Indo1ene,
ARCO [71], commercial gasolines,
CRC [21] matched volatility
methanol blends
7.1-26.4% reductions exhaust
HC in open-loop vehicles.
0-5% reductions in exhaust HC
in closed-loop vehicles.
EPA [62] a) Indolene, 6% increase in exhaust HC in
b) commercial open-loop vehicles,
gasoline
c) Oxinol blend No change in exhaust HC in
volatility closed-loop vehicles,
matched with (b)
-------
2-147
Regarding methanol blends, test programs conducted by
DuPont,[13] ARCO,[71] and the Coordinating Research Council
(CRC), [ 21 ] also showed substantial reductions in average
exhaust HC relative to gasoline for open-loop (pre-1981)
vehicles using a five percent methanol/2.5-5 percent cosolvent
blend. However, a slight decrease in HC was found for
closed-loop (post-1981) vehicles. Many of the ARCO vehicles
had high HC and CO emissions when tested with gasoline,
indicating that they were running richer than optimum, and thus
probably showed a greater benefit from the leaning effect of
the blend fuel than if the calibration had not been rich. Each
study used 8 to 11 vehicles, approximately half of which were
open-loop and half closed-loop. The study averages for
closed-loop cars ranged from a decrease of 18 percent to an
increase of eight percent. The range of results for individual
vehicles is much greater, with many showing decreases while
others show increases.
Another relevant study was conducted by EPA using 23
in-use ("Emission Factor") vehicles.[62] This study showed an
average increase of exhaust hydrocarbons of about six percent
relative to gasoline for the eight open-loop vehicles tested
with a methanol/TBA blend of similar RVP, and no change in
average hydrocarbons for the 15 closed-loop vehicles. EPA test
experience has shown that consistency in test sequence/
preconditioning of the evaporative control system is an
important factor that could lead to discrepancies in results of
different studies.
A second factor affecting exhaust HC is fuel volatility.
As discussed earlier in this chapter, there is a small but
statistically significant and measureable increase in exhaust
HC with increasing RVP. In the gasohol study mentioned above,
the 10 psi base gasoline had 19 percent greater exhaust HC than
the 9.0 psi gasoline, and the 10.6 psi blend had 26 percent
greater HC than the 9.8 psi blend. However, for some reason
the volatility adjusted 10.0 psi blend yielded HC emissions
greater than the 10.6 psi splash blend and equivalent to the
10.0 psi base gasoline. As mentioned above, there may have
been canister preconditioning differences that influenced these
results.
Based on analysis of the far larger gasoline data base of
the EPA Emission Factor (EF) program, the increase in exhaust
HC with increasing RVP is approximately 0.04 g/mile (four
percent) per psi. For blends that have not been volatility
adjusted to be equal to gasoline, the increased exhaust HC from
the greater RVP would tend to partially counteract the
leaning-out effect described above. Also, the degree of blend
volatility control would determine how much this factor affects
exhaust HC. If blend RVP is controlled to the same level as
gasoline, then this factor would not be expected to affect
exhaust HC.
-------
2-148
An analysis of these data indicates that some exhaust
hydrocarbon benefits are expected with open-loop vehicles using
the waived methanol or ethanol blends at an RVP equal to
gasoline, but much smaller exhaust hydrocarbon changes are
expected for closed-loop vehicles.[72] Likely exhaust HC
reductions were determined to range from 7.2-11.4 percent for
open loop and 0-4.7 percent for closed loop vehicles. A linear
regression of the emission data versus fuel oxygen content
yielded an average reduction of 7.4 percent for open-loop
vehicles and 4.6 percent for closed-loop vehicles. Weighting
these two factors by an estimated total exhaust HC contribution
from open versus closed-loop vehicles yields a 7.0 percent
average reduction in exhaust HC for blend-fueled vehicles.
This split was estimated by using MOBILE3 with RACT I/M and
comparing the total tons of exhaust HC from pre-1981 and 1981+
light-duty vehicles in the year 1990.
For the scenario with a 1.0 psi RVP blend margin, the RVP
effect on exhaust organics (0.04 g/mile per psi) needs to be
added to the above oxygen effect. The result of this is a four
percent greater level of exhaust organics than in the RVP
matched scenario, or a net three percent reduction due to
alcohol usage.
b. Reactivity of Exhaust Organics
This issue concerns the fact that a portion of the exhaust
emissions from gasoline and blend fueled vehicles consists of
formaldehyde (HCHO), which is approximately three times as
photochemically reactive as typical exhaust hydrocarbons.
Exhaust emission data from blend fueled vehicles have been
examined to determine if such emissions contain a different
proportion of formaldehyde than exhaust emissions from the same
vehicles fueled with gasoline.
Two studies with 10 percent ethanol blends measured total
aldehydes and the results indicated that the aldehyde
proportion of total organics could increase slightly on average
for catalyst equipped vehicles.[52,73] These are of limited
applicability since HCHO was not specifically measured and one
study was conducted at high altitude. However, another study
measured HCHO specifically using two 10 percent ethanol blends
in two oxidation catalyst vehicles and one three-way catalyst
vehicle. [74] This study found that the HCHO percent of total
HC remained essentially unchanged when switching from the base
gasolines to the ethanol blends. Therefore, no adjustment for
exhaust reactivity has been used in calculating the ozone
impact of ethanol blends.
Regarding methanol blends, there are two counteracting
factors affecting exhaust reactivity. First, there is an
increase in percentage of formaldehyde, and second, there is a
small amount of unburned methanol in the exhaust organics.
-------
2-149
Regarding aldehydes, one study by CRC for DOE using various
methanol blends measured only total aldehydes, and the results
were so scattered that no particular effect of blend use on
total aldehydes could be determined. However, an EPA study
measured HCHO specifically and found it to be two percent of
exhaust HC with a five percent methanol/five percent TBA blend
(similar to Oxinol) versus one percent with gasoline.[59 ]
Since the reactivity of formaldehyde is approximately three
times that of exhaust hydrocarbons, a one percent increase in
the formaldehyde concentration would increase the exhaust
reactivity by roughly two percent.
However, this effect is balanced by a corresponding
increase in exhaust methanol concentration with use of a
methanol blend. Various studies have found this concentration
to range from almost zero up to as high as eight mass percent,
but a typical value is two to three percent. Since the
reactivity of methanol is much less than exhaust hydrocarbons,
this concentration of methanol would decrease the exhaust
reactivity roughly counterbalancing the increase in reactivity
from the increased formaldehyde concentration. Therefore, as
with ethanol blends, the exhaust reactivity of methanol blends
is assumed to be the same as gasoline.
3. Combined Evaporative and Exhaust Effects of Alcohol
Blend Usage on Ozone
As shown in Tables 2-54 and 2-55, for 1990 and 2010 fleet
mixes respectively, the estimated exhaust organics and ozone
benefits from blend use are nearly exactly offset by the
increases in evaporative organics and ozone potential in the
matched RVP scenario for both ethanol and methanol blends.
However, in the 1.0 psi margin scenario, increases of about one
percent - 13 percent in 1990 and about one percent - ten
percent in 2010 in fleet ozone potential relative to a
gasoline-only case are predicted, depending on blend market
penetration.
Tables 2-56 and 2-57 provide a detailed breakdown of all
the contributing factors to the final results in Tables 2-54
and 2-55. Tables 2-56 and 2-57 summarize the estimated exhaust
and evaporative effects of ethanol and methanol blend use on
average organic emissions, adjusted for reactivity, from the
portion of the light duty vehicle fleet using those blends.
Note that the commingling effects attributable to blend-fueled
vehicles depend on the market share of the blend, so the
commingling values in Tables 2-56 and 2-57 were back-calculated
from fleet average values by dividing the fleet values by the
market, share.
-------
2-150
Table 2-54
Projected Fleet Ozone Impact of Blends in 1990
% Change in Total (Exhaust + Evap) LDV Ozone Formation Potential
From a Gasoline-only Baseline1
3% Share 10% Share 40% Share
RVP + 1 psi RVP +1 psi RVP +1 psi
Matched Margin Matched Margin Matched Margin
Ethanol Blends
TOTAL FLEET EVAP
OZONE EFFECT - - +0.16-.87 +1.6-3.5 +.37-2.8 +6.2-13.3%
TOTAL FLEET EXHAUST
OZONE EFFECT - - -0.36% -0.14% -1.4% -0.55%
TOTAL FLEET
OZONE EFFECT - - -0.20-(+).51 +1.5-3.4 -1.0-( + )1.4 +5.6-12.6°,
Methanol Blends
TOTAL FLEET EVAP
OZONE EFFECT -.01-(+).35 +.37-1.1 -.17-(+)0.84 +1.1-3.2 -1.7-(+).96 +3.3-10.3%
TOTAL FLEET EXHAUST
OZONE EFFECT -0.11% -0.04% -0.36% -0.14% -1.4% -0.55%
TOTAL FLEET
OZONE EFFECT -0.12-( + ).24 +. 33-1.1 -. 53-( + ).48 +0.9-3.0 -3.1-(-)0.4 +2.7-9.7%
1 Based on actual mass of HC plus alcohol rather than FID measurement.
Evap/enhaust organics contribution has been weighted by .452/.548.
-------
2-151
Table 2-55
Projected Fleet Ozone Impact of Blends in 2010
% Change in Total (Exhaust + Evap) LDV Ozone Formation Potential
From a Gasoline-only Baseline'
Ethanol Blends
3% Share 10% Share 40% Share
RVP +1 psi RVP + 1 psi RVP +1 psi
Matched Margin Matched Margin Matched Margin
TOTAL FLEET EVAP
OZONE EFFECT - - +.08 -.25 +1.4-2.3 4-.02-.49 +5.6-9. 1°.
TOTAL FLEET EXHAUST
OZONE EFFECT - - -0.36% -0.08% -1.5% -0.34%
TOTAL FLEET
OZONE EFFECT - - -0.28-(-).ll +1.3-2.2 -1.5-(-)1.0 +5.3-8.8%
Methanol Blends
TOTAL FLEET EVAP
OZONE EFFECT +0.06-.22 +.41-.77 +.04-.51 +1.2-2.3 -.79-(+).39 +3.9-7.6%
TOTAL FLEET EXHAUST
OZONE EFFECT -0.11% -0.03% -0.36% -0.08% -1.5% -0.34%
TOTAL FLEET
OZONE EFFECT -.05-(+).ll +.38-.74 -,32-(+).15 +1.1-2.2 -2.3-(-)l.l +3.6-7.4%
1 Based on actual mass of HC plus alcohol rather than FID measurement.
Evap/exhaust organics contribution has been weighted by .301/.699.
-------
2-152
Table 2-56
EVAPORATIVE
RVP
Distillation
Commingling"
Projected Organics and Ozone Impact
of IPS Ethanol Blends On Blend Fueled Vehicles
Total Organics with 1.0 psi RVP Margin1
Carbureted Fuel Iniected 1990 Mix
<•24 .0-55. OS +53 .0-76.0%
+6.1-16.8% +0%
~2.5- 8.8% +7.4-11.9%
Total Evap Mass Effect
Evap Ozone Effect (Reactivity Factor = 0.962)
EXHAUST Open Loop Closed Loop 1990 Mix5
Exhaust Mass Effect -3.4%' -0.6% -2.5%
Exhaust Ozone Effect (Reactivity Factor = 1.000) -2.5%
TOTAL OZONE EFFECT8 +14.9-33.6%
2010 Mix '
+33.0-64.0% +46.4-72.9%
+4.4-10.1% +1.6- 2.9%
+4.1-10.2% +6.3-11.5%
~41.5-84.3% +54.3-87.3%
+36.1-77.3% +46.4-77.0%
2010 Mix''
-1.2%
-1.2%
~13.1-22.3%
EVAPORATIVE
RVP
Distillation
Commingling"
Total Organics with Meeting Gasoline RVP Limit
Carbureted Fuel Imected 1990 Mix
2010 Mix
-1.0- 1.6%
+6.5-17.9%
+2.5- 8.8%
-1.6- 2.0%
+ 0%
+7.4-11.9%
Total Evap Mass Effect
Evap Ozone Effect (Reactivity Factor = 0.962)
EXHAUST Open Loop Closed Loop
Exhaust Mass Effect -7.4%7 -4.6%
Exhaust Ozone Effect (Reactivity Factor = 1.000)
TOTAL OZONE EFFECT8
-1.2- 1.8%
+4.7-15.5%
+4.1-10.2%
+7.6-23.9%
~3.5-19.2%
1990 Mix5
-6.5%
-6.5%
-1.4- 1.9%
+1.7- 3.0%
+6.3-11.5%
+6.6-12.6%
+2.5- 8.3%
2010 Mix6
-5.2%
-5.2%
-2.0-( + )5.1% -2 . 9-( -)1.1%
1 Based on actual mass of HC plus ethanol rather than FID measurement.
2 Weights .68/.32 (0 9.0 psi) to .56/.44 (@ 11.68 psi) Carb/FI total evap;
.72/.28 (0 9.0 psi) to .60/.40 (011.68 psi) Carb/FI hot-soak evap;
includes VMT & g/mile weights.
3 Weights .22/.78 (0 9.0 psi) to .14/.86 (0 11.68 psi) Carb/FI total evap;
.26/.74 (0 9.0 psi) to .17/.83 (011.68 psi) Carb/FI hot-soak evap;
includes VMT & g/mile weights.
4 Assumes a 10% alcohol blend market penetration.
5 Based on a .666/.334 OL/CL exhaust weighting including VMT & g/mile.
6 Based on a .200/.800 OL/CL exhaust weighting including VMT & g/mile.
7 Exhaust benefit lower with higher RVP than lower RVP fuels due to increase
in exhaust HC with higher RVP. (A 1.0 psi increase in RVP increases
exhaust HC by about 4%.)
8 Weights .452/.548 evap/exhaust in 1990, .301/.699 evap/exhaust in 2010.
-------
2-153
Table 2-57
EVAPORATIVE
RVP
Distillation
Commingl mg"
Projected Organics and Ozone Impact
of Methanol Blends On Blend Fueled Vehicles
Total Organics with 1.0 psi RVP Margin
Carbureted Fuel Iniected 1990 Mix
2010 Mix
+29.0-55.0% +41.4-62.8%
+1.0- 2.6% +0.4- 0.7%
+15.4-36.0% +23.5-38.3%
+45.4-93.6% +65.3-101.8%
+21.0-47.0% +47.0-65.0%
+1.4- 4.3% +0%
+10.9-34.0% +27.0-39.0%
Total Evap Mass Effect
Evap Ozone Effect (Reactivity Factor=0.851-0.87 8) +23.7-70.0% +40.7-77.2%
EXHAUST Open Loop Closed Loop
Exhaust Mass Effect -3.4%' -0.6%
Exhaust Ozone Effect (Reactivity Factor = 1.000)
TOTAL OZONE EFFECT8
1990 Mix
-2.5%
-2.5%
+ 9.3-30.3%
2010 Mix
-1.2%
-1.2%
+11.4-22.4%
EVAPORATIVE
RVP
Distillation
Commingling4
Total Orqanics with Meeting Gasoline RVP Limit
z 7^7n i
Carbureted
-3.5- 5.4%
+2.5- 8.3%
+10.9-34.0% +27.0-39.0%
Fuel Iniected
-5.4- 6.7%
+0%
1990 Mix
2010 Mix
-4.1- 6.0% -4.9- 6.5%
+1.8- 5.0% +0.6- 1.4%
+15.4-36.0% +23.5-38.3%
+13.1-35.0% +19.2-33.2%
Total Evap Mass Effect
Evap Ozone Effect (Reactivity Factor=0.851-0.878) -3.8-(+)18.5% +1.4-16.9%
EXHAUST Open Loop Closed Loop 1990 Mixs 2010 Mix6
Exhaust Mass Effect -7.4%7 -4.6% -6.5% -5.2%
Exhaust Ozone Effect (Reactivity Factor = 1.000) -6.5% -5.2%
TOTAL OZONE EFFECT8 -5.3-(+)4.8% -3.2-(+)1.5%
1 Based on actual mass of HC plus methanol rather than FID measurement.
2 Weights .68/. 32 (@ 9.0 psi) to .56/.44 (0 11.68 psi) Carb/FI total evap;
.72/.28 (@ 9.0 psi) to .60/.40 (011.68 psi) Carb/FI hot-soak evap;
includes VMT & g/mile weights.
3 Weights .22/.78 (@ 9.0 psi) to .14/.86 (@ 11.68 psi) Carb/FI total evap;
.26/.74 (@ 9.0 psi) to .17/.83 (011.68 psi) Carb/FI hot-soak evap;
includes VMT & g/mile weights.
4 Assumes a 10% alcohol blend market penetration.
5 Based on a .666/.334 OL/CL exhaust weighting including VMT & g/mile.
6 Based on a .200/.800 OL/CL exhaust weighting including VMT & g/mile.
7 Exhaust benefit lower with higher RVP than lower RVP fuels due to increase
in exhaust HC with higher RVP. (A 1.0 psi increase in RVP increases
exhaust HC by about 4%.)
8 .452/.548 evap/exhaust weighting in 1990, .301/.699 evap/exhaust weighting
in 2010.
-------
2-154
Since carbureted and fuel-injected vehicles have different
evaporative emission characteristics, a 1990 and 2010 fleet mix
of these two technologies was used to weight the evaporative
effects, including the relative vehicle-miles-traveled (VMT)
and gram/mile evaporative emissions of the two vehicle types.
Similarly, a 1990 and 2010 light duty fleet mix of open-loop
and closed-loop control systems was used to weight the exhaust
effects. As above, the weighting accounted for both VMT and
exhaust g/mile emissions, thus weighting the contributions by
total mass contribution. The relative VMT and g/mile emissions
were taken from MOBILE3. As a rough estimate of the
open-loop/closed-loop ratio, 1981 and later vehicles were
considered closed-loop, and pre-1981 vehicles were considered
open-loop. The resulting ratio is 0.666 open-loop/0.334
closed-loop in 1990, and 0.200 open-loop/ 0.800 closed-loop in
2010. If light-duty trucks were excluded from the analysis,
open-loop vehicles would account for a smaller fraction of the
total.
Considering the relative contributions of exhaust and
evaporative emissions, these factors translate into a change in
average blend-fueled vehicle total organics emissions, adjusted
for reactivity, of -5.3 percent (methanol blend, matched RVP,
minimum) up to +33.6 percent (ethanol blend, 1.0 psi margin,
maximum) in 1990. Considering the relative contributions of
exhaust and evaporative emissions, in 2010 these factors
translate into a change in average blend-fueled vehicle total
organics emissions of -3.2 percent (methanol blend, matched
RVP, minimum) up to +22.4 percent (methanol blend, 1.0 psi
margin, maximum).
C. Gasoline Distribution System
While this analysis focuses primarily on the reductions in
hydrocarbon emissions (particularly evaporative emissions) from
motor vehicles, the regulation of in-use fuel volatility will
also have an impact on emissions from gasoline-related
stationary sources. As the levels of evaporated hydrocarbons
emitted during the storage, handling, and distribution of
gasoline are a function of true vapor pressure (which in turn
is dependent on the RVP of the fuel), the control of in-use
fuel RVP will also affect the level of emissions from these
sources. Of course, the other primary aspects of this proposal
(changes to certification test fuel volatility and test
procedures) will have no impact on these sources of emissions,
since they involve only changes in the design of new vehicles.
This section discusses the effect of in-use RVP control on
two basic sources of gasoline storage and distribution
emissions: bulk storage and transfer losses (also referred to
in this analysis as the "remainder" category), and service
station underground tank loading losses (generally referred to
as the "Stage I" category). Evaporative emission rates for
-------
2-155
these two sources are commonly expressed in terms of pounds of
hydrocarbon vapor emitted to the atmosphere per thousand
gallons of gasoline stored or transferred (lb/1000 gal).
Vehicle refueling emissions, or "Stage II" emissions, were
treated as a third component of gasoline storage and
distribution losses in EPA's original volatility study.[3] In
this analysis. Stage II emissions have been removed from the
stationary source portion of the emissions inventories and are
treated (along with exhaust and evaporative emissions) as a
mobile source. Emission factors for refueling emissions,
expressed in terms of grams per mile (g/mi), are now
incorporated into EPA's emission factor modeling. The effects
of RVP control on these emissions were discussed earlier in
this chapter, and the resulting emission factors are presented
in Chapter 3.
1. Bulk Storage and Transfer ("Remainder") Losses
This category consists mainly of breathing and working
(loading and unloading) losses resulting from the storage of
gasoline in bulk terminals and the transferring of gasoline to
the ships, tankers, and barges used for transport. The
emission rates associated with bulk storage are dependent on
several factors, including tank configuration (i.e., fixed or
floating roof), tank dimensions, ambient and liquid storage
temperatures, and vapor molecular weight and true vapor
pressure (both of which are dependent on the RVP of the
gasoline). Emissions generated during the loading of carriers
are dependent on the method of filling (submerged or splash),
bulk temperature of the fuel, and the vapor molecular weight
and true vapor pressure. Equations for determining specific
evaporative losses (i.e., breathing, loading) as a function of
these and additional parameters have been developed for various
types of storage and transport media, and are published in
EPA's most recent version of the AP-42 document.[75]
The impact of controlling in-use fuel volatility on rates
of evaporative emissions from the bulk storage and transfer of
gasoline was estimated through the use of the relevant AP-42
equations. Holding the non-fuel related parameters of these
equations constant, emission factors were calculated assuming
RVP levels of 9.0, 10.0, and 11.5 psi. These emission factors
were then used to determine dimensionless adjustment factors,
which in turn were used to adjust the projected emission
inventories for this category, as described below.
First, the ratios of the emission factors calculated for
this category at 9.0 and 10.0 psi RVP to the emission factor at
11.5 psi RVP were determined. These dimensionless adjustment
factors indicated that reducing in-use RVP from 11.5 to 9.0 psi
would result in about a 17 percent decrease in emissions (i.e.,
the ratio of the emission factor at 9.0 psi to the emission
-------
2-156
factor at 11.5 psi is about 0.83). A curve was then fit to the
three points and used to determine adjustment factors at other
RVP levels, as described below.
As discussed in Section I of Chapter 3, EPA used data from
the response to comments on the gasoline marketing study to
establish the base year emission inventories for the bulk
storage and transfer and Stage I categories.[76] The RVP of
in-use fuel in the 1984 base year was assumed to be 11.5 psi in
that document, while the 1983 average in-use RVP (see Section I
of Chapter 3) was determined to be 11.0 psi. Thus, the base
year inventory was adjusted for 11.0 psi RVP fuel, using the
adjustment factor for ll.O psi from the curve described above.
The curve then was normalized to 11.0 (rather than 11.5) psi.
From this point, the adjustment factors taken from this curve
were used to adjust the projected inventories for the
"remainder" category under various RVP control levels. The
specific values of the adjustment factor corresponding to each
RVP control level evaluated are presented in reference[77].
2. Service Station (Stage I) Losses
This category includes the breathing and loading losses
associated with underground storage facilities at service
stations. Losses in this category are commonly referred to as
"Stage I" emissions, designating those emissions generated
during the transfer of fuel between the tank truck and the
service station. Emission rates from these sources are based
primarily on the same parameters as the breathing and loading
losses described in the preceding section.
The method used for developing RVP adjustments for Stage I
emissions parallels that described above for the bulk storage
and transfer category. Based on the ratio of the emission
factor at 9.0 psi RVP to the emission factor at 11.5 psi RVP, a
2.5-psi in-use RVP reduction will result in about a 23 percent
decrease in Stage I emissions. As in the case of the bulk
storage and transfer category, reference [77] presents the
Stage I inventory adjustment factors corresponding to each RVP
control level examined.
Ill. Benzene Emissions from Gasoline-Fueled Vehicles
as a Function of Fuel Parameters.
While the primary emphasis of this study is the effect
gasoline volatility has on ambient ozone, the effect of
benzene, a known carcinogen, is also important. In this
section models are developed to estimate the benzene content of
exhaust and evaporative hydrocarbon emissions for all pertinent
model year (MY) gasoline-fueled vehicles based on the fuel
parameters which can vary with RVP control. The benzene
emission models are made as specific to vehicle type as
possible based on catalyst type, fuel system, and the emissions
-------
2-157
standard under which the vehicle was produced. The benzene
content of refueling emissions is only summarized here as it
has already been addressed in detail in a separate study.[2]
The models developed below focus on the mass fraction of
benzene to total hydrocarbon emissions. This minimizes the
effect of large differences in the mass of benzene emissions
when comparing data from vehicles of different control
technologies and model years, and allows data to be used across
technologies when specific data are limited or non-existent.
Fuel parameters chosen as relevant in the model for
benzene exhaust emissions are the fuel benzene fraction, the
fuel aromatics fraction, and to a lesser degree, the RVP. For
evaporative emissions the fuel benzene fraction and the RVP of
the fuel are the most relevant fuel parameters. For refueling,
only the fuel benzene fraction is a relevant fuel parameter.
An explanation of their selection is contained later in the
discussion.
The following sections describe: 1) available data
presented in the literature, 2) development of the exhaust and
evaporative emission models from that data, 3) comparison of
those models to the additional data to support its validity,
4) the effect of RVP control on fuel composition, and finally,
5) the effect this has on the benzene fraction of exhaust and
evaporative emissions.
A. Literature Review
The information for this analysis was obtained from a
variety of different studies. Ideally, each study would have
provided information on the vehicle model year, catalyst type,
fuel intake system, type of cycle over which emissions were
taken, benzene and aromatic content and RVP of the fuel, and
both benzene and total hydrocarbon emissions. In addition,
each study would have run multiple tests varying independently
the RVP, aromatic content, and benzene content of the fuel for
each test vehicle. Unfortunately, these ideal conditions do
not exist. Many studies did not include pieces of data which
are required for this analysis. For example two studies
[78,79] gave no fuel information whatsoever, and another
study[80] gave no specification for the benzene content of the
fuel, or benzene emissions (merely the benzene fraction of
hydrocarbon emissions). When studies did provide the necessary
information, they failed to perform independent testing of some
or all of the relevant fuel parameters. Two studies [81,82]
performed no independent testing of fuel parameters, and the
remaining three [83,84,85] only performed independent testing
of fuel fraction benzene. Table 2-58 compares qualitatively
the applicability of the different studies reviewed.
-------
Table 2-58
Description of Studies Used
REFERENCE NO.
VEHICLE
INFORMATION
FUEL
SPECIFICATIONS
EMISSION
SPECIFICATIONS
INDEPENDENT
VARIATION OF
FUEL SPfcJUS
ADDITIONAL DESCRIPTION
NO. VEHICLES
CATALYST
description
is
|
1
1
i!
|
ARCMATICS
83
5
Y
8/8
Y
Y
Y
Y
Y
Y
Complete data except aromatic content and
KVP were highly correlated
84,
85
4
Y
2-3
n
Y
Y
Y
Y
Y
Y
Different fuels used on different vehicles;
Aromatic content and RVP highly correlated
81
4
Y
2/2
Y
Y
Y
Y
Y
No independent testing of fuel parameters
82
46
Y
1/2
Y
Y
Y
Y
Y
Oily 1 fuel per vehicle;
No independent testing of fuel parameters
80
15
Y
1/1
Y
Y
Y
Incomplete information;
No independent testing of fuel parameters
78
5
Y
1/1
Y
Y
No fuel information
79
23
Y
1/1
Y
Y
No fuel information
Note: Additional studies listed in reference 86 were not included for
reasons such as:
1) Incomplete infometion
2) Not properly maintained vehicles
3) Non-standard FTP cycle
4) Fuel other than gasoline used
-------
2-159
For exhaust emissions, the best available study was done
for the Coordinated Research Council (CRC) by the National
Institute for Petroleum and Energy Research (NIPER).[83] It
met all of the above mentioned requirements except that it did
not vary RVP and aromatic content independently. Thus, under
the assumption that the aromatic content is the more
significant of the two parameters for exhaust emissions, only
benzene and aromatic content of the fuel were included in this
analysis.
The data by Black and High although meeting the same
requirements as the NIPER data, tested vehicles with different
catalyst designs than NIPER. [ 84,85 ] Thus the two sets of data
could not be combined for modeling purposes. In addition,
Black and High tested no different vehicles with different
fuels introducing a high degree of vehicle to vehicle
variability into the data causing a separate modeling analysis
on this data to be impractical. Thus the following model was
developed using the NIPER data, and data from the other sources
were used to evaluate its wider applicability.
For evaporative emissions, only two studies quantified
benzene emissions. This provided a very small data base from
which to work, and was further complicated by the study done by
NIPER which showed a severe memory, or "carry-over" effect of
benzene emissions from test to test when different fuels were
used in successive tests.[83] For example, following a test
conducted with a high benzene content fuel, a test with a low
benzene content fuel will show higher than normal benzene
emissions due to the canister still being loaded with large
amounts of benzene from previous tests (see Figure 2-21). This
"carry-over" effect suggests that the accuracy of any recorded
data from other experiments may also be in error. However,
after determining the existence of the carry-over problem,
NIPER did a follow-up experiment involving four vehicles and
two fuels for which they performed repetitive tests until the
carry-over effect was eliminated. Unfortunately, one of the
fuels contained no benzene (predictively producing no benzene
emissions), and two of the vehicles had unrepresentatively low
evaporative emissions compared to those described in Section II
(i.e., extremely low diurnal emissions and low hot-soak
emissions). This left only two tests on which benzene
evaporative emissions could be modeled, one test of a
carbureted vehicle, and the other test of a fuel-injected
vehicle.
The data obtained by Black and High [84,85] could only be
used as a check of the NIPER-based model for the carbureted
vehicle, recognizing that the accuracy of the data is
questionable due to the potential presence of carry-over.
Black and High tested no fuel-injected vehicles. Therefore,
the model for fuel-injected vehicles could not be verified.
-------
2-160
Fig. 2-21- 8enzene evaporative emissions as a function
of fuel ben2ene level and number of test repetitions
(4-veh1cle average using M08ILE3 equation) [83]
10
8 -
8 -
4 -
2 -
BENZENE VOLUME PRECZNT IN RJEL
(In order of test number)
BENZENE VOLUME PERCENT IN FUEL
(In order of test number)
-------
2-161
B. Exhaust Emission Analysis
1. Selection of Relevant Fuel Parameters
Two fuel parameters are thought to have a significant
effect on the benzene content of exhaust emissions; the benzene
content, and the aromatic content of the fuel. The effect of
the RVP will also be discussed below since that is the primary
fuel parameter being addressed in this study.
The effect of benzene content is understandable since any
benzene not combusted in the engine, nor subsequently oxidized
by the catalyst, will be emitted as benzene. Therefore, the
more benzene in the fuel, the greater the amount of benzene
expected in the exhaust.
The effect of aromatic content is also understandable
since the more complex aromatics in the fuel may combust
incompletely leaving benzene as a product. Therefore, the
higher the aromatic content of the fuel, the greater the amount
of benzene expected in the exhaust. Other compounds could
possibly recombine to produce benzene, but none have yet been
specifically identified for use in a quantitative analysis.
The effect of RVP on benzene emissions is not as clear as
the previous two fuel parameters. A decrease in RVP is
essentially a measure of a decrease in the butane content of
the fuel. This has been shown in Section III to result in
decreased hydrocarbon emissions. As discussed in Chapter 4,
this effect is due primarily to periodic enrichment of the
fuel/air mixture when hydrocarbon vapors are purged from the
canisters. This enrichment would not be expected to have any
effect on the benzene fraction of hydrocarbon emissions only on
the total amounts of both benzene and non-benzene emissions.
Any other effects should be small relative to the effect of
aromatics. Thus, without data to prove otherwise, the effect
of RVP has to be ignored here, except for its effect on the
benzene and aromatic content of gasoline.
2. Model Development and Analysis
As discussed earlier, the data from the NIPER study for
1984-85 MY vehicles, presented in Table 2-59, were used for the
model development. A statistical analysis of the data
presented with Figure 2-22 showed a significant difference
between the benzene fraction of hydrocarbon emissions for
vehicles with three-way (3w) catalysts and those with 3-way
plus oxidation (3wox) catalysts. This is reasonable since the
addition of the extra oxidation catalyst and the excess air
could preferentially reduce various individual hydrocarbons.
For this reason, the data was divided into two groups; vehicles
with 3w catalysts, and vehicles with 3wox catalysts. Linear
-------
Fuel
1
2
3
4
5
6
7
8
Note
2-162
Table 2-59
NIPER Data Used for Exhaust Emission Modeling Analysis[83]
BzEx
FArom 3w cat 3wox cat
RVP
FBz
(no Bz)
CO
I
85 Chevy 85 Chevy
85 Ford 85
Chrys
10.8
0 . 03
25 . 7
3.9
1. 7
1. 4
1.3
2 . 0
10 . 6
1. 50
25.6
5.2
2.8
2 . 8
2. 1
2 . 5
10 . 6
2. 72
25 . 2
6.3
3 . 9
3 . 1
3.2
2 . 6
10 . 4
3 . 99
24 .9
7 . 8
5 . 1
4 . 2
3.9
3.8
10 . 1
0 . 03
40 . 1
6 . 2
3 . 8
2 . 1
2.5
2 . 6
9 . 8
1 . 48
39 . 4
7 . 9
4 . 9
2 . 9
3.3
3 . 3
9 . 8
2. 70
38.9
9 . 6
5.8
4 . 0
3.4
4.3
9 . 8
3 . 90
38 . 2
10 . 2
7 . 5
5 . 5
5 . 8
4 . 5
BzEx
= Weight percent benzene in the
exhaust
FBz
= Fuel
volume
percent
benzene
FArom
= Fuel
volume
percent
aromatics
excluding benzene
-------
2-163
Fig 2-22 Plot of 3way catalysts vs 3wox catalysts [83]
1/ For the above data
W%BzEx W%BzEx
3W cat 3W0X cat
x = 5.79 x = 3.21
Sx2 = 5.61 Sx2 = 1.31
*~2 = ~^—5 7T7 ^ . i/7 = 4.61 (Signal)
pn^DSy2 +(ny-l)Sy2 p 2 (Vnx + 1/ny) 2
L nx + riy - 1 J
for V = 16 + 24 - 2 = 38
C.L. = 99%
°<=(100%-C.L.)/2 = 0.005
Therefore: t, = 2.713 (Noise)
Therefore we can be atleast 99% confident that there is a
difference between the two types of catalysts. The t test dene
above is also a worst case situation, since by drawing the average
through the data as seen above, you introduce additional
variation due to the slope of the lines.
-------
2-164
regressions were performed on each set of data with the
following results:
BzEx(3w) =1.0306(FBz)+0.17588(FArom)-l.9785 (R2=.996) (1)
BzEx(3wox)=0.6796(FBz)+0.06807(FArom)-0.3468 (R2=.965) (2)
Where: BzEx = weight percent benzene in the exhaust
FBz = volume percent benzene in the fuel
FArom = volume percent aromatics in the fuel
excluding benzene
These equations were then applied to the test conditions
in the other relevant studies[81,82] and the results compared
to the actual data (see Table 2-B-l in Appendix 2-B) . As can
be seen, these models give good correlation with the data for
vehicles equipped with 3w and 3wox catalysts. However, neither
equation seemed to accurately describe emissions for cars with
only oxidation catalysts, or cars meeting pre-1980 emission
standards. Instead, it was found that for these vehicles the
average of the two equations gave the best correlation, as
shown in Table 2-60 (and Table 2-B-l in Appendix 2-B).
Therefore, for pre-1980 cars, or cars after 1980 equipped only
with oxidation catalysts, the equation
BzEx = .85511(FBz)+.12198(FArom)-1.1626 (3)
was developed. A linear regression was not performed on the
additional data to arrive at this or a similar equation for
oxidation catalyst equipped vehicles because of the small
number of fuels used in the other studies, and the lack of
independence between fuel parameters.
C. Evaporative Emission Analysis
Data on the benzene content of evaporative emissions
presented in Table 2-61, were limited to only one test each for
carbureted and fuel-injected vehicles making any linear
regression analysis impossible. Theoretical analysis was the
only viable alternative remaining.
Two fuel parameters have a significant effect on the
benzene content of evaporative emissions; the benzene content
of the fuel, and RVP. With respect to the benzene content of
the fuel, the vapor pressure of benzene is proportional to its
level in the liquid fuel, so benzene emissions to the
evaporative canister should be proportional to fuel benzene
content. Any preferential treatment of benzene by the charcoal
canister should be relatively consistent between fuel benzene
levels in the range of consideration, so benzene emissions from
the canister should also be proportional to fuel benzene
content.
-------
2-165
Table 2-60
Model Comparison for Ox-Cat and Pre-80 Cars
Actual Predicted Predicted Predicted
w% BzEx 3W EQN 3WOX EQN Avq EQN
X 3.61 4.64 2.72 3.68
S„ 1.516 1.497 .861 1.172
% error 28.5% 24.7% 1.9%
-------
2-166
Table 2-61
NIPER Data Used for Evaporative Emission Modeling Analysis[83]
Vehicle Cat Fuel Syst RVP FV% Bz V% Arom W% Bz/HC
85 Ford 3wox TBI 10.8 5.22 19.5 1.91
85 Plymouth 3wox CARB 10.8 5.22 19.5 6.36
-------
2-167
The effect of RVP, however, is more complicated. First,
the fuel benzene content (and, thus, the vapor pressure of
benzene) will tend to decrease with increased RVP (i.e.,
increased fuel butane content), but this is taken into account
through the effect of fuel benzene content, as described
above. In addition, increased RVP will increase the total
amount of evaporative emissions. This increase will be
primarily butanes. However, it will also include all other
compounds to some degree, since the additional butane vapor
generated from the temperature increase will increase the
emission of all components in the vapor. This effect may
differ for carbureted and fuel-injected vehicles since the
ratio of hot-soak to diurnal emissions from a fuel-injected
vehicle is much lower than from a carbureted vehicle, and RVP
has a stronger effect on diurnal emissions than on hot-soak
emissions.
No data are available to derive an empirical model.
Derivation of theoretical relationships is hampered by the
unknown effect of large increases in butane emissions on the
canister's ability to adsorb benzene. Thus, two different
models are proposed below. The first model is designed to
represent a lower bound estimate of the effect of RVP on
benzene emissions. The second represents an upper bound
estimate.
The first model assumes that evaporative benzene emissions
are directly proportional to the benzene content of the fuel,
and inversely proportional to RVP:
EvBz = C(FBz)/RVP
Where: RVP = Reid vapor pressure
C = constant
Other notation as previously described.
When this model is applied to the data from the NIPER
study, [83] which are shown in Table 2-61, the equations which
result for both fuel-injected and carbureted vehicles are as
follows.
EvBz = 3.952(FBz)/RVP (Fuel-Injected) (4)
EvBz = 13.159(FBz)/RVP (Carbureted) (5)
In an effort to substantiate equations 4 and 5, the
additional data by Black and High were used (see Table 2-B-3 in
Appendix 2-B).[84,85] Equation 5 was found to give reasonable
correlation at RVP levels below 10 psi, but significantly
overestimated the benzene emission fraction at higher RVP
levels. No data were available for fuel-injected vehicles to
evaluate equation 4.
-------
2-168
Based on observation of trends in the Black and High data,
a second model was developed which is more sensitive to RVP
levels:
EvBz = A(FBz)(B-RVP)
Where: A and B are constants.
With this model, for the carbureted engine, an iterative
procedure comparing values obtained from the model and the
NIPER data with those obtained from the Black and High data
gave a value of 0.5 for A and 13.23 for B. This correlates
well with the Black and High data, but it must be remembered
that the level of "carryover" effect in the Black and High data
is unknown.
For fuel-injected vehicles, assuming B remains the same as
in the carbureted case, A equals 0.151. The assumption that B
is the same is based on the belief that the effect of RVP on
evaporative emissions is the same (proportionally) for both
fuel-injected and carbureted vehicles. As indicated above,
this is probably not precisely the case. However, the lack of
additional data for fuel-injected vehicles prevents derivation
of a new value for B and the degree of error should hopefully
be small. The equations based on the more RVP-sensitive model
are then:
EvBz = 0.5(FBz)(13.23-RVP)* (carbureted) (6)
EvBz = 0.151(FBz) (13 .23-RVP)* (fuel-injected) (7)
It is believed that these last equations may be more
representative of actual emissions than equations 4 and 5 since
the added sensitivity to RVP seems to allow the model to more
closely predict trends observed in the Black and High data.
However, due to the extremely small amount of data on which the
models are based, the approach taken here is to use all four
equations, two for carbureted vehicles, and two for
fuel-injected vehicles to establish a range of projected
emissions to allow for the high degree of uncertainty.
Since the only usable data were for recent model year
vehicles, it is assumed that the benzene mass fraction of
evaporative hydrocarbon emissions remains relatively constant
for all vehicles. Therefore, any changes in benzene emissions
by model year and standard under which the vehicles were
produced will be compensated for by the change in quantity of
hydrocarbon emissions for which there is more data available.
These equations are limited to RVP levels below 13.23 psi.
-------
2-169
D. Refueling Emission Analysis
In a seperate study performed by EPA on the benzene
fraction of refueling emissions, three parameters were found to
have a significant effect: the fuel benzene content, the
dispensed fuel temperature, and the temperature difference
between the dispensed fuel and the m-tank fuel. [87] RVP was
not found to be significant. This is understandable since
refueling emissions are primarily displaced vapors from the
fuel tank and the vapor pressure of benzene is solely a
function of fuel benzene content and not RVP. The evaporation
process which is enhanced by high RVP does not come into play.
The resulting equation derived from the analysis taking
into account fuel spills during refueling is:
ReBz = 0.0451 (FBz) - 1.60 x 10~" (TD) - 4.24 xlO~" (AT)
Where: ReBz = grams benzene emitted during refueling per
gallon of fuel dispensed.
FBz = Volume percent benzene in the fuel
Td = Temperature of the dispensed fuel
AT = Temperature difference between dispensed and
in-tank fuel.
The nationwide average temperature of the dispensed fuel is
68.9 °F while the average temperature difference between
dispensed and in-tank fuel is 4.4°F. Therefore, the above
equation is reduced to:
ReBz = 0.0451 (FBz) - 0.01289 (8)
The fuel fraction benzene is the only true fuel parameter
affecting the benzene refueling emissions. However, with RVP
control, the benzene fuel fraction will change slightly. For
this reason RVP control will affect the benzene fraction of
refueling emissions.
E. Effect of RVP control on Fuel Composition
In-use control of RVP will be achieved primarily by
reducing the quantity of butane in the gasoline pool (roughly
two percent per 1.0 psi RVP reduction or up to seven percent
for a 3.5 psi RVP reduction). If the quantity of butane in the
fuel is reduced without any other compositional changes, the
percentage of all other fuel constituents, including aromatics,
would be expected to increase proportionately (i.e., by up to a
factor 1.07). However, other compositional changes are
expected which will either heighten or mitigate this general
increase. No data are available concerning the effect of RVP
control on benzene levels, in particular, but projections are
available from the Bonner and Moore model of total fuel
-------
2-170
aromatic content for the baseline and control cases allowing
investment and open NGL purchases (see Table 2-62). As can be
seen, the nationwide average aromatic content increases, by 0.6
percent and 2.8 percent for RVP reductions of 1 psi and 2 psi,
respectively, over the baseline level of 32.5 volume percent.
Assuming benzene levels increase proportionately, then current
benzene levels (roughly 1.34 percent based on the 1984 NIPER
survey[88]) would increase by as little as 0.01 volume percent
with a 1.0-1.5 psi RVP reduction from 11.5 RVP, to as high as
0.16 volume percent with a 3.5 psi RVP reduction to 8.0 (points
below 9.5 psi are extrapolated from a plot of the data, and
other points are interpolated).
F. Results
Using the equations and the estimates for fuel composition
developed above for the current base case and future controlled
RVP case, the resulting benzene emission fractions are shown in
Table 2-63. The increases in benzene and aromatic content with
maximum RVP control (i.e., 8 psi) cause a 15.2 percent increase
in the benzene fraction of exhaust emissions. The benzene
fraction of evaporative emissions, however, as they are also
dependent on RVP, more than doubled. In the case of
fuel-injected vehicles, however, this is still only a slight
increase in absolute magnitude. Refueling emissions which are
only affected by the change in fuel benzene content experienced
only a 15 percent increase. These emission fractions will be
used in Chapter 3 to estimate mobile source benzene emissions
and related cancer incidences under the various regulatory
scenarios.
-------
2-171
Table 2-62
Changes in Aromatic Content (Vol %)
Resulting from RVP Control'
With Investment, Open NGLs
PADD
Baseline
Unleaded Regular
Unleaded Premium
Leaded Regular
Weighted Average4
1 PSI RVP Reduction
Unleaded Regular
Unleaded Premium
Leaded Regular
Weighted Average
2 PSI RVP Reduction
Unleaded Regular
Unleaded Premium
Leaded Regular
Weighted Average"
38 8
31 . 6
30 .8
36. 1
37 .9
35.8
31 . 0
36.3
38.3
36.0
31. 1
36.7
4 + 5
(ex. CA)2
32 . 0
37 . 9
31 . 2
32.9
33 . 1
36.9
31.3
33 . 5
33.6
35.6
31 . 5
33.6
31.3
31 . 0
34 . 6
31.8
31.2
31.2
33 . 6
31 . 6
32 . 4
34 . 5
32 .2
32.7
31 . 7
34 . 5
32.9
32 .4
32
34
32
32
33
35
31
33
Total U.S
(ex. CA)3
32 . 2
33.3
33 . 1
32 . 5
32
33
32
32
33 . 3
35.0
31 . 9
33 . 4
1 As predicted by Bonner and Moore RPMS model.
2 Estimated as an average of PADDs 2 and 3.
3 Total U.S. (ex. CA) estimated by volumetrically weighting
of PADD aromatic content for PADDs 1,2,3,4 and 5 using
PADD specific gasoline production.
4 Average estimated by weighting the three gasoline grades
by % of sales by volume.
-------
2-172
Table 2-63
Benzene Fractions for Base and RVP Control Cases
RVP
Fuel V%
Benzene
11.5(Base) 1.34
Fuel V%
Aromatics
32. 5
Model**
A
B
C
g Bz/g HC g Bz/g HC
Exhaust
.0512
.0278
.0395
Evap
.0116-.0153
.0035-.0046
g/Bz/gal Fuel
Refueling
.0475
11.0
1. 34
32 . 5
A
B
C
.0512
.0278
.0395
.0149-.0160
,0045-.0048
,0475
10. 5
1.35
32 . 6
A
B
C
.0515
.0279
.0397
.0169-.0184
.0051-.0056
.0480
10.0
1.35
32 . 8
A
B
C
.0518
.0280
,0399
.0178-.0218
.0053-.0066
.0480
9.5
1. 37
33.2
A
B
C
.0527
.0284
.0406
.0190-.0256
.0057-.0077
.0489
9.0
1.39
33.6
A
B
C
.0536
.0289
.0412
,0203-.0294
,0061-.0089
.0498
8.5
1.43
34.6
A
B
C
.0558
.0298
.0428
.0221-.0338
,0066-.0102
.0516
.0
1. 50
36.3
A
B
C
.0595
,0314
,0455
,0247-.0392
,0074-.0118
,0548
Based on data from Table 2-62, these data are an extrapolation to
8.0 psi.
For exhaust: Model A = vehicles with 3-way catalysts
Model B = vehicles with 3-way plus oxidation
catalysts
Model C = vehicles Pre-1980 or Post-1980 vehicles
with oxidation catalysts
For evap: Model A = carbureted vehicles
Model B = fuel-injected vehicles
-------
2-173
References (Chapter 2)
1. "Refueling Emissions from Uncontrolled Vehicles,"
Dale Rothman and Robert Johnson, EPA-AA-SDSB-85-6, August
1985. (Available in Public Docket No. A-84-07.)
2. "Evaluation of Air Pollution Regulatory Strategies
for Gasoline Marketing Industry," Office of Air Quality
Planning and Standards and Office of Mobile Sources,
EPA-450/3-84-012a, July 1984. (Available in Public Docket No.
A-84-07.)
3. "Study of Gasoline Volatility and Hydrocarbon
Emissions from Motor Vehicles," U.S. EPA, OAR, OMS, ECTD,
EPA-AA-SDSB-85-0 5, November 1985.
4. "1982-84 Ozone Design Values for Regulatory Impact
Analyses," EPA Memorandum from Richard G. Rhoads, MDAD, to John
R. O'Connor, SASD and Charles Gray, ECTD, June 16, 1986.
5. "Standard Metropolitan Statistical Areas (SMSA)
Regulatory Analysis Air Quality Data Base, 1982-84," U.S. EPA,
OAR, OAQPS, Monitoring and Data Analysis Division, February
1986 .
6. "Control Characteristics of Carbon Beds for Gasoline
Vapor Emissions," EPA-600/2-77-057, Michael J. Manos and Warren
C. Kelly, Scott Environmental Technology, for EPA, ORD, IERL,
February 1977.
7. "Combustion Engine Economy, Emissions and Controls,
July 9-13, 1984," Engineering Summer Conferences, The
University of Michigan College of Engineering.
8. "Hydrocarbon Control for Los Angeles by Reducing
Gasoline Volatility," Edwin E. Nelson, Engineering Staff, GM
Corp., SAE Paper No. 690087, 1969.
9. "Effect of Fuel Front-End and Midrange Volatility on
Automobile Emissions," B. H. Eccleston and R.W. Hurn,
Bartlesville Energy Research Center, U.S. DOI, Bureau of Mines,
Report of Investigations 7707, 1972.
10. "Effect of Fuel Composition on Amount and Reactivity
of Evaporative Emissions," M.W. Jackson and R.L. Everett, SAE
Paper No. 690088, 1969.
11. "Mathematical Models for Prediction of Fuel Tank and
Carburetor Evaporation Losses," W.J. Koehl, Jr., Research
Dept., Mobil Research and Development Corp., SAE Paper No.
690506, May 1969.
-------
2-174
12. "Mathematical Expressions Relating Evaporative
Emissions from Motor Vehicles without Evaporative Loss-Control
Devices to Gasoline Volatility," William F. Biller, Michael
Manoff, Jyotin Sachdev, and William C. Zegal, Scott Research
Laboratories, Inc., and David T. Wade, Esso Research
Engineering Co., SAE Paper No. 720700, 1972.
13. "Clean Air Act Waiver Application, Section 211(f),
Volume 2," E.I. du Pont de Nemours and Company, Inc., July 11,
1984. (Available in Public Docket No. EN-84-06.)
14. "Environmental Impacts of Methanol/Gasoline Blends,"
prepared for Air Pollution Control Association by Tom Cackette
and Thomas Austin, February 16, 1984.
15. Letter from Dale F. Pollart, Texaco, Inc., to
Richard Wilson, EPA, dated May 13, 1985.
16. "Analysis of Fuel Volatility Characteristics and
Evaporative Hydrocarbon Emissions for Alcohol/Gasoline Blends,"
PLMR-43-83, E.I. du Pont de Nemours and Company, Inc., June 29,
1983 .
17. "Standard Specification for Automotive Gasoline,
D-439-83," American Society for Testing and Materials (ASTM).
18. "MVMA National Gasoline Survey, Summer Season,
1984," Motor Vehicle Manufacturers Association, Inc., October
15, 1984.
19. "Motor Gasolines, Summer 1985," Cheryl L. Dickson
and Paul W. Woodward, National Institute for Petroleum and
Energy Research (NIPER), for API, June 1986.
20. "Alcohol Outlook," May 1985.
21. "Performance Evaluation of Alcohol-Gasoline Blends
in 1980 Model Automobiles, Phase II - Methanol-Gasoline
Blends," prepared for DOE by CRC, January 1984.
22. "Physical Properties of Gasoline/Alcohol Blends,"
Frank N. Cox, U.S. DOE, Bartlesville Energy Technology Center,
September 1979.
23. "Final Report, Technical Directive #8 to EPA
Contract 68-03-3230, Effects of RVP and Temperature on 2 gm and
6 gm Vehicles," Automotive Testing Laboraties for EPA, October
1986.
24. "Exhaust and Evaporative Emissions of High Mileage
Taxicabs and Passenger Cars," EPA Technical Report, Craig A.
Harvey, Project Officer, February 1985.
-------
2-175
25. API presentation to EPA on "1984 Evaporative
Emissions Research Programs, Gasoline Volatility Assessment
Task Force," contractors: NIPER and ATL, 1985.
26. Letter from J.S. Welstand, Chevron Research
Division, to Chester J. France, EPA, April 1, 1985.
27. "Motor Vehicle Tampering Survey - 1982," EPA, OMS,
FOSD, EPA-330/1-83-001, April 1983.
28. "Motor Vehicle Tampering Survey - 1983," EPA, OMS,
FOSD, (no report number given) August 1984.
29. "Motor Vehicle Tampering Survey - 1984," EPA, OMS,
FOSD, (no report number given) Sept. 1985.
30. "Motor Vehicle Tampering Survey - 1985," EPA, OMS,
FOSD, (no report number given) November 1986.
31. "Report for Third Quarter, FY 85, In-Use Vehicle
Testing in Ann Arbor," EPA memorandum from Howard
Brasher-Frederick to Charles L. Gray, EPA, OMS, ECTD, July 23,
1985.
32. "Factors Influencing Vehicle Evaporative Emissions,"
D. T. Wade, SAE paper 670126, 1967.
33. "Evaporative Emissions from Gasolines and Alcohol
-Containing Gasolines with Closely Matched Volatilities," S. R.
Reddy, SAE paper 861556, 1986.
34. Letter from S. W. Martens, Environmental Activities
Staff, GM, to Phil Carlson, EPA, September 30, 1986.
35. "Comparison of Diurnal Mass Emissions From Fuel
Weathering Program to MOBILE3 Diurnal Emissions," Note from Tom
Darlington to Charles Gray, Director of Emission Control
Technology Division, October 22, 1985.
36. "A Service Station Test of a Vapor Balance System
for the Control of Vehicle Refueling Emissions," A.M.
Hochhauser and L.S. Bernstein, Exxon Products Research
Division, Exxon Research and Engineering Company, July 1, 1976.
37. "Efficiency Evaluation of UNION 76 Balance-Type
Stage II Vapor Control System with OPW 7VC Dispensing
Nozzles-Attendant Serve Operating Mode," Scott Environmental
Technology, Inc., August 30, 1979.
38. "Efficiency Evaluation of Healy Vapor Control
System," Scott Environmental Technology, Inc., October 30, 1981.
-------
2-176
39. "Consumer Refueling Characteristics," Michael
Lombardo, SAE paper 871085, 1987.
40. "General Motors Written Commentary on the U.S.
Environmental Protection Agency 'Study of Gasoline Volatility
and Hydrocarbon Emissions from Motor Vehicles'," Environmental
Activities Staff, General Motors Corporation, FE-3693, March
26, 1986.
41. "Fuel Volatility Trends," Southwest Research
Institute, EPA Contract No. 68-03-3192, Work Assignment 4, Task
3, Final Report dated September 28, 1984.
42. "Statistical Abstract of the United States, National
Data Book and Guide to Sources, 103d Edition, 1982-83," U.S.
Department of Commerce, BUREAU OF THE CENSUS.
43. "Local Climatological Data Monthly Summary, July
1985," for various locations. National Oceanic and Atmospheric
Administration.
44. "Effects of Distillation Point at 160°F on Hot Soak
Emissions," EPA Memo form Jonathan Adler, TSS, to Charles Gray,
ECTD, September 23, 1986.
45. "Final Report, Technical Directive No. 5 to Contract
68-03-3230, Effects of Distillation Characteristics at Fixed
RVP," Automotive Testing Laboratories for EPA, April 15, 1986.
46. "Relationship Between Exhaust Emissions and Fuel
Volatility," EPA memo from Thomas L. Darlington to Charles L.
Gray, Jr., EPA, OMS, ECTD, June 24, 1985.
47. Final Weekly Report; "Task 1 - EPA/ATL Correlation/
Temperature Effects," EPA Motor Vehicle Emission Laboratory,
Ann Arbor MI, June 20, 1986.
48. Letter to API from Exxon Research and Engineering
Company, March 19, 1986.
49. "Effect of Auxiliary Cooling on Fuel Tank
Temperatures," EPA Memorandum from Edward Barth to Robert
Maxwell, February 21, 1986.
50. Emission Factor Data Base, EPA Motor Vehicle
Emission Laboratory, Ann Arbor MI.
51. "Draft Regulatory Impact Analysis, Proposed
Refueling Emission Regulations for Gasoline-Fueled Motor
Vehicles, Volumes l and 2," U.S. EPA,OAR,OMS, (Available in
Public Docket No. A-87-11), July 1987.
5la. "Effect of Fuel Volatility and Methanol Blend Usage
on Evaporative Emissions," Peter Gabele Presentation to June
1987 APCA meeting in New York City.
-------
2-177
52. "Performance Evaluation of Alcohol-Gasoline Blends
in 1980 Model Automobiles: Phase 1 - Ethanol-Gasoline Blends,"
CRC Report No. 527, Systems Control, Inc., prepared in part by
Coordinating Research Council, Inc., Atlanta, GA, July 1982.
53. "Gasohol Test Program," Richard D. Lawrence,
EPA-AA-TAEB-79-4B, Testing and Evaluation Branch, U.S. EPA,
February 1980.
54. "Gasohol Test Program," Frank Black, Mobile Source
Emissions Research Branch, U.S. EPA, Research Triangle Park,
NC, November 19 78.
55. "Evaporative and Exhaust Emissions from Cars Fueled
with Gasoline Containing Ethanol or Methyl Tert-Butyl Ether,"
R. L Furey. and J.B. King, SAE Paper 800261, 1980.
56. "Exhaust and Evaporative Emissions from Alcohol and
Ether Fuel Blends," T. M. Naman and J.R. All sup, SAE Paper
800858, June 1980.
57. "Revised Final Data Report on Work Assignment No.
12, EPA contract 68-03-3192, Blend Vapor Analysis," Lawrence
Smith, Southwest Research Institute, letter to Craig Harvey,
April 11, 1986.
58. "Low Temperature Vapor Generation," Lawrence Smith,
Southwest Research Institute, draft final report for Work
Assignment No. 18, EPA contract 68-03-3192, August 1986.
59. "Characterization of Emissions from Vehicles Using
Methanol and Methanol-Gasoline-Blended Fuels," P.A. Gabele,
J.O. Baugh, F. Black, R. Snow, JAPCA Vol. 35, no. 11, 1985.
60. "Gasoline Headspace Over Oxygenated Fuels," L.A.
Rapp, ARCO internal memos/progress reports, subject 5149,
April-December 1983.
61. "Additional Information Regarding DuPont's
Application for a Waiver Covering Gasoline-Alcohol Fuels,"
letter from Edward Cantwell, DuPont, to Sylvia Correa, FOSD,
U.S. EPA, October 11, 1984.
62. "Twenty-three Car In-House Oxinol Blend Test
Program," EPA memo from Craig Harvey to Charles Gray, November
19, 1984.
63. "Commingling of Gasolines with Methanol/Gasoline
Blends," Charles T. Hare, Southwest Research Institute Monthly
Overall Progress Report No. 15 of EPA Contract No. 68-03-3192,
January 10, 1985.
-------
2-178
64. "Meeting Confirmation," letter from W.J. Wostl,
ARCO, to Charles Gray, February 4, 1985. (Available in Public
Docket No. EN-84-06)
65. "Analysis of Gasohol Fleet Data to Characterize the
Impact of Gasohol on Tailpipe and Evaporative Emissions," EPA
Mobile Source Enforcement Division report, December 1978.
66. "Volatility Characteristics of Gasoline-Alcohol and
Gasoline-Ether Fuel Blends," Robert L. Furey, SAE Paper 852116,
October 1985.
67. "Ozone Formation Potential of Organic Compounds,"
Joseph J. Bufalini, T. A. Walter, and M. M. Bufalini,
Environmental Science and Technology, Vol. 10, page 908,
September 1976.
68. "Impact of Methanol on Smog: A Prelimianary
Estimate," Gary Whitten and Henry Hogo, Systems Applications,
Inc. report for ARCO Petroleum Products Company, February 1983.
69. "Photochemical Modeling of Methanol-Use Scenarios in
Philadelphia," G.Z. Whitten, N. Yonkow, and T.C. Myers, Systems
Applications, Inc., EPA Report 460/3-86-001, March 1986.
70. "Effects of Methanol Fuel Substitution on Multi-Day
Air Pollution Episodes," P.L. Carter, et al, Statewide Air
Pollution Research Center, Riverside, final report for
California Air Resources Board Contract No. A3-125-32,
September 1986.
71. ARCO Oxinol waiver application, April 27, 1981.
(Available in Public Docket No. EN-81-10.)
72. "Further Analyses of the Effects of Fuel Oxygen
Content on Exhaust Emissions," memo from Jonathan Adler to Phil
Lorang, EPA/OMS/ECTD Technical Support Staff, January 7, 1987.
73. "Exhaust Emissions and Fuel Economy from Automobiles
Using Alcohol/Gasoline Blends Under High-Altitude conditions,"
TEB-79-1, Testing and Evaluation Branch, ECTD, U.S. EPA, NTIS
No. PB 290612, October 1978.
74. "Gasohol, TBA, MTBE Effects on Light-Duty
Emissions," Bruce Bykowski, Southwest Research Institute, EPA
Contract 68-03-2377 report, EPA 460/3-79-012, October 1979.
75. "Compilation of Air Pollution Emission Factors,
Volume I: Stationary Point and Area Sources," (AP-42), U.S.
EPA, OAR, OAQPS, Fourth Edition - September 1985.
76. "Evaluation of Air Pollution Regulations Strategies
for the Gasoline Marketing Industry — Response to Public
Comments," U.S. EPA, OAR, OAQPS and OMS, July 1987 (Available
in Public Docket No. A-87-11.).
-------
2-179
77. "1983 NEDS Inventory Modifications and Inventory
Projection Methodology in Support of Volatility NPRM," EPA
Memorandum from Mark Wolcott, TEB, to Terry Newell, SDSB, July
2, 1987.
78. "Benzene in Auto Exhaust," George J. Nebel, APCA
Journal Vol 25, No. 4, April 1979.
79. "Mobile Source Benzene," EPA Memorandum from C.L.
Gray, Director, ECTD to D.R. Goodwin, Director, ESED, November
1, 1982.
80. "Automotive Hydrocarbon Emission Patterns in the
Measurement of Nonmethane Hydrocarbon Emission Rates," F.M.
Black and L.E. High, SAE 770144, 1977.
81. "Impact of Low Ambient Temperature on 3-way Catalyst
Car Emissions," James N. Braddock, SAE 810280, February 1981.
82. "Volatile Organic Compound Emissions from 46 In-Use
Passenger Cars," John E. Sigsby, Silvestre Tejada, William Ray,
EPA and John Lang, John Duncan, Northrop Services, Inc., 1983.
83. "Vehicle Evaporative and Exhaust Emissions as
Influenced by Benzene Content of Gasoline," National Institute
for Petroleum and Energy Research, April 1986.
84. "Composition of Automotive Evaporative and Tailpipe
Hydrocarbon Emissions," F.M. Black and L.E. High, Journal of
the Air Pollution Control Association, V30, No. 11, November
1980.
85. "Passenger Car Hydrocarbon Emissions Specification,"
F.M. Black and L.E. High, EPA-600/2-80-085, May 1980.
86. "Analysis of Benzene Emissions from Vehicles and
Vehicle Refueling," American Petroleum Institute, SAE 841397,
1984.
87. "Factors Influencing Benzene Emissions from
Passenger Car Refueling," Paul M. Laing, SAE 861559, 1986.
88. "Motor Gasolines, Summer 1984," Ella Mae Shelton and
Cheryl L. Dickson, National Institute for Petroleum and Energy
Research (NIPER), for API, February 1985.
89. "Evaporative HC Emissions for MOBILE3," EPA
Technical Report No. EPA-AA-TEB-85-1, U.S. EPA, OAR, OMS, ECTD,
TEB, August 1984.
-------
Appendix 2-A
Breakdown of Motor Vehicle Evaporative
Emission Factors into Their Components
The evaporative emission factors used in this analysis
were derived from the results of EPA's m-use emission factor
(EF) test program. From July 1984 until March 1986, 321
vehicles were tested under this program. These vehicles were
tested on: 1) commercial fuel with a nominal RVP of 11.7 psi,
2) a blended fuel with an RVP of 10.4 psi, and 3) Indolene
fuel, with an RVP of roughly 9.0 psi, in that order. The
complete test procedure was summarized in Table 2-10 under the
heading of Post-July 1984.
The vehicles in the EF testing program have been separated
by the condition of the vehicle, and by type of fuel metering
system. Those vehicles having evaporative emission control
malfunctions considered to be tampering were placed in the
category of "tampered" vehicles. The remaining vehicles were
categorized as "non-tampered" vehicles. As a subset of this
group, those vehicles which exhibited no evaporative control
system malfunctions were categorized as "problem-free"
vehicles. Table 2-13 listed the potential malfunctions in the
evaporative control systems, and also noted the malfunctions
which were considered tampering. Within each of the above
categories, the vehicles have also been separated into
carbureted and fuel-injected vehicles.
Through consideration of the different evaporative
emission rates for each of these categories, the individual
components of evaporative emissions were determined. The
components are: 1) the standard level, 2) the insufficient
design capacity/purge effect, 3) the malmaintenance and defect
effect, 4) the excess RVP effect, and 5) the tampering
effect. The magnitude of each of the first four components was
determined from the non-tampered and problem-free EF data base
and is discussed in Section I below. The m-use EF sample is
not thought to have a representative number of tampered
vehicles, so the magnitude of the tampering effect has not been
developed from this testing and is discussed separately in
Section II. The emission components are used to determine the
total evaporative emission factors for the various control
scenarios in Chapter 3.
The third section will describe how the LDV emission rates
were extrapolated to light-duty trucks and heavy-duty
vehicles. The fourth section will describe how emissions are
projected for vehicles produced in the future after the
certification test procedure is amended.
-------
I
2-A-2
I. LDV Evaporative Emission Components
The average measured emission rates for non-tampered and
problem-free LDVs for each of the three fuels tested in the EF
program are shown in Table 2-A-l. This analysis requires that
the emission rates be known for m-use RVPs other than just
these three levels. Therefore, curves were fit through the
actual vehicle data for problem-free vehicle diurnal and
hot-soak emissions for each type of fuel metering system.
Diurnal emissions were curve-fit versus the uncontrolled
diurnal index (UDI) as defined in Section II of Chapter 2, and
hot-soak emissions were curve-fit versus RVP as defined m
Section II of Chapter 2. For the problem-free fuel-injected
diurnal emission equation, the best fit curve began rising at
indices slightly lower than 1.0. To eliminate this anomaly,
diurnal emissions on an 8.0 psi fuel were estimated to be 60
percent of the emissions on Indolene and re-curve-fit with the
extra point. This percent was, chosen so that the resulting
best-fit curve was smooth and increasing. The best fit
problem-free equations are listed in Table 2-A-l.
The non-tanpered and problem-free emission rates have been
separated into the four components listed previously. The
remainder of this section will describe the process by which
the magnitude of each component was determined.
A. Standard Levels
The standard levels represent the emission rates which
would be seen if the vehicles emitted at the current
2-gram/test LDV standard on 9.0 psi RVP fuel. As it is
necessary to separate the standard level into diurnal and
hot-soak losses (which vary from vehicle to vehicle), it is
assumed the ratio of hot-soak to diurnal emissions from
problem-free vehicles is the same which would be seen if the
standard level were met. Therefore, to determine the standard
levels, the hot-soak and diurnal emissions on 9.0 psi RVP fuel
from the problem-free sample are normalized such that their sum
equals 2 grams/test. The emission rates observed from the
problem-free vehicles and the calculated standard levels are
presented in the first part of Table 2-A-2.
B, Insufficient Design Capacity/Purge Effect
The differences between the standard levels and the
problem-free emission rates on 9.0 psi RVP fuel represent the
insufficient design capacity/purge effect. This is based upon
the assumption that a properly operating evaporative control
system with no malfunctions should meet the 2-gram/test
standard m-use. This effect is determined by subtracting the
calculated standard levels from the emission rates for
problem-free vehicles on 9.0 psi RVP fuel, and is shown at the
bottom of Table 2-A-2. Note for fuel-injected vehicles, the
-------
2-A-3
Table 2-A-l
Non-Tampered and Problem-Free 81+ LDV and LDT Evaporative
Emission Rates (g/test)
Non-Tampered EF Test Data
Fuel Metering
System
RVP(psi)
UDI*
Hot-Soak
Diurnal
CARB
9 0
1.0000
2 . 57
2 . 60
10 . 4
1.4567
3 . 28
5 . 29
11.7
2.0677
4 . 26
10 .33
FI
9 . 0
1.0000
0.88
1 . 67
10 . 4
1.4567
1 .71
4 . 11
11 . 7
2.0677
2 . 70
8. 51
Problem-Free
EF Test- Data
Fuel Metering
System
RVP(psi)
UDI*
Hot-Soak
Diurnal
CARB
9. 0
1 . 0000
1 . 66
1 .46
10 . 4
1.4567
2 . 03
3 . 79
11. 7
2.0677
3 . 32
8 . 63
FI
9 . 0
1.0000
0 . 63
1 . 17
10.4
1.4567
1 .27
3.35
11 . 7
2.0677
2 . 07
7 . 56
Best-Fit Problem-
Free Equations:
CARB Hot-Soak =
0.17677 (RVP)z - 3
.0735 (RVP) +
14.9650
CARB Diurnal
1.5498 (UDI)
2 + 1.!
6287 (UDI) - 1
. 7274
FI H01
* (RVP)2 - 1
.2099 (RVP) +
4.5867
FI Di
+ i
. 9335
B
s
build
-------
2-A-4
Table 2-A-2
Estimation of Standard Level and Insufficient
Purge Effect for 81+ LDVs and LDTs (g/test)
Problem-Free Vehicle
Average with 9.0 RVP
CARB
FI
Hot Soak Diurnal
1.66 1.46
0.63 1.17
Standard Level*
CARB 1.06 0.94
FI 0.70 1.30
Insufficient Design Capacity/
Purge Effect**
CARB 0.60 0.52
FI 0.00 0.00
Problem-free average normalized to 2-gram standard.
Problem-free average minus standard level; if negative, as
for FI vehicles, considered zero.
-------
2-A-5
insufficient design capacity/purge effect is non-existent
because problem-free average emissions on 9.0 psi RVP fuel are
under 2 grams/test. This is strong evidence that the majority
of vehicles can meet the 2 gram/test standard m-use.
C. Malmaintenance and Defect Effect
Non-tampered vehicles, on the whole, show significantly
higher average evaporative emission rates than do problem-free
vehicles (a subset of non-tanpered vehicles). This effect is
assumed to be due to improper maintenance and equipment defects
in the evaporative emission control system, but not to
deliberate tampering. The distinction between malmaintenance,/
defects and tampering was made earlier in Chapter 2 in Table
2-13.
The malmaintenance and defect (M&D) effect is determined
by subtracting the average problem-free emissions from the
average non-tampered emissions for each fuel. The magnitude of
the M&D effect was found to increase linearly with fuel RVP and
UDI. This is expected since M&D problems result in partial or
total disablement of the evaporative control system. ""he
determination of this relationship for hot-soak emissions from
fuel-injected vehicles is shown graphically in Figure 2-A-l.
The top line shows the average emission rates on the three EF
fuels for all non-tampered vehicles; the middle line shows
these rates for only problem-free vehicles; and the bottom line
shows the difference between the top two lines defined to be
the M&D curve. A linear regression has been fitted to the
curve to arrive at the M&D equation. This same method was used
to develop the relationships for fuel-injected diurnal,
carbureted diurnal, and carbureted hot-soak emissions.* The
predicted M&D equations and emissions are summarized in Table
2-A-3.
Several carbureted vehicles had unexplainably high
hot-soak emissions or hot-soak emissions which did not
increase when tested on higher RVP fuels. These vehicles
skewed the hot-soak M&D effect such that the M&D effect
did not increase with increasing RVP. The carbureted
hot-soak M&D effect was adjusted with the following
conditions. The sum of the individual M&D effects on the
three fuels was kept the same. The ratio of the slope of
the carbureted hot-soak M&D line to the slope of the
fuel-injected hot-soak M&D line was set equal to the ratio
of the slope of the carbureted diurnal M&D line to the
slope of the fuel-injected diurnal M&D line. The
resulting equation and emissions are listed in Table 2-A-3.
-------
Figure 2-A-l
FUEL—INJECTED HOT-SOAK EMISSION?
«
c
o
V)
c
o
t
UJ
4
c
0
l/J
1
•*-
0
1
2,9
2.6 "1
2,4
1.0
1.6 -
1.4 -
i n
i r Jm
1 -
o.s -
0.6 -
0.4 -
0,2 -
I
8,5
HON-TAMPERED
3,5
Fu»l RVP (psi)
¦f PROBLEM —FREE
10-5 11,5
~ MALM AI NT, 3; DEFECT
-------
2-A-7
Table 2-A-3
Malmaintenance and Defect Effect
for 81+ LDVs and LDTS
(qrams/test)
Fuel Metering
System
RVP(psi)
UDI *
Hot-Soak D
iurnal
CARB
8.0
0 .7547
0 . 62
1 . 06
8.5
0.8705
0 . 71
1 . 12
9 . 0
1.0000
0 .80
1 . 19
9 . 5
1.1454
0 . 88
1 . 26
10 . 0
1.3094
0 . 97
1 .34
10 . 5
1.4961
1 . 06
1. 44
11 0
1.7104
1 . 14
1 . 55
11.5
1.9581
1 . 23
1 . 68
FI
8 . 0
0.7547
0 . 11
0 . 42
8.5
0.8705
0 . 18
0.47
9 . 0
1 0000
0 . 25
0 . 53
9.5
1 . 1454
0 32
0 . 59
10 . 0
1.3094
0 . 39
0 . 65
10 5
1.4961
0 . 46
0 . 73
11.0
1.7104
0 . 53
0 . 82
11 . 5
1.9581
0 . 60
0 . 92
M&D Ecruations:
CARB Hot-Soak = 0.17395
(RVP) - 0.
76994
CARB Diurnal
= 0.51370
(UDI) + 0.
67194
FI Hot-Soak
= 0.14068
(RVP) - 1.
01835
FI Diurnal
= 0.41542
(UDI) + 0.
11016
* Based on
diurnal test conditions of 60-84°F
heat bi
a 40 percent full tank,
-------
2-A-8
D. Excess RVP Effect
The excess RVP effect represents the evaporative emissions
which arise from operating vehicles on a fuel of a higher
volatility than the fuel for which they were designed.
Manufacturers apparently have designed current evaporative
control systems to meet the 2-gram/test standard when operated
on Indolene, with an average RVP of 9.0 psi without regard to
the volatility of present commercial fuels. However, in-use
gasoline in many areas of the country currently has an average
volatility well above 9.0 psi. The magnitude of the in-use RVP
effect will of course depend upon the actual volatility of the
in-use fuel and in-use temperatures.
The excess RVP effect is defined as the difference between
problem-free evaporative emissions on commercial fuel and
problem-free evaporative emission on Indolene. The predicted
excess RVP effect emissions and equations are listed in Table
2-A-4.
It be should noted, in cases where RVP is less than the
vehicle's certification fuel RVP (for hot-soak emissions) and
in cases when UDI is less than 1.0 (for diurnal emissions), the
excess RVP effect is automatically set equal to zero. Also,
due to the non-1inearity of the diurnal excess RVP equation,
the incremental excess RVP emissions for high DDIs become
greater than the incremental emissions predicted by the
tampered (uncontrolled) diurnal equations presented later in
t'ns appendix in Table 2-A-6. Therefore two sets of diurnal
excess RVP equations are derived and presented in Table 2-A-4.
The first set of diurnal excess RVP equations are derived as
explained above. The second set of excess RVP diurnal
equations makes the incremental increase in excess RVP
emissions equal to the incremental increase in tampered diurnal
emissions. The UDI where the switch of equations occurs is
determined by setting the derivatives of the problem-free
emissions equation and tampered diurnal emissions equation,
with respect to UDI, equal to each other and solving for the
resulting UDI. The UDI at which the switch occurs is also
shown in Table 2-A-4.
E. Final Predicted Non-Tampered and Problem-Free
Emissions
Table 2-A-5 contains the final predicted emissions for
non-tampered and problem-free LDVs for the various test
conditions. Non-tampered emissions are the sum of the standard
level, insufficient design capacity/purge effect, the
malmaintenance and defect effect, and the excess RVP effect.
Problem-Free emissions are the sum of the standard level,
insufficient design capacity/purge effect, and the excess RVP
effect.
-------
2-A-9
Table 2-A-4
Excess RVP Effect for 81+ LDVs and LDTs (g/test)
Fuel Metering
System
CARB
FI
RVP(psi) UDI* Hot-Soak Diurnal
8 . 0
0 . 7547
0 . 00
0 . 00
8 . 5
0.8705
0 . 00
0 . 00
9 . 0
1.0000
0. 00
0 . 00
9 . 5
1.1454
0 . 10
0 . 72
10 0
1 3094
0 29
1 61
10.5
1 . 4961
0 56
2 . 73
11.0
1.7104
0 . 92
4 . 14
11 5
1 9581
1 .38
5 95
8 . 0
0 . 7547
0 . 00
0 . 00
8 . 5
0.8705
0.00
0 . 00
9 . 0
1.0000
0 . 00
0 .00
9 . 5
1.1454
0 . 19
0 .46
10.0
1.3094
0 . 42
1 . 16
10 . 5
1.4961
0 . 69
2 . 18
11 0
1 .7104
1 .00
3 . 64
11 5
1 .9581
1.36
5 . 72
Excess RVP Equations.
CARB Hot-Soak = 0.17677 (RVP)2 - 3.0735 (RVP) + 13.34313
CARB Diurnal (for UDI _< 3.8386)
= 1.5498 (UDI)2 + 1.6287 (UDI) - 3 1785
(for UDI > 3.8386)
= 13.52677 (UDI) - 26.02326
FI Hot-Soak = 0.08553 (RVP)2 - 1.2099 (RVP) + 3.96125
FI Diurnal (for UDI < 2.5806)
= 3.4304 (UDI)2 - 4.1784 (UDI) + 0.7480
(for UDI > 2 5806)
= 13.52677 (UDI) - 22.2123
Based on diurnal test conditions of 60-84°F heat build,
and a 40 percent full tank.
-------
2-A-10
Table 2-A-5
(g/test)
Fuel Metering
System
RVP(psi)
UDI *
Hot-Soak
Diurnal
CARB
8 . 0
0 . 7547
2.28
2. 52
8 . 5
0.8705
2.37
2 . 58
9.0
1.0000
2. 46
2.65
9 . 5
1,1454
2 . 64
3 . 44
10 . 0
1.3094
2.92
4.41
10 . 5
1.4961
3 .28
5. 63
11 . 0
1.7104
3 . 72
7 . 15
11.5
1,9581
4 . 27
9 . 09
FI
8.0
0 . 7547
0 .81
1 . 72
8.5
0.8705
0 . 88
1 . 77
9 . 0
1 0000
0.95
1 .83
9 . 5
1.1454
1 .21
2 . 35
10 . 0
1.3094
1 .51
3 . 11
10.5
1.4961
1 . 85
4 . 21
11 . 0
1.7104
2 . 23
5 . 76
11 . 5
1.9581
2 . 66
7 . 94
Predicted
Problem-Free
Emissions for 81+
LDVs and LDTs
(g/test)
Fuel Metering
System
RVP(psi)
UDI *
Hot-Soak
Diurnal
CARB
8.0
8.5
9 . 0
9 . 5
10 . 0
10 . 5
11 . 0
11 . 5
0 .
0 ,
1
1
1
1
1
1
7547
8705
0000
1454
3094
4961
7104
9581
1 .
1
1
1
1
2
2
3
66
66
66
76
95
22
58
04
1,
1
1
2
3
4
5
7
46
46
46
18
07
19
60
41
FI
8.0
8.5
9.0
9.5
10 .0
10 . 5
11.0
11. 5
0 .7547
0.8705
0000
. 1454
3094
, 4961
7104
9581
0 . 70
0 . 70
0 . 70
0 . 89
1 . 12
1 .39
1 . 70
2 . 06
,30
.30
.30
. 76
,46
,48
,94
,02
Based on diurnal test conditions of 60-84°F heat
build, and a 40 percent full tank.
-------
2-A-ll
II. Tampering Effect
The emission rates for tampered vehicles have been derived
from SHED testing on light-duty vehicles with removed canisters
and/or fuel caps. The SHED tests were performed using 9.G psi
and 11.7 psi RVP fuels. The emissions at other RVPs and UDls
were determined through linear interpolation and are shown in
the top portion of Table 2-A-6. Certain assumptions are made
as part of this analysis. First, uncontrolled diurnal
emissions are assumed to be the same regardless of either the
type of disablement or the vehicle's fuel metering system.
Secondly, for fuel-injected vehicles, uncontrolled hot-soak
emissions are assumed to be the same for either canister
removal or fuel cap removal due to the dominance of fuel tank
emissions relative to those from the fuel injector. Finally,
for carbureted vehicles, fuel cap removal is assumed not to
lead to any increases in hot-soak emissions (i.e., the
uncontrolled values are the sare as the non-tampered hot-soak
emissions shown in Table 2-A-5).
The difference between uncontrolled emissions and
non-tainpered vehicles is defined as the "tampering offset" to
be used in the MOBILE3 program with tampering frequency
estimates. The tampering offsets are given in the bottom half
of Table 2-A-6.
III. Extrapolation of LDV Data to LDTs and HDGVs
The extrapolation of the light-duty vehicle evaporative
emission rates to light-duty trucks and heavy-duty vehicles is
done here exactly as it is done for MOBILE3.[89] Since little
or no in-use test data exist for these vehicles, the emission
rates are extrapolated based upon their relative standard
levels. For LDTs, this means that evaporative emission rates
will be exactly the same as for LDTs, as both of these vehicle
classes must meet the same 2.0-gram/test standard under
identical test procedures.
Beginning with the 1985 model year, HDGVs must meet either
a 3.0-gram/test or 4.0-gram/test standard, depending upon their
gross vehicle weight. Therefore, the evaporative emission
rates for HDGVs under the 3.0- and 4.0-gram standards are those
for LDVs multiplied by 1.5 and 2.0, respectively, and weighted
by their respective sales fractions projected for 1987 — 81.5
percent for 3-gram and 18.5 percent for 4-gram vehicles. This
yields an overall heavy-duty/light-duty multiplicative factor
of 1.5925. MOBILE3 assumes heavy-duty vehicles are completely
carbureted, so the evaporative emission factors from only the
carbureted vehicles ate used. The resulting evaporative
emission factors and the magnitude of each of the various
components are shown in Table 2-A-7.
-------
2-A-12
Table 2-A-6
Tampered Emissions of 81+ LDVs and LDTs
Uncontrolled Emission Rates (g/test)
Fuel Metering
RVP
Canister
Removal
Gas
; Cap
Removal
System
(psi)
UDI *
Hot-Soak
Diurnal
Hot-
Soak
Diurnal
CARB
8
. 0
0
. 7547
7 . 52
12.
. 27
2 .
28
12 .27
8
.5
0
.8705
8 .94
13 .
. 75
2 .
37
13 .75
9
0
1
0000
10 .36
15,
. 40
2 .
46
15.40
9 .
. 5
1
. 1454
11 . 79
17 ,
. 25
2 .
64
17 .25
10.
,0
1
.3094
13.21
19.
.34
2.
92
19.34
10 .
. 5
1
. 4961
14 . 63
21 .
, 72
3 .
28
21.72
11 .
0
1
. 7104
16.05
24.
. 45
3.
72
24 . 45
11 .
5
1
.9581
17.47
27.
. 61
4 .
27
27 . 61
8.0
0 . 7547
3 .64
12.27
3 . 64
12 .27
8 . 5
0.8705
4 .42
13 . 75
4 .42
13.75
9 . 0
1.0000
5 . 20
15. 40
5 . 20
15 40
9 . 5
1.1454
5.98
17 . 25
5.98
17 . 25
10 . 0
1.3094
6.75
19 .34
6 . 75
19 .34
10 . 5
1.4961
7.54
21. 72
7. 54
21.72
11. 0
1.7104
8.21
24 . 45
8 .21
24 .45
11. 5
1.9581
9 .00
27 . 61
9 . 00
27 . 61
Tampering Offsets (g/test)
RVP
Canister Removal
System
(psi)
UDI*
Hot-
-Soak
Diurnal
Hot-
-Soak
Diurnal
CARB
8.
, 0
0
. 7547
5,
24
9 ,
. 75
0 .
, 00
9 .
75
8,
. 5
0
. 8705
6 ,
. 57
11,
. 17
0 ,
,00
11 .
, 17
9 .
. 0
1
. 0000
7 .
.90
12
. 75
0 .
00
12 ,
. 75
9 ,
. 5
1
. 1454
9 .
. 15
13 ,
. 81
0 .
, 00
13 .
,81
10 .
. 0
1
.3094
10 ,
. 29
14 ,
, 93
0 .
. 00
14 ,
, 93
10.
. 5
1
. 4961
11
.35
16
. 09
0
. 00
16 .
, 09
11.
0
1
. 7104
12.
,33
17.
.30
0,
,00
17.
,30
11.
. 5
1
. 9581
13 .
. 20
18 .
. 52
0 .
. 00
18 .
, 52
FI
8 .
, 0
0
. 7547
2
.83
10
. 55
2 ,
,83
10 .
, 55
8 .
, 5
0
. 8705
3
. 54
11
. 98
3
. 54
11 ,
, 98
9 ,
0
1
. 0000
4
.25
13
, 57
4 .
,25
13 .
. 57
9
. 5
1
. 1454
4
. 77
14
.90
4
, 77
14 .
.90
10 ,
. 0
1
.3094
5
.24
16
.23
5 ,
, 24
16
. 23
10
. 5
1
. 4961
5
.69
17
. 51
5
. 69
17
. 51
11 .
, 0
1
. 7104
5 .
.98
18 .
. 69
5 ,
. 98
18 ,
. 69
11 ,
. 5
1
. 9581
6
.34
19
.67
6
,34
19 ,
.67
-------
2-A-13
Table 2-A-6 (Continued)
Uncontrolled Emission Equations:
All Diurnal = 12.74 (UDI) + 2.66
Hot-Soak
CARB Canister Removal = 2.84 (RVP) - 15.2
CARB Gas Cap Removal = same as Non-Tampered Emissions
FI Canister Removal or Gas Cap Removal = 1.56 (RVP) - 8.84
Tampering Offset Emissions Equations:
= Uncontrolled Emission - Non-Tampered Emissions
* Based on diurnal testing conditions of 60-84°F heat build,
and a 40 percent full tank.
-------
2-A-14
Table 2-A-7
(g/test)
Component
RVP (psi)
UDI *
Hot-Soak
Diurna
Standard Level
-
-
1 . 69
1. 50
Insufficient Design
—
-
0 .96
0 . 83
Capacity/Purge
Malmaintenance/Defect
** 8.0
0 . 7547
0 . 99
1. 68
8.5
0.8705
1 . 13
1. 78
9 . 0
1 . 0000
1 . 27
1 . 90
9 . 5
1.1454
1 . 40
2.01
10 . 0
1.3094
1 . 54
2 . 13
10 . 5
1.4961
1 . 69
2.29
11 . 0
1 .7104
1 .81
2.47
11. 5
1.9581
1 . 96
2 . 68
Excess RVP**
8.0
0.7547
0 . 00
0 . 00
8 . 5
0.8705
0 . 00
0 . 00
9 . 0
1 . 0000
0 . 00
0 . 00
9 . 5
1 . 1454
0 . 16
1 . 15
10 . 0
1 .3094
0 .46
2 . 56
10 . 5
1 .4961
0 . 89
4.35
11 . 0
1.7104
1 . 47
6. 59
11 . 5
1.9581
2 . 20
9 . 48
Total Non-Tampered
Average
8 . 0
0.7547
3 . 64
4 .01
8 . 5
0 . 8705
3 . 78
4.11
9 . 0
1.0000
3 .92
4.23
9 . 5
1 . 1454
4 .21
5 . 49
10 . 0
1 .3094
4 . 65
7 . 02
10 . 5
1.4961
5.23
8 . 97
11 . 0
1 .7104
5.93
11 .39
11 . 5
1.9581
6 .81
14 . 49
Based on diurnal testing conditions of 60-84°F heat build,
and a 40 percent full tank.
Equations for the components are the same as LDVs
multiplied by 1.5925.
-------
2 — A — 15
LDV data on uncontrolled evaporative emissions were also
used to develop the LDT and HDGV estimates, due to limited or
lack of evaporative testing on these classes. The tampering
offsets for LDTs are assumed to be equal to those developed
from the LDV data, as indicated in Table 2-A-6. The
methodology used to develop uncontrolled estimates for HDGVs is
similar to that mentioned previously with respect to the
non-tampered average.[89] Variations from the basic M0BILE3
method of extrapolating LDV evaporative data to HDGVs are
detailed in the EPA technical report, "The Effect of Fuel
Volatility on Controlled and Uncontrolled Evaporative
Emissions," (EPA-AA-TEB-EF-86-01 ). Tampered emission estimates
and tampering offsets for HDGVs are presented in Table 2-A-8.
IV. Projection of Evaporative Emissions for Future Vehicles
with a Change in Certification Fuel RVP
The emission component equa'ions listed in the previous
sections are for all 1981+ gasoline vehicles. However, the
inputs of RVP and DDI used in M0BILE3 are dependent on
certification fuel RVP as explained in Section II of Chapter
2. A change in the certification conditions affects the
components as explained below.
The breakdown of the 2.0 gram standard level into hot-soak
and diurnal portions is assumed to be independent of the
certification fuel RVP. The insufficient design capacity/purge
effect is assumed to be eliminated in all vehicles after the
certification test procedure is amended.
The malmaintenance and defect effect is assumed to be
independent of the certification fuel RVP, only the RVP of the
fuel in the tank. The majority of problems classified as
malmaintenance and defects of the evaporative control system
are problems which result in the disablement of the evaporative
control system. Therefore, the emissions of these vehicles
should not be dependent on the certification fuel RVP. The M&D
equation inputs for RVP and UDI, as defined in Section II of
Chapter 2, are calculated as if certification fuel RVP equals
9.0 ps i .
The excess RVP effect is dependent on manufacturers'
design goals, which presently appears to be based only on
certification fuel RVP. Vehicle manufacturers will have to
adjust the size of the evaporative canister and/or the amount
of purge in order to meet the current 2.0 g standard on a
different RVP certification fuel. The excess RVP equation
input for RVP and UDI are calculated as defined in Section II
of Chapter 2 as functions of the certification fuel RVP for
which the vehicles are designed.
-------
2-A-16
Table 2-A-8
Tampered Emissions of 85+ HDGV-^"*
Uncontrolled Emission
Rates
fg/test)
RVP
Canister
Removal
Cap Removal
(psi)
UDI
Hot-Soak
Diurnal
Hot-Soak
Diurnal
8.0
0 . 7547
11 21
22 .
55
3 . 64
22 . 55
8.5
0.8705
12 . 94
24 .
22
3 . 78
24 . 22
9 . 0
1.0000
14 67
26 .
08
3 .92
26 . 08
9.5
1.1454
16 . 40
28 .
17
4 .21
28 . 17
10.0
1.3094
18. 12
30.
53
4 . 65
30. 53
10.5
1.4961
19. 86
33 .
22
5.23
33 . 22
11.0
1 .7104
21 .58
36 .
30
5.93
36 .30
11 . 5
1.9581
23 .31
39.
87
6 .81
39 .87
Tampering Offsets (g/test)
RVP
Canister
Removal
Cap Removal
(psi)
UDI
Hot-Soak
Diurnal
Hot-Soak
Diurnal
8.0
0.7547
7 . 57
18
. 54
0 . 00
18. 54
8.5
0 . 8705
9 . 16
20
44
0 . 00
20 . 44
9 . 0
1,0000
10. 75
21
. 85
0 . 00
21 .85
9.5
1.1454
12 . 19
22
. 68
0 . 00
22 . 68
10 .0
1 . 3094
13 . 47
23
. 51
0 . 00
23 . 51
10 .5
1.4961
14 . 63
24
.25
0 . 00
24 . 25
11 . 0
1.7104
15. 65
24
.91
0 . 00
24 .91
11.5
1 . 9581
16 . 50
25
.38
0 . 00
25 . 38
Included in EPA Technical
Volatility on Controlled
Emissions," by Celia
(EPA-AA-TEB-EF-86-01).
Report, "The Effect of Fuel
and Uncontrolled Evaporative
Shih and Tom Darlington,
-------
2-A-17
Tampered emissions are assumed to be independent of
certification fuel RVP, and dependent on the RVP of the fuel in
the tank. Tampering results in the disablement of the
evaporative control system and therefore the emissions from
tampered vehicles do not depend on the certification fuel RVP.
The tampered equation inputs for RVP and UDI are calculated as
defined in Section II of Chapter 2 as if certification fuel RVP
equals 9.0 psi. (The tampering offsets are affected by a
change in certification fuel RVP because non-tampered emissions
are affected by a change in certification fuel RVP.)
Table 2-A-9 contains the predicted non-tampered emissions
for an LDV designed for a 2 gram/test standard on a
certification fuel of 10.0 psi at various in-use RVPs. The
choice of a certification fuel RVP of 10.0 psi here is
arbitrary and intended only to show the emission effect of
changing certification fuel RVP to a level different from 9.0
psi, as already presented in Table 2-A-5.
-------
Table 2-A-9
Predicted Non-Tampered Evaporative Emission Rates
for LDVs Designed for a
Certification Fuel RVP of 10.0 psi
System
RVP(psi) UDI;
Hot-Soak Diurnal
CARB
FI
8.0
0.5764
1 . 68
2 . 00
8.5
0.6648
1 . 77
2 . 06
9 . 0
0.7637
1 . 86
2 . 13
9 . 5
0.8748
1 .94
2 . 20
10 . 0
1.0000
2.03
2 . 28
10 . 5
1.1426
2.20
3 . 09
11 . 0
1.3062
2 . 44
4 . 08
11 . 5
1.4954
2. 76
5.34
8 . 0
0.5764
0 .81
1 . 72
8 . 5
0.6648
0 . 88
1 . 77
9 . 0
0.7637
0 .95
1 . 83
9 . 5
0.8748
1 . 02
1 . 89
10 . 0
1.0000
1 . 09
1 . 95
10 . 5
1.1426
1 .33
2 . 48
11.0
1.3062
1 . 60
3 . 26
11 . 5
1.4954
1 . 90
4 . 39
* UDI used for excess RVP effect. Based on diurnal test
conditions of 60-84°F heat build, 40% full tank, and
certification fuel RVP equal to 10.0 psi. UDI used for
M&D effect is same as UDI based on a certification fuel
RVP of 9.0 psi.
-------
APPENDIX 2-B
Detailed Benzene Emission Data
This appendix contains all of the data accumulated for
evaluation of benzene emissions from light-duty gasoline
vehicles. Not all of it proved useful in the analysis. Below
is a description of what each table includes.
Table 2-B-l - Contains all of the exhaust data used in the
analysis to derive the emission models
presented at the top of the table.
Table 2-B-2 - Contains additional exhaust data which was not
useful in this analysis.
Table 2-B-3 - Contains the limited evaportive data
available, various modeling equations proposed
based on that data (not all are discussed in
the write up due to poor correlation), back up
data to test the models, and the constants for
some of the models as a result of application
of the back up data.
-------
2-B-2
Table 2-B-l
Exhaust Data Used for Benzene Analysis
3-way Correlation: W%BzEx = 1.0306(FV%Bz)+.17588(FV%AROM)-l.9785
3WOX Correlation: W%BzEx = .67963(FVlBz)+.0 680 7(FV!AROM)-.34679
AVG: Pre-80 Correlation* W%BzEx = .85511(FV%Bz)+.12198(FV%AROM)-l.1626
POST 80 OX Cat: W%BzEx = .85511(FV%Bz)+.12198(FV%AROM)-l.1626
Pre 1975 Cars (No Cat)
Vehicle
FV%Bz
No Bz
FVIAROM
W%BzEx
Predicted
Pre-80
63 Chevy[84,85]
63 Chevy[84,85]
1. 55
4 . 9
20.31
20 .31
3.8
7 . 0
2 . 64
5 . 50
75-79 Cars
No Cat
79 Volaret 82] 1 66
76 Celica[82] 1.66
75 Pinto[82] 1.49
76 Monarch[82] 1.49
77 Hornet[82] 1.49
77 F-10[82] 1.49
79 Rx-7[82] 1.49
36.47 6.42 4.71
36.47 4.55 4.71
25.61 2.50 3.24
25.61 1.75 3.24
25.61 3.23 3.24
25.61 1.84 3.24
25.61 2.67 3.24
OX CAT
79 Chevette[82] 1.66
78 LTD[82] 1.66
78 Mustang[82] 1.66
76 Starfire[82] 1.49
76 Regency[82] 1.49
77 Skyhawk[82] 1.49
75 Valient[82] 1.49
78 Phoenix[82]
79 Corollaf 82] 1.49
77 Chevette[82] 1.49
78 Volaret 82] 1.49
78 200SX[82] 1.49
79 Fairmont[82] 1.49
75 Cutlass[82] 1.49
78 Monarchf84,85] .24
78 Monarchf84,85] 6.0
78 Monarch[84,85] 1.29
78 Monarch[84,85] 1.65
79 LTD[84,85] 1 29
79 LTD[84,85] 6.0
79 LTD[84,85] 1.65
36.47 5.68 4.71
36.47 3.26 4.71
36.47 6.68 4.71
25.61 3.71 3.24
25.61 4.27 3.24
25.61 3.00 3.24
25.61 3.16 3.24
3.92 3.24
25.61 2.93 3.24
25.61 2.36 3.24
25.61 3.15 3.24
25.61 2.08 3.24
25.61 1.71 3.24
25.61 2.39 3.24
22.89 2.9 1.83
22.89 6.6 6.76
36.03 3.4 4.34
20.04 2.6 2.69
36.03 5.1 4.34
36.03 7.0 8.36
20.04 2.8 2.69
-------
2-B-3
Table 2-B-l (continued)
3-way + OX Cat
78 Ford[81]
1980 Cars and later
No Cat
Vehicle
2.16
FV%Bz
42 . 2
No Bz
FVIAROM
3 .1
x=3.73
Sx = l . 62
WIBzEx
3 . 99
x=3.88
S, = l.34
Predicted
Post-80
80 Cutlass[82]
1 . 49
25 61
3 . 24
3 24
OX Cat
80 Scirocco[82] 1.49
80 Electra[82] 1.49
80 GLC[82] 1.49
81 Chevette[82] 1.49
81 Lynx[82] 1.49
82 LeBaron[82 ] 1.49
81 Concord[82] 1.49
80 LeBaron[82] l. 49
25 61 3.53 3 . 24
25 61 2.71 3.24
25.61 2.04 3 24
25.61 3.53 3.24
25.61 1.72 3.24
25.61 4.82 3 24
25.61 3.27 3.24
25 . 61 __ 3.36 _ 3 . 24
x=3.13 x=3.24
S,= .91 Sx=0.0
3W Cat
81 626[82] 1.49
81 Escort[82] 1.49
82 Citation[82] 1.49
81 Escort[82] 1.49
82 Mustang[82] 1.49
80 Lincoln!81] 2.16
80 Lincoln 2.36
80 Buick[81] 2 16
80 Buick 2.36
84 VW[83] .03
84 VW 1.5
84 VW 2.12
84 VW 3.99
84 VW .03
84 VW 1.48
84 VW 2.70
84 VW 3.90
85 Chevy[83] .03
85 Chevy 1. 5
Predicted
3W
25.61 1.96 4.06
25.61 2.79 4.06
25.61 7.74 4.06
25.61 2.16 4.06
25.61 1.39 4.06
42.2 6.0 7.67
27.4 4.2 5.29
42.2 15.4 7.67
27.4 10.1 5.27
25.7 3.9 2.57
25.6 5.2 4.07
25.2 6.3 5.26
24.9 7.8 6.51
40.1 6.2 5.11
39.4 7.9 6.48
38.9 9.6 7.65
38.2 10.2 8.76
25.7 1.7 2.57
25.6 2.8 4.07
-------
2-B-4
Table 2-B-l (continued)
Vehicle
FV%Bz
No Bz
FV%AROM
W%BzEx
85 Chevy
85 Chevy
85 Chevy
85 Chevy
85 Chevy
85 Chevy
2
3
1
2
3
72
99
03
48
70
90
25 . 2
24 . 9
40 . 1
39 . 4
38.9
38.2
3
5
3
4
5
7
X=5.77
Sx=3.27
3W + OX
81 Citation[82] 1.66 36.47 6.07
82 Rx-7[82] 1.49 25.61 1.25
82 Delta 881 82 ] 1.49 25.61 3.68
81 Jetta[82] 1.49 25.61 2.80
82 chevette[82] 1.49 25.61 2.23
81 J-1000[82] 1.49 25.61 3.37
81 Citation[82] 1.49 25.61 3.46
81 Citat ion[82] 1.49 25.61 2.40
80 Caprice[81] 2.16 42.2 7.20
80 Caprice 2.36 27.4 4.5
85 Chevy[83] .03 25.7 1.4
85 Chevy 1.5 25.6 2.8
85 Chevy 2.72 25.2 3.1
85 Chevy 3.99 24.9 4.2
85 Chevy .03 40.1 2.1
85 Chevy 1.48 39.4 2.9
85 Chevy 2.70 38.9 4.0
85 Chevy 3.90 38.2 5.5
85 Ford[83] .03 25.7 1.3
85 Ford 1.5 25.6 2.1
85 Ford 2.72 25.2 3.2
85 Ford 3.99 24.9 3.9
85 Ford .03 40.1 2.5
85 Ford 1.48 39.4 3.3
85 Ford 2.70 38.9 3.4
85 Ford 3.90 38.2 5.8
85 Plymouth!83] .03 25.7 2.0
85 Plymouth 1.5 25.6 2.5
85 Plymouth 2.72 25.2 2.6
85 Plymouth 3.99 24.9 3.8
85 Plymouth .03 40.1 2.6
85 Plymouth 1.48 39.4 3.3
85 Plymouth 2.70 38.9 4.3
85 Plymouth 3.90 38.2 4_. 5
x=3.35
Sx=l.36
Predicted
3W
5 . 26
6 .51
5.11
6 . 48
7 . 65
8 . 76
x=5.56
S X = 1•68
Predicted
3W OX
3 . 26
2.41
2.41
2.41
2.41
2.41
2.41
2.41
3 . 99
3 . 12
1.42
2.42 *
3 . 22
4 . 06
2 .40
3 .34
4 . 14
4 .90
1 .42
2 . 42
3 . 22
4.06
2 .40
3 .34
4 . 14
4 . 90
1 .42
2 . 42
3 . 22
4 . 06
2 .40
3 .34
4 . 14
4 .90
x=3.09
S«= .98
-------
2-B-5
Table 2-B-2
Additional Data Not Used for Benzene Analysis
Vehicle
CAT
FTP
RVP
V%
Fuel
Bz
V%
Fuel
AROM
g/mi
HC
Emis
g/mi
Benz
Emis
%
Bz/H<
70
Delta[79]
None
X
2 . 29
.08
3 . 6
70
Dodge[7 9]
None
X
2 .75
. 09
3 2
70
Chevy[79]
None
X
3.30
. 09
2 . 8
70
Ford[79 ]
None
X
3 . 78
. 09
2 . 3
77
AMC[79]
None
X
1 . 17
. 006
5 5
78
Chevy[7 9]
OX
X
. 50
. 03
5 . 3
78
Chevy[79]
ox
X
.31
.01
2 . 6
78
Ford[79]
ox
X
. 43
.01
3 . 0
78
Ford[79]
ox
X
.51
. 02
4 . 4
78
Ford[79 ]
3WOX
X
. 11
. 003
2 . 7
78
Pont[79]
3W
X
. 26
. 001
4 . 7
78
SAAB[79]
3W
X
. 16
.01
7 . 6
79
Merc[79]
3WOX
X
.21
.01
3 . 6
78
Buick[79 ]
OX
X
1 .71
. 009
5 . 1
79
Merc[79 ]
3WOX
X
.69
. 02
2 . 6
79
Merc[79]
3WOX
X
.68
. 02
3 . 2
78
Fordt 79 ]
OX
X
1 .74
. 06
3 . 5
78
Volvo[79 ]
3W
X
.61
.05
8 . 0
78
Olds[79 ]
OX
X
.87
. 05
5 . 9
78
Chevy[79]
OX
X
. 64
. 03
5.3
78
Chevy[79]
OX
X
.45
. 02
4 . 0
78
Ford[79 ]
OX
X
. 64
.01
2 . 0
78
Chryst 79]
ox
X
4.57
.21
4 . 6
72
Chevy[80]
None
X
10 . 2
26 . 2
1.51
. 057*
74
Mazdat 80]
None
X
10.2
26 . 2
1 .47
. 0718
*
75
Fury[80 ]
OX
X
10.2
26 . 2
.49
. 0086
*
76
Fordt80]
ox
X
10 . 2
26 . 2
. 55
. 008*
75
Chevy[80]
ox
X
10.2
26 . 2
.25
. 0072
*
76
Cordoba[80]
None
X
10 .2
26 . 2
. 66
.0268
*
76
Fury[80 j
None
X
10 . 2
26 . 2
1 .45
. 045*
[80]
OX
X
10.2
26 . 2
. 92
. 0096
*
76
Fordt80]
OX
X
10 . 2
26.2
. 63
.0122
77
Chryst80]
OX
X
10.2
26.2
.26
. 0064
*
77
Fury[80 ]
OX
X
10 .2
26.2
.84
. 0236
*
77
Nova[80]
OX
X
10 . 2
26 . 2
.62
.0164
77
Vega[80]
OX
X
10 . 2
26.2
. 23
. 0096
*
77
Rabbitt 80]
OX
X
10 .2
26 . 2
. 18
. 0047
*
77
Audi[80]
OX
X
10.2
26.2
. 27
. 0101
*
75
GM[78]
OX
X
. 48
.0116
75
GM[78]
OX
X
. 70
.0154
75
GM[7 8]
ox
X
. 78
.0156
75
GM[78]
ox
X
. 73
. 0208
75
GM[78]
ox
X
. 79
. 0361
Data represents benzene plus cyclohexane
-------
2-B-6
CRC Data [83]
Vehicle
85 Ford
85 Plymouth
Table 2-B-3
Evaporative Data Used for Benzene Analysis
Fuel Syst
TBI
CARB
RVP
10.8
10.8
FV°oBz
5.22
5.22
W°oBz/HC
1.91
b . 36
A)
FIEWoBz
B)
CEVW°bBz
C)
FI EW°oBz
D)
CEVWoBz
E)
FIEW%Bz
F)
CEVWoBz
G)
FIEWoBz
H)
CEVWoBz
Modeling Equations
= 3 . 952 ( FV°oBz ) /RVP
= 13 . 159 ( FV°oBz ) / RVP
= 9 . 02 9 ( FV°,Bz ) 1 /RVP
= 30 . 064 ( FV°oBz ) "''/RVP
= . 151 ( FV°oBz ) ( 13 . 2 3 - RVP )
= . 5 ( FV°oBz ) ( 13 . 2 3 - RVP )
= . 3 ( FVSBz ) ' ' ' ( 13 . 58-RVP)
= (FV°oBz) '''''( 13. 58-RVP)
Black & High Data[84,85]
Vehicle
Fuel
System
RVP
FV°oBz
Actual Predicted Predicted Predicted Predicted
w°oBz/HC Eqn B Eqn B Eqn F Eqn H
63
Chevy
CARB
12.8
1. 55
. 6
1. 59
2.92
.33
.97
63
Chevy
CARB
12 . 8
4.9
3 . 6
5.03
5. 20
1.05
1.72
77
Mustang
CARB
9.8
1.29
1.8
1.73
3.48
2.21
4. 29
77
Mustang
CARB
9 . 8
6.0
6.2
8.06
7.51
10.29
9.25
77
Mustang
CARB
12 . 3
1.65
1.6
1.76
3.14
.77
1. 64
78
Monarch
CARB
8.4
. 24
. 5
.38
1.75
. 58
2 . 54
78
Monarch
CARB
8.4
6.0
5.7
9.40
8.77
14 . 50
12.69
78
Monarch
CARB
9 . 8
1.29
1.8
1.73
3.48
2.21
4.29
78
Monarch
CARB
12 . 3
1.65
. 6
1.76
3.14
.77
1.64
79
LTD II
CARB
9.8
1.29
2 . 2
1.77
3.48
2.21
4.29
79
LTD II
CARB
9.8
6.0
7 .1
8.06
7.51
10. 29
9.25
79
LTD II
CARB
12 . 3
1.65
. 5
1.76
3.14
.77
1. 64
-------
.28
.42
. 3 8
. 33
. 55
J 2
. / 3
. 38
.77
. 7 4
. 7 0
¦ 60
. 20
2-B-7
Table 2-B-3 (continued)
Constants Based on Black & High Data
Const Eqn B
4 . 95
9.40
13.67
10.13
11.93
17 . 5
7 .98
13 .67
4 .47
16 .71
11.60
3.73
x = 10 .48
Sx= 4.50
Const Eqn B
6. 17
20.82
15. 53
24.81
15.32
a. 57
13.55
15.53
5.75
18.-8
2S.40
4 . 79
x=15.35
Sx= 7.71
Const Eqn F
13.57
14 . 27
12.59
11.87
14 . 24
12 . 57
10. 3
12 . 59
13 .03
L 3 . 2 1
12 . 17
12.91
~=12 .78
S<= 1.07
-------
Chapter 3
Environmental Impacts
This chapter examines the environmental impacts associated
with each of the gasoline volatility control scenarios
described in Chapter l. The analyses presented in this chapter
examine the impacts of long-term control of in-use and
certification test fuel volatility to equal levels, and of
additional short-term control of in-use fuel volatility.
Section I of this chapter discusses the computer models
that were used (M0BILE3 and EKMA) and the input data. Section
II presents the results of the nationwide emissions inventory
analysis, and Section III describes the results of the city-
specific ozone air quality analyses. The chapter concludes
with an analysis of the reductions in toxic emissions {i.e.,
benzene) and cancer incidences that are projected to result
from volatility control. An appendix to this chapter discusses
butane reactivity.
I. Description of Models and Inputs
Two computer models were used in the development of the
emissions inventories and ozone air quality analyses presented
in this chapter, M0BILE3 and EKMA. M0BILE3 is EPA's model for
calculating calendar-year, fleet-average motor vehicle emission
factors for various gaseous pollutants. These factors are
calculated for exhaust, evaporative, refueling, and total
hydrocarbon (HC) emissions, and are expressed in grams per mile
(g/mi). The differences between the released version of the
MOBILE3 model and the in-house version of M0BILE3 used to
perform the modeling for this analysis are discussed below.
These motor vehicle emission factors are then used as
inputs in the projection of future HC emission inventories.
These inventory projections, in turn, are used as inputs in the
determination of cost-effectiveness and future ozone air
quality. Future year projections are made assuming no control
of volatility and assuming alternative control levels.
Emission reductions, determined by comparing the "no-control"
and "control" inventory projections for the same year, are used
to calculate the estimated cost-effectiveness of a given
control strategy. This process is detailed in Chapter 6.
Changes in emission inventories over time, determined through
comparing the projection year inventory (with or without
control) to the base year inventory, are used as input data for
the EKMA (Empirical Kinetic Modeling Approach). EKMA uses this
and other information to predict future ambient ozone
concentrations in specific urban areas. Inputs used to project
future emission inventories are discussed in Section I.B., and
inputs for the ozone air quality analysis are discussed in
Section I.e. of this chapter.
-------
3-2
A. M0BILE3: Calculation of Emission Factors
The released version of the MOBILE3 emission factor model
and the standard assumptions used as input (e.g., temperatures)
are well-documented, and are not repeated here.[l] This
discussion focuses on the differences between the released
version of the model and the "in-house" version input data used
for this analysis.
The released version of M0BILE3 uses a number of standard
input assumptions about factors that affect emissions, such as
temperatures and fuel volatility. In a nationwide analysis
like this one, these standard inputs are generally used whether
one is modeling emission factors for the nation or for a
specific urban area, thus ignoring the effects of various
climates and regional differences in fuel volatility. However,
since temperature plays such a dominant role with respect to
evaporative emissions, city-specific data were gathered to
develop city-specific emission factors for each of the 61 urban
non-attainment areas examined. As is explained below, the
nationwide emissions inventory projections are also based on
this city-specific input data.
The two primary changes to the released version of M0BILE3
are the use of local temperatures and fuel RVP. The
temperatures specified in EPA's vehicle test procedures, an
ambient temperature range of 68° to 86°F (20° to 30°C), are
used by the released version of MOBILE3 model in calculating
temperature-dependent emissions.[2] These temperatures do not
account for the wide variation in temperatures for different
areas of the country. The fuel volatility of Indolene, EPA's
test fuel for certification testing, is 9.0 pounds per square
inch (psi) Reid Vapor Pressure (RVP). Although the released
MOBILE3 uses an assumed fuel volatility of 11.5 psi as more
representative of commercial fuels, this still is not
representative of current in-use fuel volatility, which survey
data show is even higher in some areas. As was seen in the
preceding chapter, both temperatures and fuel volatility have a
significant impact on the resulting emission factors.
Considering the importance of temperatures and fuel volatility
in determining vehicle hydrocarbon emissions, particularly
evaporative emissions, EPA has used city-specific data for both
temperature and volatility in the emission factors modeling for
this analysis. The data used are described in detail in
Sections I.A. l. and 2. of this chapter.
Another significant difference between the released
MOBILE3 model and the modeling done for this analysis pertains
to the treatment of alcohol-blend fuels (i.e., gasohol,
typically a blend of 90 percent gasoline and 10 percent ethyl
alcohol). All other things equal, the addition of ethanol to
gasoline results in a fuel with higher volatility (greater
-------
3-3
RVP). The standard inputs for released M0BILE3 do not account
for the use of alcohol-blend fuels and their emissions
impacts. As is discussed in Chapter 7, EPA is considering
several alternatives for volatility controls on alcohol-blend
fuels. In order to account for the impact of these
alternatives on hydrocarbon emissions, survey data on the
market penetration of alcohol-blend fuels m the urban areas
modeled are used. This is discussed in section I.A.3. of this
chapter.
Finally, as noted above, the standard input data for
released M0BILE3 is generally used to calculate emission
factors that are then applied both nationally and to specific
urban areas. Since the city-specific input data mentioned
above result in more realistic emission estimates, the
nationwide emission factors used in this analysis were
estimated by a population-weighted average of the city-specific
emission factors. This is described in section I.A.4.
The differences between the omission factors calculated
using the standard inputs and the released version of MOBILE3
model and those calculated for this analysis are also discussed
in the document "1983 NEDS Inventory Modifications," which is
available in the docket.[3]
l. City-Specific Temperature Data
The released version of MOBILE3 assumes that the ambient
temperature range that a vehicle is exposed to is 68° to 86°F
(20° to 30 °C), which are the temperatures specified for
certification testing. Vehicle hydrocarbon emissions,
especially evaporative emissions, are dependent on
temperature. Both the maximum temperature and the diurnal
temperature range have an impact on these emissions. While
temperature correction factors are included with released
MOBILE3, they generally have not been used in EPA's previous
regulatory impact analyses.
As was seen in section I.A. of Chapter 2, ozone is mainly
a summertime problem. Summertime temperatures over much of the
nation, including many of the 61 non-California urban
nonattainment areas, frequently exceed the 86°F maximum of the
test procedure. In at least some of those areas, the diurnal
temperature range of 24 F° (13 C°) used in EPA's evaporative
emission test procedure is exceeded. EPA has accounted for
this in its modeling of city-specific emission factors through
the use of climatic data reflecting actual temperature
observations in the areas modeled.
For different analyses, two sets of city-specific
temperature data were used. Temperatures recorded on the
design-value day were used in modeling environmental impacts.
-------
3-4
and July average temperatures were used in estimating some of
the fuel credits gained as part of the economic analyses. The
details of how these temperature data were incorporated into
the calculation of emission factors were discussed in Chapter 2.
The use of these design-value date temperatures in the
modeling of environmental impacts is based on the fact that the
primary purpose of in-use fuel volatility control is to address
widespread noncompliance with the ozone NAAQS. The goal is to
estimate the number of current nonattainment areas that might
achieve compliance with the ozone NAAQS under the various
control scenarios. The design value day is the day EPA has
used as the basis for modeling emission factors; thus, the
analysis looks at whether the projected ambient ozone
concentration for a specific urban area on a day like that on
which the design value was recorded will drop below 0.125 ppm,
meeting the NAAQS. If it does, the projection indicates that
the control scenario under which this occurs will bring that
area into attainment. Inventory projections are also based on
these temperatures, with the tonnage reduction projections used
in comparing and ranking various control programs.
The use of design-value date temperatures is inappropriate
for the economic analysis, because the economic benefits of
recovering evaporative emissions occur throughout the summer,
and not only on high-ozone days. Instead, average July
temperatures were used, both because they are more
representative of average summertime conditions, and because
they correspond to the in-use fuel volatility survey data that
were available for use in this study. Summer fuel RVP survey
data are unavailable for summer months other than July, so a
5-month weighting of temperature and RVP data could not be
performed. Thus, the assumption was made that the generally
higher RVP of summer fuel outside of July would be offset by
generally lower temperatures.
Emission inventories and emission reductions obtainable in
the projection years under the various control scenarios,
calculated using July-average temperatures, are used in the
process of estimating various recovery credits. This is
discussed in section IV of Chapter 5. These temperatures may
also be more appropriate for examining additional benefits of
ozone reductions, such as welfare effects or crop damage.
Table 3-1 lists the 61 urban non-California nonattainment
areas that were modeled, along with the date on which the ozone
design value was recorded and the temperatures recorded on
those dates. The dates on which the ozone design values were
recorded were supplied by EPA's Office of Air Quality Planning
and Standards (OAQPS).[4] Temperature data were obtained from
appropriate issues of the National Oceanic and Atmospheric
Administration's (NOAA) publications of "Monthly Climatological
Data."
-------
3-5
Table 3-1
Ozone Non-Attainment Area Design Value Dates and Temperatures
Design Value
(
°F}
MSA/CMSA Name
Date
Tmin
Tma:
Boston Metro Area
July 17, 1982
73
97
Greater Connecticut
Metro Area
July 30, 1983
72
90
New Bedford MA
June 9, 1984
67
88
Portland ME
June 6, 1983
51
74
Portsmouth-Dover-
Rochester NH-ME
April 29, 1983
44
80
Providence RI
June 9,1984
73
95
Springfield MA
April 28, 1983
49
86
Worcester MA
June 14, 1983
60
87
Atlantic City NJ
July 21, 1983
70
96
New York Metro Area
June 8, 1984
72
92
Vineland-Millville-
Bridgeton NJ
August 9, 1983
69
93
Allentown-Bethlehem
PA-NJ
July 30, 1983
70
94
Baltimore MD
August 8, 1983
72
96
Erie PA
June 11, 1983
56
84
Harrisburg-Lebanon-
Carlisle PA
June 13, 1983
64
88
Lancaster PA
September 6, 1982
49
80
Philadelphia Metro Area
August 17, 1983
62
87
Pittsburgh PA
August 3, 1983
64
87
Reading PA
August 17, 1983
60
88
Richmond-Petersburg
VA
April 27, 1984
54
86
Scranton-Wilkes Barre PA
April 28, 1983
49
83
Washington DC-MD-VA
June 24, 1983
62
85
York PA
June 23, 1983
53
88
Atlanta GA
July 14, 1983
74
96
Birmingham AL
June 25, 1982
60
92
Charlotte-Gastonia-
Rockhill NC-SC
August 23, 1983
73
103
Chattanooga TN-GA
August 22, 1983
73
100
Huntington-Ashland WV-KY-OH
July 15, 1983
61
96
Louisville KY-IN
July 12, 1983
70
93
Memphis TN-AR-MS
September 5, 1982
59
88
Miami-Hialeah FL*
August 2, 1982
81
96
Nashville TN
June 12, 1984
68
95
Tampa-St. Petersburg
Clearwater FL
June 4, 1984
67
88
-------
3-6
Table 3-1
(continued)
Ozone Non-Attainment Area Design Value Dates and Temperatures
Design Value (°F)
MSA/CMSA Name
Date
Tmin
Tmax
Akron OH
August 15, 1983
62
85
Canton OH
August 24, 1983
59
92
Chicago Metro Area
July 15, 1983
77
95
Cincinnati Metro Area
June 25, 1982
50
81
Cleveland OH
August 19, 19 83
66
91
Dayton-Springfield OH
August 21, 1983
72
100
Detroit MI
August 16, 1983
66
87
Grand Rapids MI
July 23, 1984
72
91
Indianapolis IN
July 25, 1982
65
88
Milwaukee Metro Area
August 28, 1984
69
94
Muskegon MI
June 26, 1983
66
90
Baton Rouge LA
September 19, 1982
73
91
Beaumont-Port Arthur TX
August 6, 1982
74
96
Brazoria TX
July 13, 1982
71
95
Dallas-Ft. Worth TX
August 27, 1983
76
96
El Paso TX
June 21, 1984
67
94
Galveston-Texas City TX**
July 21, 1982
81
91
Houston TX
August 30, 1983
76
93
Lake Charles LA
July 3, 1982
74
91
Long View-Marshall TX
August 1, 1983
69
96
New Orleans LA
June 7, 1982
68
93
San Antonio TX
September 3, 198
69
95
Tulsa OK
August 27, 1983
80
103
Kansas City MO-KS
August 15, 1984
77
96
St. Louis MO-IL
July 22, 1984
73
95
Denver-Boulder CO***
July 22, 1982
61
99
Salt Lake City-Ogden UT***
July 31, 1982
63
97
Phoenix AZ
August 3, 1984
85
107
* Actual design value is December 24, 1982. Instead of this
date, the highest ozone date at the same monitoring site
during the summer control period was used.
** Actual design value is October 16, 1982. Instead of this
date, the highest ozone date at the same monitoring site
during the summer control period was used.
*** High-Altitude Area.
-------
3-7
Examination of Table 3-1 reveals that the maximum
temperature on the ozone design value date is greater than 86°F
(30°C) for 52 of the 61 areas. The mean of the test procedure
and standard MOBILE3 input temperatures for calculating exhaust
emission factors, 77°F (25°C), is exceeded by the mean
temperature on the design-value date in 40 of the 61 areas.
The standard input diurnal temperature range for calculating
evaporative emission factors, 24 F° (13 C°), is exceeded by the
design-value date diurnal temperature range in 54 of the 61
areas. The uncontrolled diurnal index, discussed in section
II.A.l. of Chapter 2, indicates that these temperatures result
in in-use evaporative hydrocarbon emissions exceeding those
under the standard temperature conditions.
The July average temperatures for the 61 urban areas
modeled are shown in Table 3-2. These temperatures represent
the "average" July day for each area, based on the mean of 30
years' observations (1941-1970 inclusive).[5] For those cities
that did not have July average temperature data available,
estimates were made by using the d^.ta for nearby cities, taking
into account factors such as proximity to large bodies of water
(e.g., Milwaukee data rather than Grand Rapids, MI data were
used for Muskegon, MI). The cities used to estimate the
temperature data for other cities are summarized in Table 3-3.
Relative to the design-value date temperatures, the July
average temperatures are characterized by smaller diurnal
ranges and slightly lower maximum temperatures. However, both
the ranges and the maxima still tend to be greater than those
of the EPA test procedure and standard MOBILE3 model. As will
be shown in section II, the use of July average temperatures
result in total inventories that are up to 3 percent smaller
than those based on design value day temperatures. The
emission reductions associated with each level of RVP control
are also less when based on July average temperatures, relative
to design value temperatures. The difference in the
incremental (per 0.5 psi RVP) reductions associated with each
set of temperatures is also smaller; however, this difference
decreases at each step (0.5 psi) of control. At the lowest RVP
levels evaluated, the differences in reductions are very small.
Another approach that was considered for the temperature
input data was to use the averages of the temperatures of all
days recording exceedances of the ozone NAAQS for each city.
Due to the relatively small changes in the projections of
emissions inventories and reductions between the two sets of
temperature data described above, and the fact that such data
were not available in time for these analyses, only the design
value day and July-average temperatures were used in in this
analysis. However, the use of temperature data from all NAAQS
exceedance days will be considered in future analyses.
-------
3-8
Table 3-2
July Average Temperatures for 61 Ozone Nonattainment Areas
(
° F)
(
0 F )
MSA/CMSA Name
Tmax
Tmin
MSA/CMSA Name
Tmax
Tmi
Boston MA (A)
81
65
Akron OH (E)
85
62
Greater Conn (E)
84
61
Canton OH (E)
85
62
New Bedford MA (E)
81
63
Chicago IL (A)
83
61
Portland ME (A)
79
57
Cincinnati OH-KY-IN (A)
87
65
Portsmouth-Dover-
Cleveland OH (A)
82
61
Rochester NH-ME (E)
83
57
Dayton-Sprmgfleld OH (E)
87
65
Providence RI (A)
81
63
Detroit MI (A)
83
6 3
Springfield MA (E)
84
61
Grand Rapids MI (E)
83
63
Worcester MA (E)
81
65
Indianapolis IN (A)
85
65
Milwaukee WI (A)
80
59
Atlantic City NJ (A)
85
65
Muskegon MI (E)
80
59
New York NY (A)
85
68
Vineland-Mi1lvi1le -
Baton Rouge LA (E)
90
73
Bndgeton NJ (E)
85
65
Beaumont-Port Arthur TX (E)
94
73
Brazoria TX (E)
94
73
Allentown-
Dallas-Fort Worth TX (A)
96
74
Bethlehem PA-NJ (E)
87
67
El Paso TX (A)
95
70
Baltimore MD (A)
87
67
Galveston-Texas City TX (E)
94
73
Erie PA (E)
82
61
Houston TX (A)
94
73
Ham sburg-Lebanon -
Lake Charles LA (E)
90
73
Carlisle PA (E)
87
67
Long View-Marshall TX (E)
96
74
Lancaster PA (E)
87
67
New Orleans LA (A)
90
73
Philadelphia PA-NJ (A)
87
67
San Antonio TX (E)
96
74
Pittsburgh PA (A)
83
61
Tulsa OK (E)
93
70
Reading PA (E)
87
67
Richmond-Petersburg VA(A)
88
68
Kansas City MO-KS (A)
88
67
Scranton-
St. Louis MO-IL (A)
88
69
Wilkes Barre PA (E)
87
67
Washington DC-MD-VA (A)
88
69
Denver-Boulder CO (A,H)
87
59
York PA (E)
87
67
Salt Lake City-
Ogden UT(A,H)
93
61
Atlanta GA (A) 87 69 Phoenix AZ (A) 105 78
Birmingham AL (E) 87 69
Charlotte-Gastonia-
Rockhill NC-SC (A) 88 69
Chattanooga TN-GA (E) 87 69
Huntington-Ashiand
WV-KY-OH (E) 86 64
Louisville KY-IN (A) 87 66
Memphis TN-AR-MS (A) 92 72
Miami-Hialeah FL (A) 89 76
Nashville TN (A) 90 69
Tampa-St. Petersburg-
Clearwater FL (E) 89 76
(A) = Actual 30-year (1941-70) mean temperatures.
(E) = Estimated 30-year mean temperatures, from nearest appropriate city for
which actual temperatures were available.
(H) = High-el t i t'ide areas.
-------
3-9
Table 3-3
Area for which July average
temperatures were estimated
Greater Connecticut
New Bedford MA
Portsmouth-Dover-Rochester NH-ME
Springfield MA
Worcester MA
Vineland-Millville-Bndgeton NJ
Allentown-Bethlehem PA-NJ
Erie PA
Harrisburg-Lebanon-Carlisle PA
Lancaster PA
Reading PA
Scranton-Wilkes Barre PA
York PA
Birmingham AL
Chattanooga TN-GA
Huntington-Ashland WV-KY-OH
Tampa-St. Petersburg-Clearwater FL
Akron OH
Canton OH
Dayton-Sprmgfleld OH
Grand Rapids MI
Muskegon MI
Baton Rouge LA
Beaumont-Port Arthur TX
Brazoria TX
Galveston-Texas City TX
Lake Charles LA
Longview-Marshall TX
San Antonio TX
Tulsa OK
City temperatures used
Hartford CT
Providence RI
Concord NH
Hartford CT
Boston MA
Atlantic City NJ
Philadelphia PA
Cleveland OH
Philadelphia PA
Philadelphia PA
Philadelphia PA
Philadelphia PA
Philadelphia PA
Atlanta GA
Atlanta GA
Charleston WV
Miami FL
Columbus OH
Columbus OH
Cincinnati OH
Detroit MI
Milwaukee WI
New Orleans LA
Houston TX
Houston TX
Houston TX
New Orleans LA
Dallas-Fort Worth TX
Dallas-Fort Worth TX
Oklahoma City OK
-------
3-10
2. City-Specific Alcohol-Blend Market Shares
There are two reasons for using data on the market
penetration of alcohol-blend fuels for each of the cities
modeled. In general, alcohol-blend fuels using ethanol are of
higher volatility than the gasoline from which they are made.
Thus, accounting for the fraction of gasoline use that is
represented by alcohol blends results in more accurate emission
factors. In addition, EPA is considering several alternatives
for the treatment of alcohol blends under any regulatory
program for limiting gasoline volatility, including allowing
such blends a 1.0 psi RVP allowance above the gasoline
requirements. City-specific alcohol blend market fractions can
then be used to estimate the emissions impacts of such
alternatives, which are discussed in Chapter 7.
Alcohol blend market penetration data for July 1983 and
September 1985 were used in the modeling of emission factors.
The 1983 data were used in adjusting the base year ( 1983)
inventories, while the 1985 data (the most recent available at
the time of this analysis) were assumed to apply to all of the
projection years. These values are presented in Table 3-4.
As noted above, alcohol blend fuels will have higher
volatility than the base gasoline from which they are made.
The assumption made here is that the RVP of gasohol is 1.2 psi
greater than that of gasoline (in the absence of requirements
that it be otherwise, as discussed below). This offset in
volatility represents 1.0 psi due to splash blending of alcohol
and gasoline, and 0.2 psi due to the effects of commingling (of
straight gasoline and gasohol) in vehicle fuel tanks, as
discussed in section II.B. of Chapter 2. This difference is
assumed to apply to the 1983 base-year emission factors. (The
impact of the higher RVP of alcohol-blend fuels on motor
vehicle hydrocarbon emission factors was discussed in detail in
Chapter 2. )
The latest alcohol-blend market fraction data available
for use in this modeling were for September 1985. These data
were assumed to apply to all of the projection years. Alcohol-
blend fuel production and market penetration are sensitive to a
number of factors, among them the availability of state tax
exemptions, consumer acceptance, and crude oil prices. Market
penetrations of alcohol-blend fuels can be expected to vary
considerably in the future. However, most of the factors
impacting the rate of gasohol use are very difficult to
project, especially in the long term.
In using these data for the projection years, these
assumptions were used: Where alcohol blends are required to
meet the same RVP restrictions as gasoline, the in-use RVP of
-------
3-11
Table 3-4
Base-Year (1983) and Most Recent (1985)
Ethanol Fuel Market Fraction Data
(Source: Alcohol Week[6] unless noted)
MSA/CMSA Name July 1983 Sept 1985
Boston MA 0.0 0.0
Greater Connecticut 0.0 0.0
New Bedford MA 0.0 0.0
Portland ME 0.0 0.0
Portsmouth-Dover-Rochester NH-ME 0.0 0.0
Providence RI 0.0 0.0
Springfield MA 0.0 0.0
Worcester MA 0.0 0.0
Atlantic City NJ 0.0 0.0
New York NY 0.0 0.0
Vineland-Millville-Bridgeton NJ 0.0 0.0
A1lentown-Bethlehem PA-NJ 0.0 0.0
Baltimore MD 0.0 0.0
Erie PA 0.0 0.0
Harrisburg-Lebanon-Car1 isle PA 0.0 0.0
Lancaster PA 0.0 0.0
Philadelphia PA-NJ 0.0 0.0
Pittsburgh PA 0.0 0.0
Reading PA 0.0 3.0
Richmond-Petersburg VA 2.3 19.7
Scranton-Wilkes Barre PA 0.0 0.0
Washington DC-MD-VA 0.0 0.0
York PA 0.0 0.0
Atlanta GA 0.0 0.0
Birmingham AL 0.0 11.9
Charlotte-Gastonia-Rockhi11 NC-SC 2.4 0.6*
Chattanooga TN-GA 0.0 12.7
Huntington-Ashland WV-KY-OH 0.0 0.0
Louisville KY-IN 0.0 37.9*
Memphis TN-AR-MS 0.0 12.7
Miami-Hialeah FL 8.3 8.9
Nashville TN 0.0 12.7
Tampa-St.Petersburg-Clearwater FL 8.3 8.9
-------
3-12
Table 3-4
(continued)
Base-Year (1983) and Most Recent (1985)
Ethanol Fuel Market Fraction Data
(Source: Alcohol Week[6] unless noted)
MSA/CMSA Name July 1983
Akron OH 10 6
Canton 10.6
Chicago IL 4.4
Cincinnati OH-KY-IN 10 6
Cleveland OH 10 6
Dayton-Springfield OH 10.6
Detroit MI 21.7
Grand Rapids MI 21.7
Indianapolis IN 0.0
Milwaukee Wl 0.0
Muskegon MI 21.7
Baton Rouge LA 0.0
Beaumont-Port Arthur TX 0.0
Brazoria TX 0 0
Dallas-Ft. Worth TX 0 0
El Paso TX 0.0
Galveston-Texas City TX 0 0
Houston TX 0 0
Lake Charles LA 0.0
Long View-Marshall TX 0.0
New Orleans LA o.o
San Antonio TX 0.0
Tulsa OK 2.4
Kansas City MO-KS 0.0
St. Louis MO-IL 4.4
Denver-Boulder CO** 2.4
Salt Lake City-Ogden UT** 0.0
Phoenix AZ 0.0
Nationwide 3.7
Source: Alcohol Update.[7]
High-altitude area.
Sept 1985
1 . 8
1.8
29 9
1 8
1 . 8
1 . 8
13 . 0
13 0
32 . 7
0.7*
13 . 0
10.7*
4 . 1
4 1
4 . 1
4 . 1
4 1
4 . 1
10.7*
4 . 1
10 . 7
4 . 1
0 . 0
23 . 5
29 .9
16 . 2
0.8*
0 . 0
7.4*
-------
3-13
blends is assumed to be 0.2 psi greater than that of gasoline,
as a result of commingling. This was discussed in section
11. B. of Chapter 2. Where the impact of providing
alcohol-blends with a 1.0 psi RVP allowance was examined, the
in-use blend RVP was assumed to be 1.2 psi greater than that of
gasoline, reflecting both the 1.0 psi "allowance" and the
0.2 psi commingling effect.
Alcohol-blend fuel market penetrations and in-use
commercial fuel volatility (discussed in the following section)
have time-dependent trends. This situation differs from that
of temperatures, which vary unpredictably and without
directional trends over time. For both alcohol-blend market
penetrations and in-use RVP, the 1985 data were the latest
available for this analysis and are assumed to apply in all of
the projection years. Preliminary 1986 data on these factors
showed no significant differences from the 1985 data actually
used (on an national average basis), lending additional support
to the appplication of these data to the projection years.
The discussion above refers only to ethanol-blend fuels
(gasohol). Consideration was given to taking similar steps
with respect to commercially available (i.e., to retail
consumers) methanol-gasoline blends. However, since no
methanol blends are currently being sold at the retail level,
no effort was expended to include methanol-gasoline blends in
this environmental impact analysis.
3. City-Specific In-Use Fuel Volatility
The last of the city-specific input information used in
the calculation of emission factors was data on current in-use
fuel volatility (as of July 1985). This information was
derived from two sources, the annual fuel surveys conducted by
the Motor Vehicle Manufacturer's Association (MVMA), and the
American Society for Testing and Materials' (ASTM) voluntary
fuel volatility guidelines.[8,9]
The city-specific fuel RVP data used in the analysis are
shown in Table 3-5. All of the 1983 values are average
volatilities for non-alcohol-containing unleaded gasolines
taken from the MVMA Gasoline Survey for Summer 1983. [8] For
1985, the Summer 198? MVMA Gasoline Survey average volatility
of non-alcohol-containing unleaded gasoline was used, except
for those areas where the survey RVP was less than the
corresponding ASTM limit, in which cases the ASTM limit was
used. This was done under the assumption (supported by in-use
volatility trends since the mid 1970s) that in-use volatility
in those areas would continue to increase, at least until the
ASTM limits were reached. As with average July temperatures,
-------
3-14
Table 3-5
In-Use Fuel Volatility (RVP)
Data
In-Use RVP*
(psi)
MSA/CMSA Name
1983
1985
Boston MA
10 . 7
11 . 5
Greater Connecticut
10 . 7
11 . 5
New Bedford MA
10 . 7
11 . 5
Portland ME
10 . 7
11 . 5
Portsmouth-Dover-Rochester NH-ME
12 . 7
13 . 5
Providence RI
10 . 7
11. 5
Springfield MA
12 . 7
13 . 5
Worcester MA
10 . 7
11 . 5
Atlantic City NJ
10 . 6
11 . 5
New York NY
10 . 6
11 . 5
Vineland-Millville-Bridgeton NJ
10 . 6
11 . 5
Allentown-Bethlehem PA-NJ
10 . 6
11 . 5
Baltimore MD
11 . 0
11 . 5
Erie PA
11 . 5
12 . 1
Harrisburg-Leba^non-Carlisle PA
10 . 6
11 . 5
Lancaster PA
10 . 6
11 . 5
Philadelphia PA-NJ
10 . 6
11 . 5
Pittsburgh PA
11.5
12 . 1
Reading PA
10 . 6
11 . 5
Richmond-Petersburg VA
11.0
11. 5
Scranton-Wilkes Barre PA
12 . 6
13.5
Washington DC-MD-VA
11 . 0
11.5
York PA
10 . 6
11.5
Atlanta GA
10 . 1
10.3
Birmingham AL
11.6
11 . 8
Charlotte-Gastonia-Rockhi11 NC-SC
10 . 1
10 .3
Chattanooga TN-GA
10 . 1
10 .3
Huntington-Ashland WV-KY-OH
11.0
11.5
Louisville KY-IN
11.0
11. 5
Memphis TN-AR-MS
10 . 1
10 .3
Miami-Hialeah FL
10.3
11. 5
Nashville TN
11 . 6
11.8
Tampa-St.Petersburg-Clearwater FL
10 .3
11 . 5
In-Use RVP associated with ozone design value day.
-------
3-15
Table 3-5
(continued)
In-Use Fuel Volatility (RVP) Data
n-Use RVP* (psi)
MSA/CMSA Name
1983
1985
Akron OH
11.5
12. 1
Canton OH
11 . 5
12. 1
Chicago IL
11.7
11 . 9
Cincinnati OH-KY-IN
11.5
12 . 1
Cleveland OH
11.5
12 . 1
Dayton-Springfield OH
11.5
12. 1
Detroit MI
11 .3
12.0
Grand Rapids MI
11.7
11 . 9
Indianapolis IN
11.7
11 . 9
Milwaukee WI
11.7
11 . 9
Muskegon MI
11.7
11 . 9
Baton Rouge LA
10 . 2
10 . 9
Beaumont-Port Arthur TX
9 . 9
10 . 7
Brazoria TX
9 . 9
10 . 7
Dallas-Ft.Worth TX
9.9
10 . 7
El Paso TX
8 . 8
9 . 5
Galveston-Texas City TX
10 .2
10 . 5
Houston TX
10 .2
10.5
Lake Charles LA
10.2
10 . 9
Long View-Marshall TX
9.9
10 . 7
New Orleans LA
11.7
12. 4
San Antonio TX
10 .2
10 . 5
Tulsa OK
9.7
10.3
Kansas City MO-KS
9 . 7
10 . 3
St. Louis MO-IL
10 .4
10 . 4
Denver-Boulder CO**
10 . 0
10 . 2
Salt Lake City-Ogden UT**
9 . 6
10 . 0
Phoenix AZ
8 . 6
9 . 0
In-Use RVP associated with ozone design value day.
High-altitude area.
-------
3-16
RVP information was not available for all of the 61 urban
areas. In those non-survey areas for which no in-use
volatility data were available, the RVP of the nearest survey
area, based on the gasoline distribution system network of
pipelines and commonality of ASTM classification, was used for
the volatility of the non-survey urban areas.
The RVPs listed in Table 3-5 are the RVPs for the month in
which each areas' design value day occurred. The summer
gasoline surveys are taken in July, whereas the design value
days occur in different months for the 61 urban areas For six
areas*, the ASTM classification for the month in which the
design value day occurred and the July ASTM classification are
not the same. Therefore, an additive correction was made to
the July survey RVP for these areas to adjust the RVP level to
the correct ASTM class for the month in which the ozone design
value day occurred. For example, the July 1985 survey RVP for
Nashville is 10.3 psi, during an ASTM classification C/B month
(Class B fuel was assumed, since 10.3 psi is much closer to the
Class B limit than to the Class C limit). The design value day
for Nashville occurred in June, an ASTM classification C
month. So the RVP used for this analysis was adjusted up to
11.8 psi. The additive 1.5 psi upward adjustment is the
difference between ASTM volatility specifications for Class B
and Class C fuels of 10.0 psi and 11.5 psi, respectively. As
with the city-specific alcohol-blend market penetration data
discussed in the preceding section, the 1983 RVP data were used
in adjusting the base-year inventories and the 1985 data were
assumed to apply in all of the projection years.
4. Nationwide HC Emission Factors
As mentioned in the introduction to this chapter, previous
EPA analyses of mobile source programs have generally assumed
that the same input data, and thus emission factors, apply on
both local (city-specific) and national scales. The preceding
sections have presented city-specific information used in
calculating hydrocarbon emission factors for each of the urban,
non-California ozone nonattainment areas modeled for this
analysis. This section discusses the incorporation of that
information in the calculation of nationwide emission factors.
The resulting fleetwide exhaust, evaporative, refueling, and
total hydrocarbon emission factors are presented in a series of
tables. Total hydrocarbon emission factors for each gasoline-
fueled vehicle category (light-duty vehicles, light-duty
trucks, and heavy-duty vehicles) are also presented.
Portsmouth-Dover-Rochester, NH-ME; Springfield, MA;
Scranton-Wilkes Barre, PA; Birmingham, AL; Nashville, TN;
and New Orleans, LA.
-------
3-17
The nationwide emission factors presented below are based
on a population-weighted average of the emission factors
calculated for the 61 urban non-California ozone nonattainment
areas modeled. The 1980 census populations of the 61 areas are
shown in Table 3-6.
Tables 3-7 through 3-14 present the nationwide emission
factors used in this analysis, based on the city-specific
inputs and population weighting described above. The total HC
emission factors (sum of exhaust, evaporative, and refueling HC
emissions) are presented for light-duty gasoline-fueled
vehicles (LDGVs) in Table 3-7, for light-duty gasoline-fueled
trucks {LDGTs) with gross vehicle weight ratings (GVWR) of
6.000 lbs or less (LDGTls) in Table 3-8, for LDGTs with GVWR of
6.001 lbs and up (LDGT2s) in Table 3-9, and for heavy-duty
gasoline-fueled vehicles (HDGVs) in Table 3-10.
Tables 3-11 through 3-14 present nationwide fleet exhaust,
evaporative, refueling, and total HC emission factors,
respectively. These factors also include the contributions of
diesel-powered vehicles, appropriately weighted by vehicle
miles traveled (VMT) fractions. Since the controls on m-use
gasoline volatility that are evaluated in this analysis would
have no impact on hydrocarbon emissions from diesel vehicles,
the diesel vehicle emission factors that are weighted into the
fleetwide factors are those calculated using standard MOBILE3
inputs.[1] The VMT fractions used for the different vehicle
categories and for urban and rural travel are those developed
using EPA's MOBILE3 Fuel Consumption Model (M3FCM), which is
further discussed in the following section.[10]
All of the hydrocarbon emission factors for gasoline-
fueled vehicles presented below include the assumption that an
onboard refueling emission control requirement will be
effective in 1990 and later model years. Emission factors have
also been calculated under the assumption that no onboard
refueling controls will be required. These factors, which were
used in the sensitivity and economic analyses, are not
reproduced here.
B. Calculation of Inventories
Calculation of emission inventories for each of the
projection years is necessary m order to estimate the emission
reductions resulting from various control strategies, and as
input for the ozone air quality modeling. A "No RVP Control"
-------
3-18
Table 3-6
Populations of the 61 Non-California
Urban Non-Attainment Areas Used in
Calculation of Nationwide Emission Factors
Populat ion
MSA/CMSA Name
(1980
Census)
New York Metro Area
17
190
ii
221
Chicago Metro Area
7
794
851
Philadelphia Metro Area
4
716
818
Detroit MI
4
488
072
Boston Metro Area
4
136
612
Washington DC-MD-VA
3
250
822
Dallas-Fort Worth TX
2
930
516
Houston TX
2
735
766
Pittsburgh PA
2
218
870
Greater CT Metro Area
2
206
318
Baltimore MD
2
199
531
Atlanta GA
2
138
231
Cleveland OH
1
898
825
St Louis MO
1
808
621
Cincinnati Metro Area
1
660
278
Miami-Hialeah FL
1
625
781
Denver-Boulder CO
1
618
461
Tampa-St. Petersburg-Clearwater FL
1
613
603
Phoenix AZ
1
509
052
Milwaukee Metro Area
1
498
078
Kansas City MO-KS
1
433
458
New Orleans LA
1
256
256
Indianapolis IN
1
166
575
San Antonio TX
1
071
954
Charlotte-Gastonia-Rock Hill NC-SC
971
391
Louisville KY
956
756
Dayton-Springfield OH
942
083
Memphis TN
913
472
Salt Lake City-Ogden VT
910
222
Birmingham AL
883
946
Nashville TN
850
505
Richmond-Petersburg va
761
311
Scranton-Wilkes-Barre PA
728
796
Is
-------
3-19
Table 3-6
(continued)
Population
MSA/CMSA Name (1980 Census)
Akron OH 660 328
Tulsa OK 657 173
A1 lentown-Bethlehem PA 635 481
Providence RI 618 514
Grand Rapids MI 601 680
Harrisburg-Lebanon-Carlisle PA 555 158
Springfield MA 515 259
Baton Rouge LA 494 151
El Paso TX 479 899
Chattanooga TN 426 540
Canton OH 404 421
Worcester MA 402 918
York PA 381 255
Beaumont-Port Arthur TX 375 497
Lancaster PA 362 346
Huntington-Ashland WV 336 410
Reading PA 312 509
Erie PA 279 780
Atlantic City NJ 276 385
Galveston-Texas City TX 195 940
Portland ME 193 831
Portsmouth-Dover-Rochester NH-ME 190 938
Brazoria TX 169 587
Lake Charles LA 167 223
New Bedford MA 166 699
Muskegon MI 157 589
Longvlew-Marshall TX 151 752
Vineland-Millville-Bridgeton NJ 132 866
-------
3-20
Table 3-7
Weighted National Average Total HC
Emission Factors — LDGVs (q/mi)
Strategy Proiection Year
(RVP, in psi)
1983
1990
1995
2000
2010
No RVP Control
5 . 03
2 . 70
1 .94
1 72
1 . 68
11.5
2. 50
1 . 5b
1 . 20
1 . 07
11.0
2 . 29
1 . 44
1 . 14
1 . 04
10 . 5
2 . 11
1 .34
1 . 09
1 . 01
10 . 0
1 .98
1.27
1 . 05
0 . 99
9.5
1 . 88
1.21
1 . 02
0 . 97
9 . 0
1 . 79
1 . 16
0 . 99
0 .95
8.5
1 . 76
1 . 14
0.97
0.93
8.0
1 . 72
1.11
0 . 95
0 . 92
-------
3-21
Table 3-8
Weighted National Average Total HC
Emission Factors — LDGTls (q/mi)
Strategy
Projection
Year
(RVP, in psi)
1983
1990
1995
2000
2010
No RVP Control
7 . 08
4 . 40
3 . 04
2 .32
2 . 10
11.5
4 .20
2 . 73
1 . 87
1 . 56
11.0
3 .98
2 . 57
1 . 80
1 . 53
10. 5
3 79
2 .45
1 . 73
1 . 49
10 0
3 64
2 . 36
1 68
1 46
9.5
3 . 52
2. 29
1 . 63
1 . 43
9.0
3.42
2.22
1 . 58
1 .40
8.5
3.36
2 . 18
1 . 55
1 .37
8.0
3 .32
2 . 14
1 . 52
1.35
-------
3-22
Table 3-9
Weighted National Average Total HC
Emission Factors — LDGT2s (q/mi)
Strategy Proiection Year
-------
3-23
Table 3-10
Weighted National Average Total HC
Emission Factors — HDGVs (q/mi)
Strategy Projection Year
(RVP, in psi)
1983
1990
1995
2000
2010
No RVP Control
14 . 63
7 . 94
5.68
5.03
4 . 77
11.5
7 . 63
4 .88
4 . 09
3. 78
11.0
7.21
4 . 71
3 .99
3 . 71
10 . 5
6.83
4 . 56
3 .89
3 . 63
10 . 0
6.51
4 . 42
3 . 80
3 . 57
9 . 5
6.23
4 . 29
3 . 72
3 . 50
9.0
5.97
4 .17
3.64
3.43
8.5
5 . 77
4 . 07
3 . 56
3 . 37
8.0
5 58
3.98
3.49
3 .31
-------
3-24
Table 3-11
Weighted National Average Exhaust HC
Emission Factors — All Vehicles (q/mi)
Strategy
Projection Year
(RVP, in psi)
1983
1990
1995
2000
2010
No RVP Control
3.60
1. 63
1 . 12
0 .96
0 . 93
11. 5
1 .62
1 . 09
0.90
0 . 86
11. 0
1 . 61
1. 08
0 . 90
0 . 86
10 . 5
1 . 59
1 . 07
0 . 90
0 . 86
10. 0
1 . 58
1 . 06
0 . 89
0 . 86
9.5
1 . 57
1 . 06
0 . 89
0.86
9 . 0
1 . 55
1 . 05
0 .89
0 . 86
8.5
1 .55
1 . 05
0 . 89
0.86
8,0
1 . 55
1 . 05
0 . 89
0 . 86
-------
3-25
Table 3-12
Weighted National Average Evaporative HC
Emission Factors — All Vehicles (g/mi)
Strategy
(RVP, in psi)
1983
Projection Year
1990
1995
2000
2010
No RVP Control
2 . 08
1 . 36
1.02
0 . 89
0 . 86
11.5
1 . 19
0 . 71
0 . 51
0 . 43
11. 0
1 .01
0 . 62
0 . 46
0 . 40
10.5
0 . 86
0 . 54
0 42
0 . 37
10.0
0 . 76
0 . 49
0 .38
0 . 35
9.5
0 . 67
0 .44
0 .35
0 . 33
9.0
0 . 61
0 .40
0 . 33
0.31
8.5
0 . 57
0.37
0 . 30
0 . 29
8.0
0 . 54
0.35
0 . 28
0 . 27
-------
3-26
Table 3-13
Weighted National Average Refueling HC
Emission Factors — All Vehicles (q/mi)
Strategy Projection Year
(RVP, in psi)
1983
1990
1995
2000
2010
No RVP Control
0.36
0 .27
0 . 13
0 . 06
0 . 03
11. 5
0 .27
0 . 13
0 . 06
0 . 03
11.0
0 . 26
0 . 12
0 . 06
0 . 03
10 . 5
0 . 25
0 . 12
0 . 06
0 . 03
10 0
0 . 24
0.11
0 05
0 03
9.5
0.22
0.11
0 . 05
0 . 03
9.0
0.21
0 . 10
0 .05
0 . 03
8 . 5
0 .20
0 . 10
0 . 05
0 . 03
8.0
0 . 19
0 . 09
0 . 05
0 . 03
-------
3-27
Table 3-14
Weighted National Average Total HC
Emission Factors — All Vehicles (g/mi)
Strategy
Proiection Year
(RVP, in psi)
1983
1990
1995
2000
2010
No RVP Control
6 . 04
3.27
2.26
1.91
1 . 83
11. 5
3 . 08
1.92
1 .47
1 .32
11. 0
2.88
1. 82
1 . 42
1 . 29
10 . 5
2 . 71
1. 73
1 .37
1 . 26
10 . 0
2 . 57
1 . 66
1.33
1 . 24
9 . 5
2 . 46
1 . 60
1 .30
1 . 22
9 . 0
2.37
1 . 55
1 . 26
1 . 19
8.5
2.32
1 . 52
1 . 24
1 . 17
8.0
2 . 28
] . 49
1 . 22
1 . 16
n p 10 '
m
rD1^
-------
3-28
case is developed in which no additional emission controls
beyond those already implemented are assumed to be in effect.
Comparison of projected inventories under this base case to
those projected under different control cases provides
estimated emission reductions.
Inventories for the projection years under the "No RVP
Control" case and the various control cases under consideration
are also a key input for the ozone non-attainment analysis.
EPA used the Empirical Kinetic Modeling Approach (EKMA) to
estimate the ambient ozone concentrations m the urban ozone
non-attainment areas evaluated in this analysis This is
discussed m the section I.e., below.
Projection of emission inventories requires that a
base-year inventory (1983 for this analysis) be supplied, along
with base-year ozone design values (ambient concentrations,
expressed in parts per million (ppm)), and assumptions
describing future growth and anticipated emission reductions
for each component of the inventory.
The inventory is divided into mobile and stationary source
subtotals, each of which m turn is further divided into a
number of individual source categories. This section of the
chapter describes the base-year emissions inventory, the growth
rates and controls assumed in the projection of future emission
inventories, and the other inputs to the EKMA for the ozone
non-attainment analysis. These are followed by sections
presenting the results of the nationwide emissions analysis and
the ozone air quality (EKMA) analysis.
1. Base Year (1983) Emissions Inventory
The base-year non-methane hydrocarbon (NMHC) emissions
inventory used for these analyses is for calendar year 1983,
the most recent year for which official design values were
available at the time of this analysis. This base-year
inventory is based on that developed by EPA's National Air Data
Branch within the Office of Air Quality Planning and Standards
(OAQPS), and is referred to as the NEDS (National Emissions
Data System) inventory.
Table 3-15 presents the nationwide base year inventory
used in this analysis, and Table 3-16 presents the base year
inventory for the 61 non-California urban non-attainment areas
analyzed. The differences between the inventories taken from
NEDS and the inventories presented in Tables 3-15 and 3-16
involve the gasoline-related stationary source and the mobile
source components. These adjustments are described below.
More detailed information on the adjustments to the inventory
is available in an EPA technical memorandum.[3]
-------
3-29
Table 3-15
Base Year (1983) Nationwide NMOC Emissions Inventory
NMOC Emissions
Source Category (x 1000 T)
Light-Duty Vehicles (LDV) 5946.3
Light-Duty Trucks (LDT) 2686.4
Heavy-Duty Gasoline-Fueled Vehicles (HDGV) 868.2
Heavy-Duty Diesel-Powered Vehicles (HDDV) 486.4
Off-Highway Vehicles* 1209.5
Stage I 252.5
Remainder (Bulk Storage) 446.9
Area Sources* 9340.7
Point Sources* 2188.8
Mobile Source Subtotal 11196.8
Stationary Source Subtotal 12228.9
Gasoline-related Subtotal** 10200.3
Non-gasoline-related Subtotal** 13225.4
TOTAL EMISSIONS 23425.7
* The components of these categories are described in the text
** Gasoline-related = LDV + LDT + HDGV + stage I + Remainder;
non-gasoline-related = HDDV + Off-highway + Area + Point.
-------
3-30
Table 3-16
Base Year (1983) NMOC Emissions Inventory for
61 Non-California Urban Ozone Non-Attainment Areas
NMOC Emissions
Source Category (x 1000 T)
Light-Duty Vehicles (LDV) 2332.9
Light-Duty Trucks (LDT) 752.7
Heavy-Duty Gasoline-Fueled Vehicles (HDGV) 300.6
Heavy-Duty Diosel-Powered Vehicles (HDDV) 184.0
Off-Highway Vehicles* 311.5
Stage I 101.4
Remainder (Bulk Storage) 181.2
Area Sources* 2815.4
Point Sources* 915.1
Mobile Source Subtotal 3881.7
Stationary Source Subtotal 4013.1
Gasoline-related Subtotal** 3668.8
Non-gasoline-related Subtotal** 4226.0
TOTAL EMISSIONS 7894.7
* The components of these categories are described in the text.
** Gasoline-related = LDV + LDT + HDGV + Stage I + Remainder;
non-gasoline-related = HDDV + Off-highway + Area + Point.
-------
3-31
The NEDS base-year inventories for light-duty vehicles
(LDVs), light-duty trucks (LDTs), and heavy-duty gasoline-
fueled vehicles (HDGVs) were adjusted to account for
city-specific data on temperatures, market penetration of
alcohol-blend fuels, and in-use volatility, as described in
sections I.A.I. , 2., and 3. of this chapter, respectively.
This adjustment consisted of multiplying the NEDS inventories
for each vehicle class by the ratio of the emission factors
incorporating the city-specific input data to the emission
factors using standard input data, which are used by NEDS.
These inventories, as well as that for heavy-duty diesel
vehicles (HDDVs), are also based on the vehicle miles traveled
(VMT) data from the MOBILE3 Fuel Consumption Model (M3FCM)
rather than on NEDS' VMT data.[10] This adjustment is
performed by multiplying the inventories by the ratio of the
1983 M3FCM VMT to the NEDS VMT.
The other significant change is in the Stage I and
Remainder (bulk storage) categories' inventories. As described
in section I.B.3. of this chapter the base-year inventories
for these categories are based on information developed in
EPA's response to comments received on the gasoline marketing
study.[11]
The categories of concern in this analysis are, of course,
those directly related to gasoline. As listed in Tables 3-15
and 3-16, these are LDVs, LDTs, HDGVs, Stage I, and Remainder
(bulk storage). Emissions from HDDVs are included in the
mobile source emissions totals, but not as gasoline-related
emissions. The "off-highway" mobile source category, which
also is assumed not to be gasoline-related, consists of
emissions from aircraft, locomotives, inboard- and outboard-
powered vessels, small general-utility engines, agricultural
and heavy-duty construction equipment, and snowmobiles. The
bulk of these sources are unregulated and are not gasoline-
fueled. More information on these sources and their emissions
is available in the most recent AP-42 document.[1]
The vast majority of stationary source emissions are in
the categories shown in Tables 3-15 and 3-16 as "Area" and
"Point" sources. These categories include a number of
industrial processes (e.g., manufacture of plastic and rubber
products, polymers, resins, synthetic fibers, pharmaceutical
products), many industrial surface coating processes (e.g.,
large appliances, various metal products, large ships and
aircraft), the use of solvents (e.g., dry cleaning, adhesives,
degreasing, graphic arts), and other miscellaneous sources
(e.g., solid waste disposal, open forest and agricultural
burning). For these many stationary sources, 1983 NEDS
inventories are used in this analysis. None of these "area"
and "point" source emissions will be affected by gasoline
volatility controls.
-------
3-32
2. Mobile Source VMT Growth Rates and Emission Controls
Given the base-year inventories for the various mobile
source categories, future inventories are projected using
assumptions about anticipated VMT growth rates and emission
controls. The emission factors calculated by MOBILE3 and
presented in the previous section incorporate the effects of
the existing mobile source emission control program (i.e.,
fleet turnover leading to replacement of older vehicles with
later-model vehicles subject to more stringent emission
standards). The "No RVP Control" emission factors shown in
Tables 3-7 through 3-14 assume continuance of the existing
mobile source control program; the various RVP control
scenarios include the effects of both the existing program and
various levels of in-use and certification fuel volatility
(RVP) control.
In all of these cases, the existence of vehicle inspection
and maintenance (I/M) programs is assumed for all ozone
non-attainment areas. As noted previously, these emission
factors also assume implementation of onboard (vehicle-based)
controls on refueling emissions, beginning in the 1990 model
year.
The growth assumed for the LDV, LDT, HDGV, and HDDV
categories is based on the urban VMT projections from the
MOBILE3 Fuel Consumption Model (M3FCM).[10] The VMT for the
projection years is calculated by the M3FCM by applying an
annual compound VMT growth rate to the base-year VMT.
Projected VMT by category and projection year, based on the
M3FCM, are summarized in Table 3-17.
Another factor that has an important influence on the
mobile source emission projections is the division of VMT into
urban and rural fractions and the growth rates assumed to be
applicable to urban VMT. In the NEDS base-year mobile source
inventories, fleet emission factors are based on urban and
rural VMT breakdowns for only four vehicle categories (LDV,
LDT, HDGV, and HDDV). Thus, the M3FCM VMT projections, which
are calculated for seven vehicle categories (LDGV, LDDV, LDGTi,
LDGT2, LDDT, HDGV, and HDDV), are weighted to correspond to
these categories. These weightings are then used with in the
adjustment of the bare-year inventories.
In this analysis about 54.6 percent of total 1983
nationwide VMT is urban VMT, while in the 61 urban areas
modeled about 81.7 percent of total 1983 VMT is urban.[3] For
both the nationwide and urban area analyses, the urban VMT
fraction is assumed to increase by 0.358 percent annually. The
base-year NEDS inventories for the mobile source categories
were adjusted using fleetwide emission factors that reflect
-------
3-33
Table 3-17
Annual VMT by Vehicle Class (x 109 mi)
Year
Vehicle
Class
TOTAL*
LDV
LDT
HDGV
HDDV
1983
1031.80
341.01
55.45
97 . 48
1525.74
1990
1187.53
397.67
55. 17
120 . 44
1760.81
1995
1298.75
438.60
55.93
137.81
1931.08
2000
1409.98
479 .29
58 . 70
154.28
2102.25
2010
1632.44
560.23
68.21
182.83
2443 .71
* Class VMTs may not add exactly to totals due to rounding.
-------
3-34
this urban/rural division of VMT and urban VMT growth rate.
More information on urban/rural VMT breakdowns, urban VMT
growth, and the resulting adjustment to the base-year inventory
is provided in reference [3].
3. Stage I and Remainder Growth Rates and Controls
The impacts of in-use RVP control on Stage I and Remainder
(bulk storage) emissions were discussed in section II.C. of
Chapter 2. This section discusses three other aspects of the
modeling of these emissions: the base-year inventories used,
future growth rates assumed applicable to these categories, and
emission reductions expected to occur as a result of additional
application of control technology to these sources (excluding
RVP control).
The base-year emissions inventories for Stage I and bulk
storage were taken from EPA's Response to Comments to the
Gasoline Marketing Study, with a few minor adjustments that are
described below.[11] These inventories, which are referred to
in this section as the "GEMS inventories" (to distinguish them
from the NEDS inventories used for other stationary sources),
were presented m reference [11] in terms of thousands of
megagrams (x 1000 Mg) for the year 1984. In addition, these
inventories assume an average summertime RVP of 11.5 psi. Each
of these results in an adjustment to the GEMS inventories being
required before substituting them for the corresponding NEDS
inventories.
These inventories were adjusted from the 11.5 psi RVP used
in the gasoline marketing analysis to 11.0 psi RVP, which is
the population-weighted national average RVP for 1983 used in
this analysis (see section I.A.3. of this chapter). An
adjustment was also made to base the inventories on 1983, the
base year of this analysis, rather than 1984 as presented in
reference [11]. The adjustment factor for changing the base
year is the ratio of fuel consumption in 1983 to fuel
consumption in 1984, using fuel consumption results from the
MOBILE3 Fuel Consumption Model.[10] The use of fuel
consumption to model growth in these emissions rests on the
idea that gasoline storage and distribution losses should be
directly proportional to total gasoline throughput, all other
factors being constant. This is also discussed under growth
rates, below.
The resulting base year inventories for these categories
are 252,500 tons from Stage I and 446,900 tons from bulk
storage. These figures are included in the base-year
nationwide nationwide inventories on which future projections
are based. The city-specific base-year emissions inventories
-------
3-35
for these two categories were developed by adjusting the
corresponding city-specific NEDS inventories in the same way.
The growth rates used to model future Stage I and
Remainder inventories are based on fuel consumption projections
from the M0BILE3 Fuel Consumption Model. Previous modeling of
these categories has generally assumed positive annual growth
rates based on economic growth projections for large sectors of
the economy. However, Stage I and Remainder emissions
logically are more a function of total gasoline throughput (and
thus of total gasoline consumption) than of general economic
growth. Therefore, in addition to substituting the "GEMS"
Stage I and Remainder inventories for those supplied by NEDS,
growth rates derived from MOBILE3 Fuel Consumption Model
projections of total gasoline consumption are used for these
categories.
The fuel consumption projections used predict that
gasoline consumption will gradually decline through the
remainder of this century, from base-year consumption of about
94.16 x 109 gallons in 1983 to about 74.73 x 109 gallons in
2000. Projected gasoline consumption reaches a minimum of
about 74.51 x 109 gallons in 2002, after which it is
predicted to slowly increase, to about 78.63 x 109 gallons in
2010. Thus, the growth rate applied to Stage I and Remainder
emissions is negative through 2002. then positive through 2010.
The net effect of the use of GEMS inventories and MOBILE3
Fuel Consumption Model-based growth rates on the Stage I and
Remainder categories is to reduce these emission considerably,
both in absolute terms and as a fraction of total NMHC
emissions, relative to the levels included in the original
volatility study.[12]
4. Stationary Source Growth Rates and Controls
Since Stage I and Remainder emissions have been handled
separately in this analysis, the remaining stationary sources
of NMHC emissions are the "Area" and "Point" categories of
Tables 3-15 and 3-16. The distinction between the area and
point categories is made on the basis of annual emission
totals, with smaller individual emission sources designated as
"area" and larger sources as "point."
There are many individual contributors to area and point
source NMHC emissions. They can broadly be classified as
emissions from industrial manufacturing processes (e.g.,
plastic and rubber products, polymers and resins, etc.),
emissions from surface coating operations (e.g., of large
appliances, ships, and airplanes), and miscellaneous sources
-------
3-36
(e.g., open burning and solid waste disposal). These numerous
categories share in common the fact that they are not
gasoline-related, and thus that emissions from these sources
will be unaffected by any RVP controls.
The emissions inventories from these categories for the
base year were taken from the 1983 NEDS inventory. These
inventories are shown for the nation m Table 3-15 and for the
61 non-California urban non-attainment areas in Table 3-16.
Growth and replacement rates applied to these categories are
based on information supplied by EPA's Office of Air Quality
Planning and Standards, and are described briefly here.
As noted above, there are many individual sources included
in the "area" and "point" categories of this analysis. The
annual growth rate applied to industrial manufacturing
processes, industrial surface coating operations, and other
solvent uses are 3.3, 3.1, and 0.8 percent, respectively.
These rates are applied to the relevant emission sources listed
under both area and point sources. Remaining point sources
(not included in the three sub-categories noted above) are
assumed to increase by 1.9 percent annually, and area sources
not covered by these three sub-categories have no growth
assumed.
Regardless of whether m-use RVP control is implemented,
EPA, states, and local areas have established equipment-related
controls for stationary sources that must be accounted for in
modeling future hydrocarbon emissions. The Clean Air Act as
amended in 1977 requires that hydrocarbon emissions from both
new and existing stationary sources in ozone non-attainment
areas be controlled to the lowest achievable levels. EPA has
interpreted this as those levels achievable with RACT, which
varies from source to source.
To assist the states in developing control regulations
consistent with RACT levels, EPA's Office of Air Quality
Planning and Standards (OAQPS) issued a series of control
technique guideline (CTG) documents relevant to various
sources. These CTGs assessed the technology available to
control HC emissions from various sources, and provided
estimates of the emission rates achievable with the full
implementation of RACT-based controls.
The Clean Air Act Amendments of 1977 also stated that all
areas were to be m compliance with the ozone National Ambient
Air Quality Standard (NAAQS) by December 31, 1982; therefore,
this date was originally projected as the year by which RACT
levels of control would be fully implemented on sources in
-------
3-37
the non-attainment areas of the late 1970s and early 1980s. In
the 1985 volatility study, EPA made certain assumptions
regarding the rate at which RACT-based emission controls would
be implemented; essentially, the assumption was that one to
five percent of the emission reductions available through the
application of RACT controls were in effect as of 1982 (the
base year used in that analysis), and that all remaining
available controls would be implemented by 1988.[12] This
resulted in the highest projections of future emission
reductions expected to result from additional implementation of
RACT and the lowest projected ozone values.
EPA's earlier assumptions regarding the degree to which
RACT-based controls have already been applied and the extent of
emission reductions still available from further RACT
application were erroneous. EPA's current best estimates
indicate that, while RACT-based emission reductions have
already been achieved from some stationary sources, further
reductions are still available from other components of the
stationary source inventory.
Specifically, the modeling now assumes that no further
emission reductions can be obtained through the implementation
of RACT controls on area source solid waste disposal,
miscellaneous area sources, or other solvent use (i.e., solvent
use other than industrial surface coating operations). CTGs
yet to go into effect are assumed to yield reductions amounting
to two percent of the total 1983 emission inventory, from these
components of the stationary source inventory: Stage I and
remainder (bulk storage), petroleum refineries, industrial
processes (mostly chemical manufacturing), and industrial
surface coating operations. A reduction of two percent in the
total inventory is equivalent to a 12 percent reduction from
these specific sources. Compared to the previous analysis,
these assumptions result in smaller reductions being obtained
through RACT, and greater projected inventories for the
affected stationary source categories.
EPA is continuing to evaluate the entire issue of RACT
control implementation for stationary sources as part of the
Agency's efforts to comprehensively address ozone control
strategies; however, the assumptions made in this analysis
represent the best estimate at this time.
Replacement rates refer to the projected replacement of
older (i.e., higher emission) equipment and facilities by newer
(i.e., low emission) equipment and facilities. Emission
reductions due to application of RACT apply to such older
equipment (see below), while the newer replacement equipment is
subject to new source performance standards (NSPS). In this
-------
3-38
analysis, emissions from a given source under full
implementation of RACT are assumed to be equal to emissions
from a replacement for that source subject to NSPS; therefore,
the replacement rates used do not affect projected emissions
from stationary sources.
C. EKMA: Ozone Air Quality Analysis
The ozone air quality analysis for this study was
conducted using the Empirical Kinetic Modeling Approach
(EKMA). EKMA uses projected future emission inventories and
other input data to estimate ambient ozone concentrations,
NAAQS compliance status, and numbers of expected exceedances
for specific urban areas future years. Additional EKMA input
data are described below. The results of the EKMA modeling are
presented in section III of this chapter.
1. Background Ambient Ozone Concentration
One of the inputs to EKMA is the background ozone level.
This is an estimate of the ambient ozone concentration that
would exist as a result of naturally occurring emissions of
ozone precursors, m the absence of the anthropogenic
(man-made) emissions included in the inventory modeling.
Previous analyses, including those performed as part of the
original volatility study, have assumed no background ozone
concentration. This analysis has been revised to incorporate
the assumption of a 0.07 parts per million (ppm) background
ozone concentration, reflecting EPA's current best estimate.
2. NMOC:NOx Ratios
Another input required for the EKMA model to project
future ambient ozone concentrations is the ratio of non-methane
hydrocarbon (NMHC) emissions to oxides of nitrogen (NOx)
emissions, or the NMHC:NOx ratio. Since much of the inventory
from stationary sources is composed of photochemically reactive
volatile organic compounds (VOC) that are not hydrocarbons,
this ratio is also referred to as the non-methane organic
compound (NMOC)-to-NOx ratio.
The NMOC:NOx ratios used in these analyses are based on
the average of the median ratios observed in 1984 and 1985, for
those cities where data for both years were available. For
cities where data from only one of the years were available,
that value is used. The ratios used in these analyses were
obtained from EPA's Office of Air Quality Planning and
Standards (OAQPS), and are summarized in Table 3-18.[13]
-------
3-39
Table 3-18
NMOC:NOx Ratios Used in Air Quality Modeling [1]
NMOC:NOx
MSA/CMSA Name
Ratio
Akron, OH
12. 7
Atlanta, GA
10 . 4
Baton Rouge, LA
14 . 9
Beaumont-Port Arthur, TX [2]
33. 0
Birmingham, AL
11. 7
Boston Metro Area
7 . 6
Charlotte-Gastonia-Rock Hill, NC-SC [3]
10 . 4
Chattanooga, TN
16.8
Cincinnati Metro Area
9 . 1
Cleveland, OH
7 . 5
Dallas-Fort Worth, TX [4]
12.8
El Paso, TX
13 . 5
Galveston-Texas City, TX [5]
33.2
Houston, TX
12.9
Indianapolis, IN
11.3
Kansas City, MO-KS
9.3
Lake Charles, LA
24 .3
Memphis, TN
13.9
Miami-Hialeah, FL [6]
13 .3
Philadelphia, PA [7]
8.1
Portland, ME
11. 6
Richmond-Petersburg, VA [8]
10.9
Scranton-Wilkes-Barre, PA [9]
14.3
St. Louis, MO-IL
9.6
Washington, DC-MD-VA
9 . 0
[1] NMOC:NOx ratio of 11.6 used for cities not listed.
[2] Beaumont observations.
[3] Charlotte observations.
[4] Average of Dallas and Ft. Worth observations.
[5] Texas City observations.
[6] Miami observations.
[7] Average of observations from 2 sites, both in 1985
[8] Richmond observations.
[9] Wilkes-Barre observations.
-------
3-40
Only 25 of the 61 non-California urban ozone
non-attainment areas are listed in Table 3-18. For those
cities not shown, a default NMOC:NOx ratio of 11.6 is
used.[13] This value is based on the median of both year's
observations for all of the cities listed with the exception of
Beaumont-Port Arthur and Galveston-Texas City. These areas are
not included in determining the median, and thus the default
value, since their NMOC:NOx ratios are extremely high due to
intense concentrations of petroleum processing facilities in
these areas, and thus are not indicative of the rest of the
country.
3. Ozone Design Values
As was noted earlier, the inventory projections and air
quality analyses in this chapter use 1983 as the base year.
The 1983 design values are based on air quality monitoring data
recorded in 1982, 1983, and 1984. These design values, which
are shown in Table 3-19 under the heading "82-84 design
values," are the basis for determining which urban areas are in
non-attainment status for the ozone NAAQS. They are based on
the 1982, 1983, and 1984 monitoring data without modification.
In other words, the values shown in Table 3-19 under the
heading "82-84 design values" represent the fourth highest
one-hour ambient ozone concentrations measured in the three
years of monitoring data examined.
Previous ozone air quality analyses have used design
values determined in this way directly as input for EKMA.
However, the air quality analyses for this chapter include a
refinement which leads to the development of a second set of
"design values" being used to represent the base year ambient
concentrations. These values are also shown in Table 3-19,
under the heading "EKMA design values."
This second set of design values reflect modifications
that take into account the transport of ozone. This is
accomplished by a computer program known as MASH (Multiple
Airshed) . MASH uses wind directions and the times that ozone
exceedances (i.e., one-hour concentrations in excess of the
0.12 ppm NAAQS) occur to determine whether an exceedance is
predominantly the result of emissions originating in the local
area, or of emissions originating m a nearby major
metropolitan area upwind. It then assigns the measured ozone
levels to either the upwind metropolitan area or the local area
as a function of selected criteria.[14] In regions where ozone
transport is significant (e.g., the northeastern part of the
country), this process tends to lower the design values in
smaller cities or MSAs and to raise the design values in the
larger metropolitan areas. These MASH-processed design values
-------
3-41
Table 3-19
61 Non-California Urban Ozone Non-Attainment
Areas and Associated Design Values (ppm)
82-84 Design EKMA Design
Area Values Values
EPA Region 1
Boston Metro Area 0.19 0.18
Greater Connecticut Metro Area 0.23 0.18
New Bedford, MA 0.19 0.19
Portland, ME 0.15 0.10
Portsmouth-Dover-Rochester, NH-ME 0.13 0.09
Providence, RI 0.16 0.15
Springfield, MA 0.19 0.17
Worcester, MA 0.14 0.12
EPA Region 2
Atlantic City, NJ 0.19 0.15
New York Metro Area 0.23 0.24
Vineland-Millville-Bridgeton, NJ 0.14 0.14
EPA Region 3
Allentown-Bethlehem, PA 0.15 0.14
Baltimore, MD 0.17 0.17
Erie, PA 0.13 0.13
Harrisburg-Lebanon-Carlisle, PA 0.13 0.13
Lancaster, PA 0.14 0.09
Philadelphia Metro Area 0.18 0.20
Pittsburgh, PA 0.14 0.14
Reading, PA 0.13 0.13
Richmond-Petersburg, VA 0.14 0.14
Scranton-Wilkes Barre, PA 0.13 0.13
Washington, DC-MD-VA 0.16 0.17
York, PA 0.13 0.13
EPA Region 4
Atlanta, GA 0.17 0.17
Birmingham, AL 0.15 0.15
Charlotte-Gastonia-Rock Hill, NC-SC 0.13 0.13
Chattanooga, TN-GA 0.13 0.13
Huntington-Ashland, WV-KY-OH 0.14 0.14
Louisville, KY-IN 0.15 0.15
Memphis, TN-AR-MS 0.13 0.13
Miami-Hialeah, FL 0.13 0.13
Nashville, TN 0.13 0.13
Tampa-St. Petersburg-Clearwater, FL 0.13 0.13
-------
3-42
Table 3-19 (cont'd)
Area
EPA Region 5
Akron, OH
Canton, OH
Chicago Metro Area
Cincinnati Metro Area
Cleveland, OH
Dayton-Springfield, OH
Detroit, MI
Grand Rapids, MI
Indianapolis, IN
Milwaukee Metro Area
Muskegon, MI
EPA Region 6
Baton Rouge, LA
Beaumont-Port Arthur, TX
Brazoria, TX
Dallas-Fort Worth, TX
El Paso, TX
Galveston-Texas City, TX
Houston, TX
Lake Charles, LA
Longview-Marshall, TX
New Orleans, LA
San Antonio, TX
Tulsa, OK
EPA Region 7
Kansas City, MO-KS
St. Louis, MO-IL
EPA Region 8
Denver-Boulder, CO
Salt Lake City-Ogden, UT
EPA Region 9
Phoenix, AZ
82-84 Design
Values
0 .13
0.13
0 . 20
0. 15
0 . 14
0 . 13
0 . 14
0. 13
0 . 13
0.17
0 . 14
0 . 17
0.21
0 . 14
16
17
0.17
0
0
25
15
0 .15
0 . 15
0
0
14
13
0. 14
0 . 17
0 . 14
0 .15
0 . 15
EKMA Design
Values
0 .13
0 . 13
25
17
0 . 14
0 . 13
0 . 14
0 . 13
0 .13
0.17
0 . 14
0
0
17
21
0 . 14
0.16
0 . 17
0. 17
25
15
15
0 . 15
0 . 14
0 . 13
0. 14
0 . 17
0 . 14
0 . 15
0 . 15
-------
3-43
are intended for use in ozone air quality modeling for
Regulatory Impact Analyses (RIAs), and are not appropriately
used in determining attainment/non-attainment status for
individual areas.
It should also be noted that there is relatively little
difference in the two sets of design values, with the "82-84
design value" and "EKMA design value" being equal for 46 of the
61 areas modeled. Of the 15 areas where the application of the
MASH program led to a change in the design value, the value
used for input to EKMA is lower than the unmodified value in 10
of the areas and higher in five of them. The areas with
differences between the two values are concentrated in the
Northeast and along the central East Coast (EPA Regions 1, 2,
and 3), where many of the MSAs and CMSAs are contiguous. The
greatest increase m design value resulting from the MASH
program occurs for the Chicago Metropolitan Area. The increase
of 0.05 ppm represents a 25 percent increase over the
unmodified value, reflecting the fact that some of the
ozone-precursor emission inventory generated in this area is
transported downwind before it is photochemically converted to
ozone.
11. Nationwide and Non-Attainment Area Inventory Projections
This section presents the results of the nationwide
emissions inventory projections, as well as inventory
projections for the 61 non-California urban non-attainment
areas. Projected emission inventories have two purposes in
this study. First, the inventories are used to generate the
projected emissions reductions associated with each control
strate
-------
3-44
day conditions by annual activity rates (i.e., VMT). Thus,
while the inventories are expressed in units of tons/year, they
are more accurately thought of as design value day inventories
times 365. This is why annual reductions can appropriately be
generated for 5-month m-use RVP control. While the emission
factors for non-fuel-related sources may not be specific to the
design value day, they generally are summer (and thus
comparable) figures.
Thus, though "annual" inventories are used as input for
EKMA, daily inventories are effectively being used. For
control and cost-effectiveness ranking, it does not matter
whether a daily or an annual figure is used provided that the
same period is used for all sources and the same fraction of
each source's emission reduction is in the relevant (ozone
producing) period.
Inventory projections were also made using mobile source
emission factors based on July-average temperatures for the 61
non-attainment areas. The July-average temperatures are more
representative of the entire summer control period than are the
design value day temperatures, and thus are more relevant to
the calculation of economic recovery credits (e.g., recovery
and subsequent use of evaporative emissions), which accrue
throughout the control period. Only the nationwide inventory
projections based on July-average temperatures are presented in
this chapter, as their only use is in the nationwide
cost-effectiveness analysis.
The projection years for both cases (design value day and
July average temperatures) are 1988, 1990, 1992, 1995, 1997,
2000, and 2010. A total of nine regulatory scenarios are
presented, ranging from no control (i.e., certification fuel
RVP at 9.0 psi, with no regulation of in-use RVP) to eight
control scenarios (8.0 to 11.5 psi RVP, in 0.5-psi increments,
for both certification and m-use fuel). Changes to the
volatility of certification test fuel are modeled as beginning
in the 1990 model year. Control of in-use fuel volatility
begins in 1988 (for RVP levels down to 10.5 psi), with
additional in-use control (to RVP levels of 10.0 psi and lower)
beginning in 1992.
The base year (1983) emissions inventories used in the
projections were shown in Tables 3-15 and 3-16 for the nation
and for the 61 urban areas, respectively, and discussed in the
preceding section. All of the necessary inputs describing
growth rates and controls on the various source categories were
also discussed in the preceding section of this chapter.
-------
3-45
The "No Control" scenarios, as discussed under the
development of emission factors, assume that m-use RVP will
rise to the ASTM recommended limits[9] in those areas where
that limit has not yet been reached, and will remain at 1985
levels (see Table 3-5) in those areas where that limit has
already been exceeded. The population-weighted national RVP
level for the "No Control" cases, calculated in the same way
that the nationwide emission factors were calculated (see
section I.A.4. of this chapter), is 11.7 psi for ASTM Class C
areas.
The "No Control" case and all of the control cases (RVP
between 11.5 and 8.0 psi) that are the focus of this analysis
assume that onboard refueling emission controls will be
required beginning in model year 1990. "No Control" and
control projections that do not include onboard refueling
controls have also been generated for both the nationwide and
61-area cases. The nationwide inventory projections without
onboard are also presented in this section.
The control cases are designated by the RVP level that
would apply to certification fuel and to m-use fuel in ASTM
Class C areas, which represent much of the country in terms of
land area, population, and ozone non-attainment. RVP control
levels of 11.5, 11.0, 10.5, 10.0, 9.5, 9.0, 8.5, and 8.0 psi
are evaluated. Proportional RVP reductions are assumed for
ASTM Class A, B, D, and E areas. The proportional control
levels for these areas, corresponding to the listed values for
Class C areas, are shown in Table 3-20.
As noted above, the modeling of the urban ozone non-
attainment areas is limited to the 61 such areas outside
California. The MOBILE3 emission factor model used by EPA is
not applicable for California vehicles, which are subject to
different emission standards and regulations than are Federal
(49-state) vehicles. While the use of nationwide emission
factors from the MOBILE3 model only introduces a relatively
small error in the nationwide inventories, it is a significant
problem when modeling of individual urban areas within
California is considered.
A. Nationwide Inventory Projections
The emission inventories presented below, and the ozone
air quality projections that follow, are based on phased-in
control of in-use fuel volatility, as was stated in the
introduction to this section. Due to feasibility
considerations, control of in-use volatility can be to levels
-------
3-46
Table 3-20
Proportional Reduction Control RVP Levels
ASTM Class
BCD
Current ASTM
RVP Limit (psi): 9.0 10.0 11.5 13.5 15.0
Controlled RVP
Levels (psi):
6.3
7 . 0
8.0
9.4
10 . 4
6 . 6
7.4
8.5
10 . 0
11 . 1
7.0
7.8
9.0
10. 6
11.7
7 . 4
8.3
9.5
11.2
12 .4
7.8
8.7
10.0
11 . 7
13 . 0
8.2
9 . 1
10 . 5
12.3
13.7
8.6
9.6
11.0
12.9
14.3
9.0
10.0
11.5
13.5
15.0
-------
3-47
no lower than 10.5 psi {in ASTM Class C areas) between 1988 and
1991 inclusive. In-use volatility control down to 8.0 psi (in
ASTM Class C areas) begins in 1992. Reasons for this two-
tiered implementation of in-use control are discussed m
Chapter 5. Changes in the volatility of certification test
fuel, which will result m vehicle design modifications, are
modeled as beginning in the 1990 model year.
Tables 3-21 through 3-24 present the results of the
nationwide inventory projections for each of the control
scenarios evaluated. Tables 3-21 and 3-22 are based on design
value day temperatures, while Tables 3-23 and 3-24 are based on
July-average temperatures. Tables 3-21 and 3-23 assume that
onboard vehicle refueling emission control systems are already
in piece, beginning with model year 1990, while Tables 3-22 and
3-24 present the same scenarios without assuming prior
implementation of onboard refueling controls. The discussion
that follows focuses on the results presented in Table 3-21 and
the assumption of onboard control requirements, however, the
points made generally are also applicable if Table 3-22 (no
onboard requirement) or Tables 3-23 or 3-24
-------
3-48
Table 3-21
Nationwide NMOC Emissions Inventories
Assuming Prior Implementation of Onboard
Refueling Control in MY 1990
based on design-value temperatures
Control NMOC Emissions (x 1000 T/yr)
Scenario
1990
1995
2000
2010
No RVP Control
20319
19504
19947
22538
RVP =11.5
19925
18732
18884
21156
(1.9)*
(4.0)
(5.3)
(6.1)
11.0
19481
18481
18739
21060
(4.1)
(5.2)
(6.1)
(6.6)
in
o
19108
18265
18609
20967
(6.0)
(6.4)
(6.7)
(7.0)
O
O
19108**
18089
18499
20885
(6.0)
(7.3)
(7.3)
(7.3)
9 . 5
19108
17949
18409
20815
(6.0)
(8.0)
(7.7)
(7.6)
9.0
19108
17825
18324
20744
(6.0)
(8.6)
(8.1)
(8.0)
8.5
19108
17744
18258
20681
(6.0)
(9.0)
(8.5)
(8.2)
8.0
19108
17672
18198
20620
(6.0)
(9.4)
(8.8)
(8.5)
* Figures in parentheses represent percent reduction in
inventory relative to "No RVP Control."
** Certification fuel changes in 1990. In-use RVP control to
less than 10.5 psi is delayed until 1992 (see Chapter 5).
-------
3-49
Table 3-22
Nationwide NMOC Emissions Inventories
Assuming No Prior Onboard Refueling Control Requirement
based on design-value temperatures
Control NMOC Emissions (x 1000 T/yr)
Scenario
1990
1995
2000
2010
No RVP Control
20334
19766
20325
22999
RVP =11.5
19940
18994
19263
21616
(1.9)*
(3.9)
(5.2)
(6.0)
11.0
19496
18731
19100
21499
(4.1)
(5.2)
(6.0)
(6.5)
m
o
19121
18501
18952
21384
(6.0)
(6.4)
(6.8)
(7.0)
O
O
19121**
18313
18824
21280
(6.0)
(7.4)
(7.4)
(7.5)
9.5
19121
18160
18715
21190
(6.0)
(8.1)
(7.9)
(7.9)
9 . 0
19121
18024
18612
21097
(6.0)
(8.8)
(8.4)
(8.3)
m
CO
19121
17932
18530
21012
(6.0)
(9.3)
(8.8)
(8.6)
Co
o
19121
17848
18452
20929
(6.0)
(9.7)
(9.2)
(9.0)
Figures in parentheses represent percent reduction in
inventory relative to "No RVP Control."
Certification fuel changes in 1990. In-use RVP control to
less than 10.5 psi is delayed until 1992 (see Chapter 5).
-------
3-50
Table 3-23
Nationwide NMOC Emissions Inventories
Assuming Prior Implementation of Onboard
Refueling Control in MY 1990
based on July
-average
temperatures
Control
NMOC
Emissions
(x 100 0 T/yr)
Scenario
1990
1995
2000
2010
No RVP Control
19687
18975
19443
21974
RVP =11.5
19425
18444
18707
21018
(1.3)*
(2.8)
(3 8)
(4.4)
11. 0
19091
18256
18596
20939
(3.0)
(3.8)
(4.4)
(4.7)
10 . 5
18796
18083
18490
20860
(4.5)
(4.7)
(4.9)
(5.1)
10 . 0
18796**
17940
18396
20788
(4.5)
(5.5)
(5.4)
(5.4)
9.5
18796
17824
18319
20725
(4.5)
(6.1)
(5.8)
(5.7)
9.0
18796
17719
18244
20661
(4.5)
(6.6)
(6.2)
(6.0)
8.5
18796
17649
18186
20601
(4.5)
(7.0)
(6.5)
(6.2)
8 . 0
18796
17581
18129
20545
(4.5)
(7.3)
(6.8)
(6.5)
* Figures in parentheses represent percent reduction
inventory relative to "No RVP Control."
** Certification fuel changes in 1990. In-use RVP control
less than 10.5 psi is delayed until 1992 (see Chapter 5).
-------
3-51
Table 3-24
Nationwide NMOC Emissions Inventories
Assuming No Prior Onboard Refueling Control Requirement
based on July
-average
temperatures
Control
NMOC
Emissions
(x 10 00 T/yr)
Scenario
1990
1995
2000
2010
No RVP Control
19701
19237
19820
22434
RVP = 11.5
19440
(1.3)*
18706
(2.8)
19086
(3.7)
21480
(4.3)
11 . 0
19105
(3.0)
18505
(3.8)
18956
(4.4)
21378
(4.7)
10 . 5
18810
(4.5)
18320
(4.8)
18832
(5.0)
21277
(5.2)
10 . 0
18810**
(4.5)
18166
(5.6)
18721
(5.5)
21183
(5.6)
9 . 5
18810
(4.5)
18037
(6.2)
18627
(6.0)
21100
(5.9)
9 . 0
18810
(4.5)
17920
(6.8)
18533
(6.5)
21013
(6.3)
8.5
18810
(4.5)
17836
(7.3)
18457
(6.9)
20932
(6.7)
8 . 0
18810
(4.5)
17757
(7.7)
18382
(7.3)
20854
(7.0)
* Figures in parentheses represent percent reduction in
inventory relative to "No RVP Control."
** Certification fuel changes in 1990. In-use RVP control to
less than 10.5 psi is delayed until 1992 (see Chapter 5).
-------
3-52
are larger m the ll.5-to-9.0 psi range than in the 9.0-to-8.0
psi range. As discussed in section III of Chapter 2, this is
primarily because EPA has estimated that no further benefit in
terms of exhaust hydrocarbon emissions will result from the
control of in-use RVP to levels lower than 9.0 psi in ASTM
Class C areas.
Finally, it can also be noted from Table 3-21 that the
reductions, in terms of percent relative to no control,
increase through the projection period for control scenarios of
11.5, 11.0, and 10.5 psi. These percent reductions are
approximately equal m all 3 projection years at the 10.0 psi
control level. At 9.5 psi or lower, the percent reductions are
greatest for 1995, and are slightly lower in 2000 and again in
2010. Similar trends were reflected in the original volatility
study analyses. 112 ] If long-term RVP control is in the range
of 10.5-11.5 psi, vehicle-based control (i.e., increased
canister volume) is the dominant effect. There is an inherent
delay in the effectiveness of vehicle-based controls, which
depend primarily on fleet turnover to achieve the desired
reductions in emissions. Thus, this effect increases over
time. Fuel control, which is effective immediately, is the
dominant effect when long-term RVP control in the 9.5 psi or
lower range is examined. The downward trend reflects the
continuing conversion of the fleet from carbureted to
fuel-injected engines, and the somewhat lower emissions of the
latter.
B. Non-Attainment Area Inventory Projections
Much of the discussion of the preceding section, on the
nationwide inventory projections, is also applicable to the
inventory projections for the 61 non-California urban ozone
non-attainment areas. These projections are summarized in
Table 3-25.
The most notable difference between the impact of
volatility control on the nationwide inventories and the impact
on the urban non-attainment area inventories can be seen in the
percent reduction estimates associated with each control
scenario. For example, control of in-use fuel volatility to
9.0 psi is projected to reduce the nationwide inventory by 8.1
percent in 2000, while the combined inventories for the 61 non-
attainment areas would drop by 9.0 percent. Similar
differences in the percent reduction obtained from a given
control scenario in a specific projection year are apparent
when Tables 3-21 and 3-25 are compared cell-by-cell.
The reason that volatility control provides greater
proportional reductions in the non-attainment areas than on a
-------
3-53
Table 3-25
Total NMOC Emissions for 61 Non-California
Urban Ozone Non-Attainment Areas
(Assuming Prior Implementation of
Onboard Refueling Control in MY 1990)
Control NMOC Emissions (x 1000 T)
Scenario
1990
1995
2000
2010
No RVP Control
6784
6524
6717
7699
RVP =11.5
6640
6238
6321
7182
-
(2.1)*
(4.4)
(5.9)
(6.7)
11 . 0
6476
6145
6266
7147
(4.5)
(5.8)
(6.7)
(7.2)
10 . 5
6339
6065
6218
7112
(6.6)
(7.0)
(7.4)
(7.6)
10 . 0
6339**
6000
6177
7081
(6.6)
(8.0)
(8.0)
(8.0)
9 . 5
6339
5948
6144
7055
(6.6)
(8.8)
(8.5)
(8.4)
9 . 0
6339
5903
6113
7028
(6.6)
(9.5)
(9.0)
(8.7)
8.5
6339
5873
6088
7004
(6.6)
(10.0)
(9.4)
(9.0)
8 . 0
6339
5846
6065
6983
(6.6)
(10.4)
(9.7)
(9.3)
Figures in parentheses represent percent reduction in
emissions, relative to "No RVP Control."
Certification fuel changes in 1990. In-use RVP control to
less than 10.5 psi is delayed until 1992 (see Chapter 5).
-------
3-54
nationwide basis can be seen through comparison of the base-
year inventories shown in Tables 3-15 and 3-16. Table 3-15
shows that, of the nationwide inventory, 43.5 percent of the
total NMOC emissions are from gasoline-related sources. Of the
NMOC inventory for the 61 urban areas, 46.5 percent of
emissions are from gasoline-related sources. Thus, applying
the same proportional reduction to those components of the
inventories that would be affected by volatility control (i.e.,
gasoline-related sources) yields greater reductions, as a
fraction of the total inventory, in the 61 urban areas than in
the nation taken as a whole.
Figure 3-1 presents the reductions in emissions for the 61
non-California urban ozone non-attainment areas by source. The
reductions shown are for RVP control to 9.0 psi, relative to
the "No Control" case. As has been noted, only gasoline-
related sources will be affected by RVP control; thus, there
are no entries on Figure 3-1 for the "area," "point," or "off-
highway" source categories. The most important point made
clear m Figure 3-1 is that reductions m motor vehicle
evaporative emissions account for the bulk of the total
reductions m both years. In 1995, the reduction in
evaporative emissions represents about 77 percent of the total
reduction, and this proportion increases to almost 87 percent
of the total in 2010.
Ill. Ozone Air Quality Analysis
Because of the complex relationship between hydrocarbon
emissions and ambient ozone concentrations, the rollback
approach used by EPA to model other pollutants (i.e., NOx and
CO) is inappropriate for ozone. Instead, EPA uses EKMA
(Empirical Kinetic Modeling Approach) to predict future ambient
ozone concentrations in urban areas. EKMA utilizes a series of
ozone isopleths which depict downwind ozone concentrations as a
function of initial NMOC and NOx concentrations, subsequent
NMOC emissions, meteorological conditions, reactivity of the
precursor mix, and concentrations of ozone and precursors
transported from upwind areas.
It should be noted that EKMA as used by EPA is primarily a
nationwide-average model. In other words, city-specific
information is not generally used as input for the model, with
the exception of the base-year ozone concentrations (design
values) and NMOC:NOx ratios from which future concentrations
are projected. Meteorological conditions (e.g., mixing height,
angle of incidence of sunlight) are based on data from one of
three cities: Los Angeles, CA, Denver, CO, and St. Louis, MO.
The Los Angeles data are used for modeling California coastal
cities, and thus are not used in this analysis; Denver data are
-------
700 n
600 H
£ 500
o
o
o
o
tn
c
o
o
ZJ
~u
400-
O 300-
200-
100-
624.7
622.0
74.3
477.5
83.1
341.0
0.5-
20.7
25.1-
a
Ul
I
Ln
Ul
Legend
m EXHAUST
CD EVAP
£3 STAGE II
EJ STAGE I
ZZ1 REMAINDER
1995
2010
-------
3-56
used for cities located in Arizona, Colorado, Nevada, New
Mexico, and Utah; and St. Louis data are used for all other
areas.[15] (The design values for the 61 urban areas in this
analysis are shown in Table 3-19, and the NMOC:NOx ratios are
shown in Table 3-18. For more details on EKMA, see References
[16] and [17] . )
Using EKMA and the NMOC emissions inventory presented in
Table 3-16, projections of future ozone conditions in the 61
non-California urban ozone non-attainment areas were made. The
results of this modeling are discussed below.
Tables 3-26 through 3-28 present summaries of the EKMA-
based predictions of future ambient ozone conditions in the 61
non-California urban ozone non-attainment areas under various
levels of RVP control, assuming prior implementation of onboard
refueling control in the 1990 model year. (Tables 3-29 through
3-31 present the same information, except that no control of
refueling emissions is assumed. The following discussion is
based on Tables 3-26 through 3-28 and the assumption of onboard
control of refueling emissions.)
Table 3-26 shows the average percent change in ambient
ozone concentration with respect to the base level in 1983.
The reductions expected to occur under the "No Control"
scenario are, of course, in response to programs other than
gasoline volatility control. Programs already in place that
are projected to result in reductions in ozone precursor
emissions, and thus improvements in ozone air quality, include
the current Federal Motor Vehicle Control Program (FMVCP) and
some controls on stationary sources (see section I.B.4 of this
chapter). The current FMVCP accounts for most of the
reductions that are projected in the "No RVP Control" case.
However, the additional ozone reductions shown under the
eight RVP control scenarios in Table 3-26 are due solely to
NMOC reductions attained through the control of in-use and
certification fuel RVP. For example, if in-use RVP was
controlled to 10.0 psi in 1992 (10.5 psi in 1988), and
certification RVP was raised to 10.0 psi in 1990, ambient ozone
concentrations by the year 1995 would be expected to decrease
an additional 3 percent beyond the baseline RVP scenario (i.e.,
11 percent instead ot 8 percent lower than 1983 levels).
Estimates of the total annual exceedances of the ozone
NAAQS are presented in Table 3-27 for each of the RVP control
scenarios. The NAAQS for ozone calls for an effective limit of
0.125 ppm on the fourth highest daily maximum one-hour ozone
concentration in any three-year period; the numbers of
exceedances listed m the table represent the total number of
days this maximum hourly ozone concentration is expected to
-------
3-57
Table 3-26
Average Percent Change* in Ambient Ozone Concentrations
in 61 Non-California Urban Non-Attainment Areas
{Assuming Prior Implementation of
Onboard Refueling Control in MY 1990)
Control Projection Year
Scenario
1990
1995
2000
2010
No RVP Control
-7
-8
-8
-4
RVP =11.5
-7
-10
-9
-6
11. 0
-8
-10
-10
-6
10.5
-9
-11
-10
-7
o
o
_g * *
-11
-10
-7
9 . 5
-9
-11
-10
-7
9 . 0
-9
-11
-11
-7
8.5
-9
-12
-11
-7
CO
o
-9
-12
-11
-7
Relative to the base-year (1983) levels.
Certification fuel changes in 1990. In-use RVP control
less than 10.5 psi is delayed until 1992 (see Chapter 5).
-------
3-58
Table 3-27
Total Number of Exceedances of the Ozone NAAQS in
61 Non-California Urban Non-Attainment Areas
(Assuming Prior Implementation of
Onboard Refueling Control in MY 1990)
Control Projection Year
Scenario
1990
1995
2000
2010
No RVP Control
246
237
241
283
RVP =11.5
244
228
225
261
11.0
236
219
220
261
10 . 5
232
212
216
258
O
o
r-4
232*
206
214
256
9 . 5
232
203
208
256
9 . 0
232
200
208
256
8 . 5
232
198
206
253
o
00
232
198
205
252
Certification fuel changes in 1990. In-use RVP control
less than 10.5 psi is delayed until 1992 (see Chapter 5).
-------
3-59
Table 3-28
Projected Number of Non-California
Urban Ozone Non-Attainment Areas
(Assuming Prior Implementation of
Onboard Refueling Control in MY 1990)
Control Proiection Year
Scenario
1990
1995
2000
2010
No RVP Control
40
37
40
49
RVP = 11.5
38
35.
34
41
11 0
37
33
33
41
10 . 5
36
32
33
41
10 . 0
36*
31
32
41
9 . 5
36
30
30
41
9.0
36
29
30
39
in
00
36
29
30
39
8.0
36
29
30
39
Certification fuel changes in 1990. In-use RVP control
less than 10.5 psi is delayed until 1992 (see Chapter 5).
-------
3-60
Table 3-29
Average Percent Change* in Ambient Ozone Concentrations
in 61 Non-California Urban Non-Attainment Areas
(Assuming No Control of Refueling Emissions)
Control Projection Year
Scenario
1990
1995
2000
2010
No RVP Control
-7
-8
-7
-3
RVP =11.5
-7
-9
-9
-6
11.0
-8
-10
-9
-6
10 . 5
-9
-10
-9
-6
10.0
_9 * *
-11
-10
-6
9.5
-9
-11
-10
-6
9.0
-9
-11
-10
-6
8.5
-9
-11
-10
-6
8.0
-9
-11
-10
-6
Relative to the base-year (1983) levels.
Certification fuel changes in 1990. In-use RVP control
less than 10.5 psi is delayed until 1992 (see Chapter 5).
-------
3-61
Table 3-30
Total Number of Exceedances of the Ozone NAAQS in
61 Non-California Urban Non-Attainment Areas
(Assuming No Control of Refueling Emissions)
Control Projection Year
Scenario
1990
1995
2000
2010
No RVP Control
246
239
250
292
RVP =11.5
244
232
233
265
11. 0
236
226
230
264
10. 5
232
220
225
262
10 . 0
232 *
213
222
262
9.5
232
208
219
261
9 . 0
232
203
218
261
in
00
232
203
213
260
o
00
232
201
210
258
Certification fuel changes in 1990. In-use RVP control to
less than 10.5 psi is delayed until 1992 (see Chapter 5).
-------
3-62
Table 3-31
Projected Number of Non-California
Urban Ozone Non-Attainment Areas
(Assuming No Control of Refueling Emissions)
Control Projection Year
Scenario
1990
1995
2000
2010
No RVP Control
40
38
41
51
RVP = H.5
38
35
35
43
11 . 0
37
35
34
43
10 . 5
36
33
34
41
o
o
36*
33
34
41
9 . 5
36
31
33
41
9.0
36
30
33
41
8.5
36
30
32
41
o
00
36
30
31
41
Certification fuel changes in 1990. In-use RVP control
less than 10.5 psi is delayed until 1992 (see Chapter 5).
-------
3-63
exceed 0.125 ppm m the 61 areas modeled. Only the peak
monitoring site in each of the 61 areas is considered. Thus,
365 is the maximum possible number of annual exceedances per
area. Table 3-27 shows, for example, that control of both
in-use and certification fuel RVP to 9.0 psi is estimated to
reduce the combined total number of ozone exceedances in the 61
urban areas by approximately 16 percent relative to the
baseline RVP scenario in 1995 (i.e., 200 rather than 237 total
exceedances).
Finally, Table 3-28 estimates the total number of
non-California urban areas expected to be in non-attainment of
the ozone NAAQS under the various RVP control options. As
shown, RVP control to 9.0 psi is projected to enable about 8
more areas to come into attainment in 1995. One significant
limitation associated with evaluating control options on the
basis of number of non-attainment areas must be kept in mind:
only those areas that fall below the NAAQS as a result of the
action are distinguishable. In other words, the value of
bringing an area closer to attainment is not recognized unless
attainment is actually projected to occur. Therefore,
estimated overall emissions reductions or changes in average
ambient concentrations as a result of a particular action are
more indicative of the environmental impact of that action than
is the projected number of non-attainment areas.
IV. Benzene Emissions and Incidence Analysis
The control of in-use fuel volatility is likely to result
in changes to the composition of gasoline, which in turn will
result in changes in the composition of hydrocarbon emissions
from motor vehicles. This section uses the benzene mass
emission fractions (mass of benzene as a fraction of the total
mass of hydrocarbons) developed in section III of Chapter 2 to
estimate the impact of fuel volatility control on mobile source
benzene emissions. This is followed by a discussion of the
impact of these changes in benzene emissions on the cancer
incidences resulting from mobile source benzene exposure.
A. Benzene Emissions
In section III of Chapter 2, benzene mass fractions
associated with each level of RVP control were developed.
Separate fractions were derived for exhaust, evaporative, and
refueling hydrocarbon emissions. At each RVP level, distinct
exhaust benzene emission fractions are presented for three
vehicle categories based on catalyst technology: vehicles
equipped with 3-way catalysts; vehicles equipped with 3-way-
plus-oxidation catalysts; and pre-1980 vehicles, and those
post-1980 vehicles equipped only with oxidation catalysts.
Similarly, distinct evaporative benzene emission fractions were
provided for carbureted and fuel-injected vehicles. A single
refueling benzene emission fraction was given for each RVP
level. These fractions were summarized in Table 2-63.
-------
3-64
Relevant data from M0BILE3 {e.g., registration and
catalyst technology distributions) were used to weight the
exhaust and evaporative benzene emission fractions by the
fraction of vehicle miles travelled (VMT) for each technology
category. The resulting fleet VMT-weighted benzene emission
fractions were then multiplied by the emission factors
generated in section I.A. of this chapter to produce exhaust
benzene and evaporative benzene emission factors in milligrams
per mile (mg/mi). These are presented in Table 3-32, with
evaporative benzene emissions shown as a range because of
uncertainties in the relationship between fuel benzene content
and evaporative benzene emissions. Only the emission factors
corresponding to 11.5, 10.5, 9.5, and 9.0 psi RVP are included
in the table.*
The procedure for estimating the refueling benzene
emission factors shown in Table 3-32 begins with the refueling
benzene fractions shown in Table 2-63 (g Bz/gal fuel
dispensed). These were divided by the corresponding total
refueling emission factors (g HC/gal fuel dispensed),
interpolated from those presented in Table 2-45, to yield
benzene mass fractions (g Bz/g HC) for refueling emissions.
Multiplying these mass fractions (which are the same for all
vehicles) by the appropriate refueling emission factors (g/mi),
determined as part of the analysis in section II of this
chapter, yields the refueling benzene emission factors shown m
Table 3-32.
Finally, the exhaust, evaporative, and refueling emission
factors for each vehicle category are summed to give the total
benzene emission factors presented in the bottom section of
Table 3-32. The evaporative benzene ranges are incorporated in
these totals. The relatively large ranges in these factors,
relative to the incremental change as a function of RVP level,
are reflective of some of the uncertainties inherent in these
estimates.[18]
These benzene emission factors, when multiplied by the
nationwide VMT for each category, yield the estimated mobile
source benzene emissions shown in Table 3-33.[10] The ranges
in these estimates again reflect the ranges in the evaporative
RVPs of 8.0 and 8.5 psi were not evaluated since the
available Bonner and Moore refinery modeling, from which
the fuel benzene content was estimated, did not evaluate
RVPs below 9.0 psi. Simple extrapolation was not deemed
appropriate due to the complex refinery operations
involved, some of which increase fuel aromatic content and
some of which do not. Also, the effect of RVP control was
so small that presenting results for 10.0 and 11.0 psi
serves no useful purpose.
-------
3-65
Table 3-32
Benzene Emission Factors* (mq/mi)
(Assuming Prior Implementation of
Onboard Refueling Controls in MY 1990)
Light-Duty Vehicles
RVP 1990**
(psi) Exht Evap Refu
11.5
44
12-15
2
10.5
43
13-14
2
9.5
43
11-15
2
9.0
42
11-16
2
1995
Exht
Evap
Refu
33
7-8
1
30
7-8
1
30
6-8
1
30
6-9
1
2000
Exht
Evap
Re f u
31
5-o
11
27
5-b
< 1
27
5-6
c 1
28
5-6
< 1
Light-Duty Trucks
RVP 1990 1995 2000
(psi) Exht Evap Re f u Exht Evap Refu Exht Evap Refu
11.5 135
10.5 133
9.5 133
9.0 132
23-23 3
26-27 3
22-31 3
23-31 3
98 14-17
93 15-16
92 14-18
92 14-19
2 74
2 b6
2 66
2 66
10-11 <1
10-11 <1
10-12 <1
10-12 <1
Heavy-Duty Vehicles
RVP
(psi)
1990
1995
2000
Exht
Evap
Refu
Exht
Evap
Refu
Exht
Evap
Reft
11.5
130
53-68
4
101
40-50
3
91
36-45
1
10.5
130
63-67
4
99
41-44
3
88
35-37
1
9.5
132
55-77
4
101
38-52
3
90
33-45
1
9.0
133
58-79
4
101
41-55
3
90
36-48
1
Total Benzene Emission Factors
RVP 1990 1995 2000
(psi) LDV LPT HuV LDV LPT HDV LDV LPT HDV
11.5 58-61 161-166 187-202 41-42 114-117 144-154 36-37 84-85 128-137
10.5 58-59 162-163 197-201 38-39 110-111 143-146 32-33 76-77 124-126
9.5 56-60 158-167 191-213 37-39 108-112 142-156 32-33 76-78 124-136
9.0 55-60 158-166 195-216 37-40 108-113 145-159 33-34 76-78 127-139
Exhaust and evaporative emission factors from ln-house MOBILE3 (Run 028,
March 3, 1986 ) .
Calendar year
-------
3-66
Table 3-33
Projected Mobile Source Benzene Emissions
(Assuming Prior
Refuel mg
Implementation of Onboard
Controls in MY 1990)
RVP
Benzene Emissions* (x 1000T)
(psi)
1990
1995
2000
11. 5
149-155
108-111
91-93
10 . 5
150-152
102-104
82-84
9.5
150-152**
100-105
82-85
9.0
150-152**
100-107
83-86
From gasoline-fueled mobile sources only.
In-use RVP control to less than 10.5 psi is delayed until
1992 {see Chapter 5).
-------
3-67
benzene emission fractions discussed above. These estimated
emissions do not include benzene emissions from diesel
vehicles, which are very low relative to those from gasoline-
fueled vehicles and will not be affected by RVP control.[19]
Although the projected emissions in Table 3-33 show a
slight tendency to be lower as RVP is reduced, the reduction is
never greater than 10 percent, and is on the order of the
ranges in the estimates. Although the benzene emission
fractions increase slightly with decreasing RVP, the decrease
in the total hydrocarbon emissions slightly overcompensates for
this effect. This effect is also small in comparison to the
decrease in benzene emissions over time at all RVP levels.
These estimates illustrate that mobile source benzene emissions
will drop by more than 50 percent from 1982 levels of about
250,000 Mg/yr,[18] regardless of the RVP level assumed.
While exhaust benzene emissions dominate both the
nationwide and mobile source inventories, it is important to
note that as fleet turnover occurs and exhaust hydrocarbon
emissions decrease, the size of tae exhaust benzene emission
inventory is expected to drop by about 50 percent between 1985
and 2000 even without further vehicle or RVP controls. The
same is true for evaporative benzene emissions. Assuming the
prior implementation of onboard refueling emission controls,
benzene emissions from refueling would be expected to decrease
by about 12 percent in 1990, and by about 80 percent by 2000.
B. Incidence Analysis
Benzene is a known human carcinogen at low exposure
levels, and has been categorized as a hazardous air pollutant
by EPA under Section 112 of the Clean Air Act. This analysis
estimates the number of cancer incidences which may occur due
to exposure to mobile source benzene emissions and the effect
of RVP control on the number of such incidences.
In general, there are two types of exposures involved with
mobile source incidences. The first is exposure of the general
public to ambient benzene concentrations (ambient exposures),
and the second is exposure of individuals to more highly
localized concentrations of benzene, such as during refueling
operations (individual exposures). Exhaust, evaporative and
refueling emissions from mobile sources all contribute to
ambient concentrations of benzene and thus to ambient
exposures. However, in addition to their contribution to
ambient exposures, refueling emissions also pose an additional
individual exposure risk. The Gasoline Marketing Study
presented an estimate of incidences due to both individual and
ambient exposures to refueling benzene emissions, so refueling-
-------
3-68
related incidences were not recalculated for this
analysis.[19] These refueling incidences are presented along
with the incidences due to exposures to ambient benzene
concentrations arising from exhaust and evaporative emissions.
This analysis was conducted for a base case and four
likely RVP control cases, for exhaust and evaporative emissions
only.* In-use RVP control was assumed effective in 1988, and
vehicle controls were assumed effective for the 1990 model
year. Base case incidences from community exposure to benzene
refueling emissions were taken from the analysis supporting the
gasoline marketing analysis, as noted above, while
refueling-related incidences for the RVP control cases were
estimated using the benzene fractions of Table 2-63 and the
changes in total refueling hydrocarbon emissions from section I
of this chapter.
Several steps are necessary to estimate the number of
incidences which may result from exposure to exhaust and
evaporative benzene emissions. First, the exhaust and
evaporative mass fractions were fleet-weighted for each model
year, and the results were used with the MOBILE3 HC input data
to calculate benzene composite emission factors for calendar
years 1985, 1988, 1990, 1995 and 2000. These factors were
shown in Table 3-32. The resulting MOBILE3 composite emission
factors were then used as inputs to the NAAQS Exposure Model
(NEM) to estimate total annual person hours of exposure at
different ambient benzene concentrations.[20] The exposure
estimates from NEM were then multiplied by the benzene unit
risk factor to calculate incidences.
Using the methodology described above, the expected number
of cancer incidences was calculated for the base case and four
RVP control cases. The results are shown in Tables 3-34 and
3-35 and are discussed below.
First, it is apparent from the base-case results that
exhaust and evaporative-related benzene incidences decrease by
roughly 55 percent between 1985 and 2000 with or without
vehicle or RVP control. This decrease is largely due to the
overall reduction in total hydrocarbon emissions as fleet
turnover occurs, as illustrated by Tables 3-21 through 3-25.
Also for the base case, it is worth noting that exhaust-related
incidences outnumber evaporative-related incidences by about
4:1, and that refueling-related incidences are considerably
less than those related to evaporative emissions.
Base case: in-use RVP = 11.5 psi, certification fuel
RVP = 9.0 psi; control cases: in-use RVP = certification
fuel RVP = 11.5, 10.5, 9.5, and 9.0 psi RVP.
-------
3-69
Table 3-34
Incidences Due to Mobile Source Benzene Exposure
Sub
Grand
Year
Evaporative*
Exhaust
Total
Refuelinq** Total
Base
Case***
1985
16-20
62
78-82
7 85-89
1988
11-14
49
60-63
7 67-70
1990
10-12
42
52-54
6 58-60
1995
6-8
32
38-40
3 41-43
2000
5-6
30
35-36
1 36-37
RVP =
11.5 psi
1988
11-14
49
60-63
1990
10-12
42
52-54
1995
7-8
30
37-38
2000
4-5
29
33-34
RVP =
10.5 psi
1988
13-15
47
60-62
1990
10-11
41
51-52
1995
6-7
30
36-37
2000
5-6
28
33-34
RVP =
9.5 psi
1988
11-16
48
59-64
1990
9-12
41
50-53
1995
6-7
31
37-38
2000
5-6
28
33-34
RVP =
9.0 psi
1988
12-16
48
60-64
1990
9-12
41
50-53
1995
6-8
31
37-39
2000
5-7
28
33-35
* Low and high range estimates.
** Refueling incidences were taken directly from analysis
supporting the Gasoline Marketing Study, with the 1990 and
later values adjusted for the impact of onboard controls.
*** In-use RVP = 11.5 psi, certification RVP =9.0 psi.
-------
3-70
Table 3-35
Differences in Number of Incidences: Control Cases vs Base Case
Year/
RVP Control
Level**
Category-
11
. 5 psi
10.5 psi
9.5 ps i
9.0 ;
Low Evaporative Emission
Range
* *
1988:
Evaporative
0
+2
0
+ 1
Exhaust
0
-2
-1
-1
Total
0
-1
-2
-2
1990:
Evaporative
0
0
-1
-1
Exhaust
0
-1
-1
-1
Total
0
-1
-3
-3
1995:
Evaporative
+1
0
0
0
Exhaust
-2
-2
-1
-1
Total
-1
-2
-1
-2
2000 :
Evaporative
-1
0
0
0
Exhaust
-1
-2
-2
-2
Total
-2
-2
-2
-2
High Evaporative Emission Range**
1988:
Evaporative
0
+ 1
+2
+2
Exhaust
0
-2
-1
-1
Total
0
-2
0
-1
1990 :
Evaporative
0
-1
0
0
Exhaust
0
-1
-1
-1
Total
0
-2
-2
-2
1995:
Evaporative
0
-1
-1
0
Exhaust
-2
-2
-1
-1
Total
-2
-3
-2
-2
2000 :
Evaporative
-1
0
0
+ 1
Exhaust
-1
-2
-2
-2
Total
-2
-2
-2
-1
Certification fuel RVP = in-use fuel RVP.
Evaporative emission range depends on benzene emissions
model used (see section III of Chapter 2).
-------
3-71
Second, the vehicle-control approach to reducing total
m-use hydrocarbon emissions (in-use fuel RVP = certification
fuel RVP = 11.5 psi) shows only a slight reduction in the
number of incidences (one or two) in 1995 and 2000. Vehicle
control is dependent on fleet turnover to reduce in-use
emissions, so no differences are noted in 1990, the year such
controls are assumed to begin. Since the benzene mass
fractions (exhaust and evaporative) are the same for the base
case and the vehicle-control case, any reduction of in-use
benzene emissions and related incidences arises as a result of
the reduction in total hydrocarbon emissions which would
accompany vehicle controls. As the results in Table 3-35
indicate, the overall impact of vehicle controls would be small.
The third area for discussion is the effect of in-use RVP
control on the estimated number of incidences. As can be seen
in Tables 3-34 and 3-35, in-use RVP control has little or no
impact on the number of incidences. However, when there is a
small effect, it is nearly always a reduction in incidences.
The explanation is that decreases in benzene emissions due to
decreases in hydrocarbon emissions are largely offset by the
increases in the benzene mass fraction (shown in Table 2-63).
As explained earlier, RVP control is likely to lead to
increases in the benzene and total aromatic content of
gasoline, which in turn increases the benzene mass fraction in
exhaust and evaporative emissions.
If gasoline composition with respect to benzene and total
aromatic content remained at current levels under RVP control,
a greater decrease in the number of incidences would likely be
noted. Given the accuracy of the modeling techniques used
here, however, it is most reasonable to say that RVP control
will have no little or impact on exhaust- or evaporative-
related incidences.
In summary, the analysis shows that approximately 80
incidences occur currently due to exhaust and evaporative
benzene emissions and that this number will drop by more than
50 percent to about 35 in the year 2000 as fleet turnover
occurs and total hydrocarbon emissions decrease. Vehicle
control or RVP control would have essentially no effect on the
number of incidences or the relative contribution of the
emissions (exhaust vs. evaporative) leading to the incidences.
More information on how this analysis was performed and some of
the areas of uncertainty may be found in an EPA technical
memorandum.[18]
-------
3-72
References for Chapter 3
1. "Compilation of Air Pollutant Emission Factors,
Volume II: Mobile Sources" (AP-42), U.S. EPA, OAR, OMS, Fourth
Edition - September 1985.
2. Federal Register, 42 FR 32954, June 28, 1977.
(Codified in the Code of Federal Regulations, Title 40, Part
86, Subpart B, July 1, 1986.)
3. "1983 NEDS Inventory Modifications and Inventory
Projection Methodology m Support of Volatility NPRM," EPA
Memorandum from Mark Wolcott, TEB, to Terry Newell, SDSB,
July 2, 1987.
4. "Data Request from Office of Mobile Sources," EPA
Memorandum from Keith Baugues, AMTB, to Ray Vogel, CPOB, August
5, 1986.
5. "Statistical Abstract of the United States,
1982-83," Bureau of the Census, U.S. Dept. of Commerce, 103d
edition.
6. "Alcohol Week," March 5, 1984 and March 3, 1986.
7. "Alcohol Outlook," January 1986.
8. "MVMA National Gasoline Survey, Summer Season
(Sampling Date: July 15, 1983)," October 15, 1983, and "MVMA
National Gasoline Survey, Summer Season (Sampling Date: July
15, 1985)," October 15, 1985, Motor Vehicle Manufacturer's
Association, Inc.
9. "Standards Specification for Automotive Gasoline:
D-439-83," American Society for Testing and Materials.
10. "MOBILE3 Fuel Consumption Model," Wolcott, Mark A.,
U.S. EPA, OAR, OMS, and Dennis F. Kahlbaum, Computer Sciences
Corporation, February 1985, EPA-AA-TEB-EF-85-2.
11. "Evaluation of Air Pollution Regulatory Strategies
for the Gasoline Marketing Industry - Response to Public
Comments," U.S. EPA, OAR, OAQPS and OMS, July 1987. (Available
in Public Docket No. A-87-11.)
12. "Study of Gasoline Volatility and Hydrocarbon
Emissions from Motor Vehicles," U.S. EPA, OAR, OMS, ECTD,
November 1985, EPA-AA-SDSB-85-5.
-------
3-73
References for Chapter 3 (cont'd)
13. "Updated NMOC:NOx Ratios for Regulatory Impact
Analyses (RIAs)," EPA Memorandum from Keith Baugues, AMTB, to
Tom McCurdy, SASD, February 21, 1986.
14. "1982-84 Ozone Design Values for Regulatory Impact
Analyses," EPA Memorandum from Richard G. Rhoads, MDAD, to John
R. O'Connor, SASD and Charles Gray, ECTD, June 16, 1986.
15. "Use of City-Specific EKMA in the Ozone RIA,"
Gipson, Gerald L. and Warren P. Freas, U.S. EPA, OAR, OAQPS,
MDAD, July 1983.
16 "Uses, Limitations, and Technical Basis of
Procedures for Quantifying Relationships Between Photochemical
Oxidants and Precursors," U.S. EPA, OAR, OAQPS,
EPA-450/2-77-02la, November 1977.
17. "Guidelines for Use of City-Specific EKMA in
Preparing Ozone SIPs," U.S. EPA, OAR, OAQPS, EPA-450/4-80-027,
March 1981.
18. "Mobile Sources Benzene Emissions and a Preliminary
Estimate of Their Health Impacts," EPA Memorandum from Charles
L. Gray to Richard D. Wilson, May 15, 1986.
19. "Evaluation of Air Pollution Regulatory Strategies
for the Gasoline Marketing Industry, U.S. EPA. OAR, OAQPS and
OMS, EPA-450/3-84-012a, July 1984.
20. "Improved Mobile Sources Exposure Estimation," M.N.
Ingalls, EPA-460/3-85-002, March 1985.
-------
Appendix 3-A
Butane Reactivity
Ozone formation in the troposphere is a complex process
involving many reactions and compounds. Hydrocarbons are one
group of these compounds which lead to the formation of ozone.
However, each hydrocarbon species (i.e., methane, butane,
benzene, etc.) reacts at varying rates defined here as the
reactivity, resulting in different amounts of ozone formation
under the same atmospheric conditions. Reactivity has also
been categorized by the rate of hydrocarbon depletion and the
rate of nitrogen oxide oxidation, as well as by other criteria,
but these are not the focus here. The purpose of this appendix
is to present the ozone-related reactivity of butane in
relation to other hydrocarbons. Butane is the focus of the
appendix since volatility control results in emission
reductions which consist primarily of butane.
The majority of evidence in existence supports a high
reactivity of butane with respect to ozone formation.
Historically, the first evidence was obtained in indoor smog
chamber testing.[1] Butane was mixed with NOx at
ppm-concentrations and irradiated with artificial sunlight.
Ozone was produced at concentrations exceeding the air quality
standards of 0.12 ppm. Isolated studies have given conflicting
results (i.e., ozone yields less than 0.12 ppm) but such
conflicts can be explained by differences in butane-to-NOx
ratio and other conditions used in the studies. [2,3] The early
indoor smog chamber results have been discredited because real
atmosphere conditions were not simulated in the indoor smog
chambers and because in the real atmosphere, butane reacts in
the presence of other hydrocarbons, also. However, butane was
found again to be a significant producer of ozone in more
realistic outdoor smog chamber studies in which butane was
tested mixed with other organics.[4]
In an attempt to rate the reactivity of different
chemicals, researchers have developed several reactivity
classifications based on different properties as previously
mentioned. The classifications used by EPA are based on both
reactivity (rated on smog chamber data) and volatility
properties and were developed in 1984. The overall reactivity
of the chemicals is designated by one of three classes
categorized I, II or III corresponding to "unreactive",
"borderline", or "reactive", respectively.[5] It is believed
within EPA that the "borderline" classification occurs at a
reactivity level roughly equal to that of ethane.[6] Table
-------
3-A-2
3-A-l presents the classification of several organic
compounds. Butane is definitely in the "reactive" category
based on this scale.
A second reactivity scale developed in the mid-1960s,
called the GM scale, is based on the rate of formation of NOz
(equal in magnitude but opposite in direction of the rate of NO
disappearance) under fixed test conditions.[7] (The
alkyl-peroxy radical, R02, formed from the reaction of
hydrocarbons in the atmosphere, competes with ozone to react
with NO to form NO?. The alkyl-peroxy radical reacts faster
than ozone, thus resulting in an accumulation of ozone.) The
GM scale is a relative scale with the specific reactivity of
2,3-dimethyl-2-butane, a highly reactive compound, equal to
100. The reactivities of several hydrocarbons are shown is
Table 3-A-2. Butane is also classified as a reactive compound
under the GM scale, though in the lowest category of reactive
compounds.
A third reactivity scale, developed by the U.S. Department
of Health, Education, and Welfare (HEW) in the mid-1960s, is
based on six criteria: equilibrium oxidant, peroxyacetyl
nitrate (PAN), formaldehyde, and aerosol formation, eye
irritation and plant damage.[8] The hydrocarbons were rated
for each of the six categories (0 = no effect and 10 = maximum
effect). The average of the six ratings is the relative
reactivity. The averaged response for various hydrocarbons are
presented in Table 3-A-3. On the HEW scale butane is rated as
non-reactive along with other paraffins up through the pentanes.
Besides the reactivity scales, other experimental and
modeling work has been done to show the effects of butane on
ozone. Modeling has shown butane to be from 14 percent to 34
percent as reactive as propene, a highly reactive
chemical.[9,10] Based on existing smog chamber and fuel data,
one report estimated that nine percent of the ozone generated
results from reactions involving butane.[10]
In summary, it appears butane is not one of the most
reactive compounds. However, the issue is whether or not
butane reacts fast enough to affect ozone in the breathing
zone. Both the EPA and GM reactivity scales rate butane as a
reactive compound. (The older HEW scale is based on a number
of criteria, not all of which pertain to ozone.) Modelling and
smog chamber data also verify the contribution of butane to
ozone formation in the troposphere. Based on available
information, it is EPA's position that butane is a
photochemically reactive compound which contributes to ground
level ozone formation. Therefore, reductions in butane
emissions through in-use fuel volatility control are expected
to lead to subsequent reductions in ambient ozone levels.
-------
3-A-3
Table 3-A-l
EPA Classification of Hydrocarbon Reactivity
II III
Ethane Benzene Propane
Acetylene i-Butane
n-Butane
Xylenes
Formaldehyde
Methanol
Cyclohexane
Ethylene
Propene
-------
3-A-4
Table 3-A-2
GM Scale of Hydrocarbon. Reactivity Classes
and Class Specific Reactivity
Specific
Class reactivity Hydrocarbons
I 0 methane
ethane
propane
acetylenes
benzene
II 2 mono alkyl benzenes
C4 and higher molecular
weight paraffins
ortho and para dialkyl
benzenes
cyclic paraffins
III 5 ethylene
meta dialkyl benzenes
formaldehyde and higher
aldehydes
IV 10 l-olefins (other than
ethylene)
diolefins
tri and tetra alkyl
benzenes
internally bonded
olefins
V 30 internally bonded
olefins
VI 100 internally bonded olefins
with substitution at the
double bond
cyclo olefins
-------
Table 3-A-3
HEW Scale of Hydrocarbon Reactivity
Substance or Averaged
Sub-Class Response
Ci-Cs paraffins
0
acetylene
0
benzene
0
C6 paraffins
1
toluene (and other
monoalkylbenzenes)
3
ethylene
4
diolef ins
6
dialkyl and tnalkyl-
benzenes
6
internally double-
1-alkenes
7
bonded olefins
8
-------
3-A-6
Appendix 3-A References
1. Heuss, J.M,, and W.A. Glasson, "Hydrocarbon
Reactivity and Eye Irritation," Environmental Science &
Technology, 2, p. 1109, 1968.
2. Yanagihara, S.I., et al. , "Photochemical
Reactivities of Hydrocarbons," Proceedings of the Fourth
International Clean Air Congress, Tokyo, Japan, 1977.
3. Dimitnades, et al., "Development and Utility of
Reactivity Scales From Smog Chamber Data," Bureau of Mines,
Report of Investigations RI 8023, Department of Interior, 1975.
4. Kamens, R.M. , et al., "Smog Chamber Experiments to
Test Oxidant-Related Control Strategy Issues,"
EPA-600/3-82-014, 1982.
5. Singh, H.B., et al., "Reactivity/Volatility
Classification of Selected Organic Chemicals: Existing Data,"
EPA-600/3-84-082, 1984.
6. "Reactivity of Butane," EPA memo form Dr. Basil
Dimitnades, ACPD, to Charles L. Gray, Jr., ECTD, April 22,
1986.
7. "Smog Chemistry Points the Way to Rational Vehicle
Emission Control," J.D. Caplan, SAE Transactions, Vol. 74
(1966).
8. "Emissions from Combustion Engines and Their
Control," D.J. Patterson, and N.A. Henein, Ann Arbor Science
Publishers, 1979.
9. Bufalini, J.J. and M.C. Dodge, "Ozone-Forming
Potential of Light Saturated Hydrocarbons," Environmental
Science & Technology, 12, page 308, 1983.
10. Singh, H.B., et al., "Assessment of the
Oxidant-Forming Potential of Light Saturated Hydrocarbons in
the Atmosphere," Environmental Science & Technology, 15, page
113, 1981.
-------
CHAPTER 4
Vehicle Controls
This chapter will focus upon the modifications to current
evaporative emissions control systems (ECSs) needed in order to
improve their ability to control evaporative emissions. The
need for such improvement stems from the proposed changes in
the certification test procedure to underscore EPA's intent to
have manufacturers design the evaporative control systems to
operate effectively m-use. These changes are: 1) eliminate
the discrepancy between certification and m-use fuel
volatility, and 2) begin the test with a canister loaded to
breakthrough. The discussion is broken up into three
sections- the technological feasibility of the potential
modifications - Section I; the costs associated with such
modifications - Section II; and the overall conclusions
Section III.
I. Technology
As has been indicated in Chapter 2, vehicular evaporative
control systems may require modification in order to meet the
requirements imposed by the potential changes in the
certification test procedure. As will be shown, there is a
strong interrelation between the canister size and design, and
purge. Depending on the actual situation, it is no longer
considered necessary to fully increase both canister size and
purge rate. Although some of these potential design options
will be mentioned, this section will take a worse case approach
in the canister/purge discussion. This worse case approach
will assume a minimal research and development (R&D) effort by
the automobile industry. Therefore, the change in capacity of
the canister required to adsorb and desorb hydrocarbon vapors
will be discussed separately from the changes required for the
control system to purge the evaporative canister.
A. Canister Modifications
The following section describes the adsorption and
desorption characteristics of an activated carbon canister.
The succeeding section describes the necessary working capacity
required to capture the increased evaporative emissions from a
higher RVP fuel.
1. Canister Characteristics
The carbon canister adsorbs the gasoline vapors that would
otherwise escape into the atmosphere during hot soaks and
diurnals. Under predetermined operating conditions, the engine
burns these vapors by pulling air through the canister and into
the engine. The amount of gasoline vapors that the canister
can cyclically adsorb (adsorb/desorb) is called the working
capacity. This capacity is dependent upon internal and
-------
4-2
external factors to the evaporative control system. The
internal factors are: 1) charcoal type and particle size*, 2)
the volume of charcoal in the canister, and 3) the
configuration of the canister. The external factors are: 1)
the ambient temperature and humidity, 2) the volume of purge
air drawn through by the control system, 3) the purge
temperature, and 4) the composition of fuel vapors being
adsorbed. These factors will now be discussed further.
Vapor adsorption and desorption is a purely physical
process, with Van der Wall's bonding acting as the force
holding the vapor to the charcoal.[1] The effectiveness of
this bonding will depend upon the average particle size, the
ratio of surface area to volume of the individual particles,
and the size of the pores in which the vapor molecules are
adsorbed. These various elements determine what is defined as
the charcoal "working capacity" (measured as grams HC per 100
cubic centimeters (cc) of charcoal).
In a given volume of charcoa^, smaller particles can be
compacted more tightly, thus increasing the likelihood that the
vapors will come in contact with the surface of the particles.
The tighter compaction of particles also allows for a greater
mass of charcoal in a given volume, resulting in increased
volumetric working capacity. This increase in charcoal working
capacity has been seen in a designed exper lment. [ 2 ] This same
compaction, however, decreases permeability and reduces the
flow rate through the canister at a specific pressure drop. It
is for this reason that, in current practice, the maximum
pressure drop across the canister at a specified flow rate is
usually the determining factor in the selection of particle
size.[3]
A greater surface area for a given particle implies more
sites to which vapor molecules can be bonded. Therefore a
particle with a highly convoluted surface having a large
surface area to volume ratio will necessarily be a better
adsorber than a particle of similar size with a smaller surface
area to volume ratio. Increasing this ratio by disturbing the
surface of a particle is the process that is referred to when
carbon is defined as being activated.[4]
Particle size of a given charcoal type is usually
expressed in terms of mesh size and is expressed in the
form A x B. A and B are divisions per inch and A x B size
particles are able to pass through (1/A) inch by (1/B)
inch openings. Thus, larger values of A and B represent a
smaller mesh size and therefore smaller particles.
-------
4-3
Finally, pore size can affect the charcoal working
capacity, although the effect is not as clear as with particle
size or the surface area-to-volume ratio. In general, vapor
molecules will adsorb into pores of a size comparable to their
own. It is possible, however, that larger particles can become
lodged in pores in such a manner as to block access to more
pore space that smaller particles could occupy. Since these
larger particles are heavier and the bonding forces are
stronger, they may not be removed during the purge cycle, and a
portion of the carbon's adsorptivity will be lost.[1 ]
A charcoal's working capacity is measured as its ability
to adsorb and desorb HC vapor, on a volumetric basis, under a
relatively standardized test procedure. Working capacity is
usually measured as grams of butane per 100 cc of charcoal, as
butane is easy to work with, yields consistent results and is
the dominant compound in vehicular evaporative emissions.
Typical levels of "butane working capacity" for various
charcoals are presented in Table 4-1. These values will not
necessarily translate directly into charcoal working capacities
for vehicular evaporative emissions, as these HC vapor mixtures
are more complex than simple butane. Since HC species differ
in their physical characteristics, they will be adsorbed
preferentially and the degree of preference may depend upon
carbon type. Thus, equal butane working capacities will not
always result in equal gasoline working capacities.[2,4]
Assuming that the characteristics of the charcoal remain
the same, there should be a direct relationship between the
volume of charcoal used and the system (total) working
capacity. This should be clear from the definition of the
charcoal working capacity.
Canister configuration may play a role in determining the
system working capacity, although this theory is disputed. In
a study by Scott Environmental Technology, the specific working
capacity of a canister was increased by 12.3 percent by
doubling the length of a cylindrical canister while holding the
diameter, and therefore the cross sectional area, constant.[2]
The theoretical explanation for this increase in working
capacity centers on the concept of a "mass transfer zone"
(MTZ). This theory states that when the carbon bed is charged,
there will exist a zone which is fully saturated with
hydrocarbons and a zone where the saturation decreases from
fully saturated to completely devoid of hydrocarbons. This MTZ
will only operate at half of its theoretical capacity when
breakthrough occurs, and its length will not vary with the
-------
4-4
Table 4-1
Charcoal Properties*
Apparent
Surface
Dens i ty
Capaci
ty
(q/100 cc)
Charcoal
Manufacturer
Base
Mesh
Size Area (m /q)
(lb/ft )
Test
I*
Test II**
BPL-3
Calgon
Coal
6 X
14 800-1000
23-24
5 . 8
8.08
WV-A
Westvaco
Wood
10 X
25 1500-1700
15-18
8 . 5
8.31
WV-A
Westvaco
Wood
14 X
35 1600-1800
16-19
9.0
8.89
Extruded***
Westvaco
Wood
****
20-21
10. 5
--
* Specified by charcoal manufacturer in "Mestvaco's Wood-Base Carbons Improve
Evaporative Emission Control," Billy Kornegay, Ph.D., P.E., September 1980.
** From a test performed by Westvaco.
*** Data from phone conversation with Bill Kornegay of Westvaco Corporation on November
5, 1984.
**** Not specified, but equivalent pressure drop of mesh size b x 14.
-------
4-5
length of the carbon bed. A carbon bed whose length is twice
that of the MTZ will have a capacity at breakthrough of 75
percent of maximum.* By doubling the length of the carbon bed,
with the length of the MTZ remaining constant, there will be
less under-utilization of capacity, and the capacity at
breakthrough will be 87.5 percent of maximum.** This
theoretical increase of 12.5 percent is quite close to that
observed m the testing.[2]
Using a different approach, a 1969 SAE paper by R.S. Joyce
et al, indicated that system working capacity is not
significantly related to canister configuration.[3] While both
approaches acquired data from adsorption/desorption cycles,
Joyce defined working capacity as the amount of butane
(hydrocarbons) desorbed during a 20 minute of time, rather than
the amount adsorbed to the point of breakthrough. It is not
known if the same conclusion would have been reached using a
different period of time. Also, it is not known how different
charcoals affect the capacity/configuration results.
Assuming that canister configuration generally does affect
working capacity, future designs could optimize the canister
shape. To increase the working capacity and still remain
within overall size limitations found in some engine
compartments, future designs potentially could incorporate
baffles to increase the effective length to diameter ratio.
External factors also affect the system working capacity.
For a given canister, the total amount of hydrocarbons desorbed
is dependent on the total volume of purge air. Working
capacity is also dependent on the desorption and adsorption
temperatures. For a common Calgon charcoal, increasing the
desorption/adsorption temperature difference from 0 to 75°F
increases the carbon working capacity by 48 percent.[5]
Therefore, the working capacity can be increased with higher
temperature purge air.[2,4] This too could potentially
minimize canister size and purge rate. Due to the non-linear
nature of HC desorption, the increase in hydrocarbons desorbed
by a heated purge possibly could offset the usual decrease with
time, creating a more constant purge density and therefore a
more manageable purge.
Increased humidity will decrease the working capacity of a
charcoal canister. Fortunately this is not a lasting effect as
charcoal has a higher affinity for hydrocarbons than for water
vapor.[2]
Figured as: [0.5 (length) x 1.0 (capacity utilized)] +
[0.5 (length) x 0.5 (capacity utilized)].
Figured as: [0.75 (length) x 1.0 (capacity utilized)] +
[0.25 (length) x 0.5 (capacity utilized)].
-------
4-6
The adsorptivity of charcoal has been found to be
dependent upon the vapor composition of the evaporative
emissions.[2] This is illustrated in Figure 4-1, showing the
average charcoal working capacities for two coal-based
charcoals tested on 8.7 and 13.8 RVP (psi) fuels. According to
the Scott study, hydrocarbon molecules larger than C6 are
more easily retained in the carbon than the "lighter"
hydrocarbons.[3] They felt the larger molecules were firmly
lodged in the carbon structure and therefore required a higher
energy to release. Since high RVP fuel produces more
lighter-end hydrocarbons than low RVP fuel, the working
capacity will be higher. The increase in charcoal working
capacity with increasing RVP will translate into a higher
system working capacity.
2. Required Working Capacity
Changes in fuel volatility may require changes in system
working capacity. To ensure that the changes in canister
capacity are not underestimated, e'.isting canisters are assumed
to have barely sufficient capacity for the emissions from 9.0
RVP fuel. The change in system working capacity necessary to
meet the requirements of the potential changes in fuel
volatility can therefore be estimated by considering the
resulting changes in uncontrolled emissions. These must be
adjusted, however, to reflect the changes in working capacity
that will occur automatically due to the relation to fuel
volatility. The difference will represent the amount of extra
control that the Emission Control System (ECS) designer will
have to develop.
Figure 4-2 shows how the required amount of extra system
working capacity is determined. Curve 1 shows the relationship
between uncontrolled evaporative emissions and fuel
volatility. The curve has been normalized so that emissions
from a 9.0 RVP fuel, typical of current certification fuel,
will equal 1.0. Curve 2 is a reproduction of Figure 4-1,
showing the relationship between charcoal working capacity and
fuel volatility. This curve has also been normalized such that
the value for 9.0 RVP fuel equals 1.0. Curve 3 is the ratio of
Curves 1 to 2, and indicates the additional system working
capacity required for a given increase in RVP. As an example,
a change in the certification fuel to 11.5-psi RVP, while
generating 81 percent more emissions, will only require the
development of a 60 percent increase in system working
capacity. The remaining 13 percent required increase will be
obtained from the increase in specific charcoal working
capacity.
-------
7.0
6.9
6.8
6.7
6.6
6 5
6.4
6 3
6 2
6 1
6 0
5.9
5 8
5.7
5.6
5.5
5.4
5.3
5.2
5.1
5.0
i. O.VJLiIT0 -» -»-
CHARCOAL WORKING CAPACITY vs. RVP
i 1 1 1 r
.0 10.0 12.0 14
RVP (psi)
~ CHAR WORKING CAP.
-------
.90
80
.70
60
50
40
30
20
10
00
Figure 4-2
NOMINAL CANISTER EFFECTS vs. RVP
>IONS
RVP (psl)
CHAR. WC
DESIGN CAPACITY
-------
4-9
Increasing the canister volume is the easiest way to
increase system capacity. Changing the canister configuration,
increasing the desorption-adsorption temperature differential
and changing the specific charcoal working capacity also affect
the system capacity.
In 1984, a new type of charcoal was introduced to the
market. The properties of this "extruded" charcoal are shown
in Table 4-1, along with the properties of other charcoals.
Because of the greater working capacity of the extruded
charcoal (10.5 vs. 6.8-9.0 g/100 ml), canisters with an equal
volume of the extruded charcoal would have a higher working
capacity and, thus, be able to support emissions from a higher
RVP fuel. This is shown in Table 4-2, taking into account the
current carbon type of various manufacturers and curves 1 and 2
of Figure 4-2. By switching to this type of charcoal, then,
present systems could support 9.7 - 11.7 RVP fuels without
requiring an increase in canister size. An increase m
pressure drop with the extruded charcoal may occur However,
the charcoal manufacturer expects to be able to accommodate
these concerns by manufacturing various particle sizes.[6]
The option of switching to this extruded charcoal to meet
an increase in the RVP certification fuel standard is feasible
for some vehicle manufacturers (i.e., those using large
particle, coal-based charcoal). However, the remainder of this
report will only consider the alternative option of changing
canister capacity by changing canister size. This is not meant
to imply that charcoals with specific working capacities
greater than those traditionally used could not be developed.
It is more an indication of a decision not to make any
assumptions about their development at this time.
Thus, the likely maximum change in designed canister
capacity that may be needed to certify with a new RVP fuel is
shown in Table 4-3. For an 11.5 RVP certification fuel, this
would be a 60 percent increase in carbon corresponding to: 775
ml for LDVs, 1010 ml for LDTs and 2400 ml for HDVs. To
estimate this for LDVs and LDTs, an average industry-wide
canister size was determined from 1985 certification records by
assuming an equal sales distribution between evaporative
families for a given manufacturer and averaging
corporate-average canister sizes by projected 1990 sales for
each corporation. Only dominant manufacturers were included:
General Motors, Ford, Chrysler, Toyota, and Nissan. For HDVs,
an average canister size of 4000 ml is used, as it is expected
that General Motors will use two canisters totaling 4000 ml of
charcoal on all cf its HDVs.[7] Details of these calculations
are shown in Appendix 4-A.
-------
4-10
Table 4-2
Canister
Canister Equivalents With Extruded Charcoal
Ratio of
Increased
Working
Capacity
Present to Present
Carbon Mesh Working
Type Size Capacity
Equivalent
RVP
Control*
Ford
Chrysler
GM
Nissan* *
Calgon BPL-3
Westvaco WV-A
Westvaco WV-A
Calgon BPL-3
6 x 14
14 x 35
10 x 25
6 x 14
1 . 54
1 . 17
1 . 24
1 . 54
11.65
9.85
10.20
11 . 65
From Curve 3 of Figure 4-2.
Estimated on the basis of charcoal type.
-------
4-1 1
Table 4-j
Average Canister Volume Changes"
Cert1firat inn Fuel RVP fosi)
8.0
8 5
9.5
10 0
10 5
1 0
11.5
Vehicle Class
ml
°L.
ml
%
ml
%
ml
%
ml
%
Oil
%
ml
LDV
-284
-22
-129
-10
129
10
284
22
439
34
594
46
775
60
LDT
-371
-22
-169
-10
169
10
371
22
574
34
77b
46
1013
60
HOV
-880
-22
-400
-10
400
10
880
22
1360
34
1840
46
2400
60
Assuming no change in type of charcoal or canister configuration
Based on curves ) and 3 of Figure 4-2. No change is needed at 9 0
RVP
-------
4-12
It was assumed that all canisters would have to be
proportionally increased in size to accommodate an increase in
emissions. However, due to the non-linear desorption
characteristics of canisters, some manufacturers may decide to
oversize their canisters to allow for lower purge rates. At
present, many manufacturers use identical canisters on vehicles
with differing uncontrolled emissions to reduce manufacturing
and inventory costs. Therefore, oversized canisters may have a
dual benefit. The relation between canister size and purging
will be developed further in the following section.
B. Purge Recruirements
Along with a change m vapor storage capacity, a similar
change may be required in the ability to desorb these
hydrocarbons from the canister. For 11.5 psi RVP fuel, the
amount of hydrocarbons purged will have to increase by 81
percent. In addition, a further increase in purging may be
required for the proposed change in certification test
procedure. This resulting increase in purged HCs must be
accomplished without an increase in tailpipe exhaust emissions.
Designing a purge system to handle this increase in
canister loading involves the consideration of canister
desorption characteristics, along with a number of engine
parameters. Due to the non-linearity of canister desorption,
the engine may receive a purge stream that varies from an
extremely rich mixture to an extremely lean (almost pure air)
mixture. If not properly controlled, the initial "slug" of
purge may cause rich misfire, or some other driveability
problem. However, such improper fuel/air mixtures may also
result in exhaust emission "spikes".
Overall, before a vehicle can be designed to handle an
increase in purged hydrocarbons, several technical questions
must be answered: 1) what is the range of control of existing
fuel metering/exhaust treatment (catalyst) systems; 2) what
vacuum sources are available; 3) under what conditions can
purge occur; and 4) what methods are available to increase the
HCs purged? An expansion of each of these questions will now
be made.
1. Fuel Metering/Exhaust Treatment Systems
As shown in Chapter 2, an increase m unmetered fuel can
increase exhaust emissions. Therefore, the effects of
increased purge must be within the range of control of fuel
metering systems. An exhaust oxygen sensor can be used to send
a rich/lean feedback signal to an onboard computer, which then
uses this input to calculate a minor correction for the
-------
4-13
air/fuel ratio (fuel delivery). Essentially all light-duty
gasoline vehicles presently being manufactured use feedback
control. With increased concern for fuel economy, and with
tighter exhaust standards, feedback systems are expected to
become common on light-duty trucks and light heavy-duty
vehicles. Heavier heavy-duty vehicles have less stringent
exhaust emission standards and therefore will probably not be
using feedback systems in the near future, nor will they need
to be as concerned with small increases in exhaust emissions.
Since they are without catalytic converters, oxygen
deficiencies can not cause an increase in tailpipe HC levels,
as found in LDVs and LDTs. Also, HDVs have very low fuel
economy ratings where unmetered fuel will be only a small
fraction of their total fuel consumed. Therefore, the effects
of purge will not be as critical with them.
For vehicles with an oxygen sensor, the engine never
operates right at stoichiometry, but cycles around it, usually
within ±0.5 air/fuel ratios. The range of authority of the
oxygen loop is also around ±0.5-1.0, which smooths
transitions between operating points, and also minimizes any
problems that may occur with sensor degradation or failure.
Therefore, either the perturbations in the air/fuel ratio due
to the introduction/increase of purge should be well within
this range of authority, or a method of satisfactorily
expanding this range must be developed.
Due to the transient behavior of typical driving patterns,
exhaust emissions and therefore exhaust emission controls have
a very transient behavior. The delay time between a signal to
change the air/fuel ratio and detection of the resulting change
by the oxygen sensor is on the order of 0.1 seconds. In
comparison, a single cycle in a 4-stroke engine takes only
0.060 seconds to complete at 2000 rpm (a typical FTP speed).
Therefore, an engine may go through several combustion cycles
before receiving a correction signal. Due to this finite time
for feedback control, a "slow" (1-2 second) introduction/
increase of the purge becomes ideal. Presently, a few vehicle
models use dampening canisters (small canisters connected
between the evap canister and the purge vacuum source) to
slowly initiate the purge. Other methods will be suggested
later.
2. Available Vacuum Sources
The purge vacuum source must be adequate to purge the
canister over an FTP while considering the pressure drop in the
evaporative control system. Also, a constant purge to intake
air ratio allows increased ingestion of unmetered hydrocarbons
during times of increased metered fuel consumption. This
proportional purge design maximizes the total purge volume
attainable over an LA-4 while staying within the air/fuel
perturbation limits.
-------
4-14
Existing systems obtain their vacuum from the air cleaner
housing, throttle-housing, or intake manifold. The largest
vacuum source is located at the intake manifold and is largely
dependent on throttle position. Unfortunately, the maximum
vacuum occurs during closed throttle position (idle) which is
also the period of minimum fuel consumption. Therefore, a
method to limit purge (with the associated unmetered fuel) is
required. Presently, the common restraining methods use a
two-stage purge valve or purge through the PCV valve. The air
cleaner and throttle-housing provide a vacuum source
proportional to air flow, but the maximum vacuum is much
smaller than that available at the intake manifold.
Consequently, the intake manifold is the most common vacuum
source used for purging. To purge the increased canister
loadings associated with higher RVP fuel, the intake manifold
will probably continue to be the most common vacuum source.
To optimize the purge flow rate through this inversely desired
vacuum source, a variable flow purge restraining device would
be required.
3. Purge Timing
By starting the certification test procedure with a
canister loaded to breakthrough, two conditions must be
considered in the purge/canister design. These conditions are
as follows.
First, the loaded canister must be adequately purged
during the LA-4 prep cycle to allow the vapors from the
resulting hot soak and the following diurnal to be fully
captured. This condition requires the greatest amount of
purge. However, since emissions are not measured during the
prep cycle, some emissions effect could occur without penalty.
Second, the canister at a lighter loading must be
adequately purged during the ensuing FTP to allow vapors from
10 minutes worth of hot soak between bags 2 and 3, and a full
hot soak at the end of the FTP to be captured. In this case,
increases in exhaust emissions will have to be generally
avoided, due to the stringency of LDV, LDT and lighter HDV HC
emission standards.
The LA-4 prep cycle is the most difficult purge condition
to meet. The entire cycle lasts only 1370 seconds. Since LDVs
and LDTs use feedback control to keep the engine around
stoichiometry (the point of highest catalyst efficiency),
purging usually occurs only during these closed-loop periods.
Therefore, the first 1-3 minutes of the LA-4 typically preclude
purging due to the required warm-up time for the catalytic
converter and oxygen sensor. In addition, most systems do not
-------
4-15
allow purging during periods of minimal fuel consumption, i.e.
idles and decelerations. This results approximately in 760
seconds (12.7 minutes) available for purging during the LA-4.
The prep cycle condition can be met using one of two
strategies. The first strategy is to purge much more than one
hot-soak and diurnal loading during the LA-4 prep cycle (where
emissions are not measured) by greatly increasing the purge
rate and/or canister size. In this way, a lightly loaded
canister will be purged during the FTP. The light loading will
result in a high air/fuel ratio in the purge line, which will
cause minimal perturbation to the engine's scheduled air/fuel
ratio.
The second strategy is to increase the purge rate only
enough to purge a diurnal and hot soak loading during the LA-4
prep cycle. In this case, the canister will be fully loaded at
the beginning of the FTP, and an improved method of purge
control will be required to manage the higher density of
unmetered hydrocarbons without causing an exhaust effect.
Since this strategy is the more difficult of the two, it will
be discussed in this section.
4. Methods to Increase Purge
Increasing the amount of hydrocarbons purged can be
accomplished by increasing the duration of the purge and/or by
increasing the rate of purge. However, due to the limitations
previously mentioned, significantly increasing the purge time
during an FTP is not viable. The rate of purge, on the other
hand, can be increased by increasing the concentration in the
purge stream or by increasing the volume of purge air through
the canister.
Increasing the purge concentration versus increasing the
purge flow rate requires separate design strategies. The
purged hydrocarbon concentration is partly dependent on the
canister size, load level, and type of charcoal. The canister
desorption curves (HC mass purged per total volume of air
flow), are always non-linear, initially showing a high rate of
desorption and then quickly changing to a slower rate. By
increasing the canister size, the canister/purge system can be
designed to desorb a hot soak and diurnal using the relatively
constant, high desorption rate of the initial portion of the
canister curve, along with a low flow rate.
Figure 4-3 graphically shows the effect of increasing the
canister size to decrease the total purge air volume required.
The calculated hot soak and diurnal loading which must be
purged is 32 grams (11.5 RVP fuel). All three canisters are
assumed to be loaded to breakthrough at the start of purge,
-------
Figure 4-3
CANISTER PURGE CURVES FOR 3 CANISTERS
"5? grams
pu cged
0
2 2L
1 .31
1 5L
1
A
8
Purge Air Volume (Cuhir Feel)
-------
4-17
which may be close to the "equilibrium point" for the given
purge rate and diurnal/hot soak loads. (Actual systems will
probably have extra capacity for a factor of safety). As seen
by Figure 4-3, by increasing the canister size from 1.5Q. to
2.22., the 32 grams can be desorbed with a relatively constant
HC purge rate. The resulting "constant" purge concentration
eliminates large swings, and therefore eliminates the problem
of large "HC slugs" reaching the engine.
Due to the continued high HC concentration at the end of
the desired purge, a smaller total volume of purge air can also
be used to desorb the 32 grams. As shown in Figure 4-3, the
total purge volume decreased from 6.2 ftJ/LA-4 to 1.6
ft3/LA-4. This smaller volume of purge air also reduces the
"vacuum leak" that the primary fuel metering system must be
designed to accommodate. Therefore, an oversized canister can
help minimize HC density swings in the purge stream, along with
reducing the required volume of purge air. Due to low flow
rates, deviation in the scheduled air/fuel ratio should be
small, even when the canister is empty.
The other method to increase the total amount of
hydrocarbons purged requires increasing the total volume of air
flow through a "normally sized" canister. This can partly be
accomplished by designing a proportional purge system using a
variable flow purge valve. In addition, the flow rate schedule
may also have to be increased to obtain the total volume of
required purge air. Unfortunately, this increase in purge air
may complicate the control strategy of the air/fuel ratio. For
these situations, two possible solutions are discussed below.
One possible solution to an air/fuel ratio complication is
to increase the total range of authority of the oxygen loop.
In this way, the purge can be introduced within the normal
air/fuel ratio tolerance, and slowly increased as the feedback
system adjusts the metered fuel to obtain stoichiometry. To
accomplish this type of control, a slow-opening, variable flow
purge valve could be used, along with the corresponding
modifications in the fuel scheduling algorithm. The valve
could be a diaphragm operated control valve in conjunction with
a vacuum delay valve, or an electric solenoid valve with
frequency modulation.
The diaphragm valve could receive a damped vacuum signal
proportional to the air flow (e.g., from the air cleaner), and
then limit the purge accordingly. However, with the improved
control from electronics, pneumatic systems are being phased
out. Some manufacturers are already starting to upgrade their
purge control by installing electrical solenoids with frequency
modulation. The electric solenoid can be operated by the
onboard computer, which allows more flexibility and precision
-------
4-18
in system control. Since the purge HC concentration decreases
with increased purge time, future onboard computer systems
could be designed to monitor the purge time from engine start
up and then slowly increase the scheduled flow rate as time
progressed. According to an SAE seminar on sensors and
actuators, plenty of program/memory space is available m
onboard computers, so there should be no need to increase the
amount of computer hardware for software requirements.[8]
Another possible solution to the air/fuel ratio
complication involves a change in fuel control strategy, along
with the modulating purge solenoid valve. Since the purge
solenoid valve is controlled by the onboard computer, the
scheduled air/fuel ratio can be adjusted anytime purging
occurs. By going slightly lean with the metered fuel, a larger
amount of purged fuel can be introduced to the engine within
the air/fuel ratio tolerance. The oxygen sensor can then make
the "fine-tune" adjustments to account for the actual
variability in unmetered fuel, along with the typical fuel/air
fluctuations. The purge algorithm can even be designed with a
time variance for the metered fuel adjustment, to account for
the decrease in purged hydrocarbons with increasing purge
time. The approximate amount of purge air flow can also be
incorporated into the lean air/fuel algorithm, so that a purge
failure will force the resulting air/fuel ratio towards
stoichiometry (via the reduction in net air flow).
Both of the above solutions involve changes in the fuel
control strategy, along with a modulating solenoid valve. Due
to potential driveability and emission complications in
expanding the total range of authority of the oxygen sensor,
shifting the scheduled air/fuel ratio during purge may be the
preferred method. To learn if this solution is viable, a
theoretical calculation is shown below for a "worst-case"
vehicle (small fuel consumption with large canister loading),
using the modified air/fuel ratio algorithm. The hypothetical
carbureted vehicle has a city fuel economy of 40 mpg and a 13
gallon fuel tank resulting in a canister loading of 32 grams
(diurnal and hot soak) on 11.5 RVP fuel. By finding the amount
of stoichiometric air consumed by the engine during purge, the
necessary deviation m the air/fuel ratio required to purge 32
grams can be calculated. The stoichiometric air consumption
can be estimated by multiplying the approximate fuel consumed
during purge times, by 14.7 (the stoichiometric air/fuel
ratio). For this theoretical situation, purging is assumed to
occur only during cruise and acceleration modes of the LA-4.
The modal fuel consumption values for this hypothetical
vehicle were prorated from a modal test of a 1987 GMC van (city
fuel economy of 21.7 mpg). Table 4-4 shows the measured fuel
economy, the prorated fuel economy, and the calculated fuel
-------
4-19
Table 4-4
Fuel Consumed by a High Mileage Vehicle
During Potential Purge Times Over LA-4*
GMC
Prorated
LA-4
Mode
Time
Distance
Economy
Economy
Fuel
Hill
(**)
(sec)
(miles)
(mpg) * * *
(mpg)* * * *
(gal)
2
A
42
. 298
13 . 9
25 .6
.0116
C
95
1.39
28 . 7
52.9
. 0263
3
A
20
. 119
13 . 6
25. 1
. 0047
C
17
. 167
29 . 0
53. 5
. 0031
4
A
13
. 064
11. 0
20.3
. 0032
5
A
17
. 105
12 . 5
23 . 0
. 0046
C
27
. 261
36 . 2
66. 7
.0039
6
A
19
.072
12.9
23 . 8
.0030
C
14
. 096
24 . 9
45.9
. 0021
7
A
8
. 024
10.2
18. 8
. 0013
C
35
. 188
20.2
37 . 2
. 0051
8
A
15
. 063
12 . 0
22. 1
. 0029
9
A
22
. 093
14 . 5
26 . 7
. 0035
10
A
18
.086
13.7
25.3
. 0034
11
A
17
. 086
13 . 6
25.1
. 0034
C
163
1.22
32.2
59 . 4
. 0206
12
A
18
. 077
13 . 0
24 . 0
.0032
C
32
.219
34.0
62. 9
. 0035
13
A
18
. 090
14 . 4
26.5
.0034
14
A
22
.088
14 .4
26.5
. 0033
C
19
. 137
25.4
46.8
. 0029
15
A
9
.031
11.9
21.9
. 0014
16
A
20
.066
13.7
25.3
. 0026
C
17
. 099
27.2
50 . 1
. 0020
17
A
23
. 077
13 . 0
24 . 0
. 0032
C
23
. 160
25 . 4
46.8
. 0034
18
A
14
.054
13 .3
24.5
. 0022
Total
757
. 1338
* Potential purge times for this example are the cruise and
acceleration modes of LA-4 "Hills" 2-18 (Hill l - 125
seconds - is allotted for sensor and catalyst warm-up
time).
** A = acceleration, C = cruise. Note: Some hills do not
have a cruise mode.
*** Fuel economy values are based on a modal test of an GMC
Tl 15 (21.7 mpg - city). These values will vary some
between vehicles, but are still useful for approximate
calculations.
**** Fuel economy values prorated for a hypothetical vehicle
with a city rating of 40 mpg.
-------
4-20
consumption, during the allowed purge times in the LA-4 .
Summing the fuel consumed during these periods, this
hypothetical vehicle will consume approximately .13 gallons
(360 grams) of fuel. The stoichiometric air consumed during
this same period is therefore 5,290 grams.
An acceptable air/fuel ratio perturbation can now be
derived by calculating the air/fuel ratio for two extreme
cases: a HC laden purge, and a purge with no hydrocarbons (air
only purge). The purge volume flow rate is assumed to be
constant for both cases.
Preliminary Mass Equations
Air-rotal = Airnetered + Airpurged
Fue 1 T o t a 1 Fuelfteiered + Fuel Purged
HC Laden Purge (Purge air/fuel ratio = 1.5, typical for
a well-loaded canister)
AirPurged = 32g x 1.5 = 48g
Resulting air/fuel ratio to engine
= ( Air-To t a 1 )/ (Fuel To t a 1 )
= (5290 + 48)/(360 + 32) = 13.6
Deviation from stoichiometry
= 13.6 - 14.7 = -1.1
Air Only Purge
Volume flow rate for HC laden purge
= (48g)/(29g/mole) + (32g)/(58g/mole)
= 2.2 moles
Equivalent mass of air
= (2.2moles)/(29g/mole) = 64g
Resulting air/fuel ratio to engine
= (5290 + 64)/(360) = 14.9
Deviation from stoichiometry
=4.9 -14.7 = +0.2
-------
4-21
The total perturbation range is 1.3 air/fuel ratios
(14.9-13.6), or —. 65 from the set point. The -.65
deviation is within a -1 air/fuel ratio range of authority of
the feedback loop which was previously mentioned as already
existing in some cars. In this example, the scheduled air/fuel
ratio could be set to 15.1, which will still allow rich and
lean deviations (due to purge), well within the 14.7 + 1 range
of authority of the fuel control loop. Therefore, shifting the
scheduled air/fuel ratio during purge should be sufficient to
provide adequate purge control with 11.5 RVP fuel.
As was shown by this theoretical calculation, purging a
diurnal and hot soak canister load during an LA-4 should be
feasible without causing an increase in exhaust emissions.
Since this is a worse case purge condition, the purging of the
one-plus hot soak canister load during the FTP should cause
even less problems. Therefore, post-1989 vehicles should still
be able to certify to their HC emissions standards, despite a
change to 11.5 psi RVP certification fuel, with only these
minor, purge-related modifications.
5. Summary
This section has discussed the possible canister and purge
changes that may be needed due to a change in certification
fuel RVP and test procedure. Increases in canister capacity
can be accomplished by using improved charcoals, changing the
canister configuration, using a heated purge, or most
obviously, by increasing the canister size.
While canister size and purge rates are interrelated,
changes may also be required in purge rates. However, problems
arise with purge when the resulting air/fuel ratio exceeds the
capability of the feedback system to maintain stoichiometry.
Therefore, purge control is also a key issue.
Improved control for higher RVP certification fuels can be
accomplished by a variety of options. Designs may incorporate
combinations of both increased canister size and improved purge
control. For computer-equipped vehicles, use of a solenoid
purge valve with frequency modulation, and possibly a shift of
the scheduled air/fuel ratio during purge (both controlled by
the onboard computer) was investigated here and shown to be
sufficient. None of the necessary changes are expected to
incorporate computer hardware changes. The costs of changing
the canister and purge control is discussed in the following
section.
-------
4-22
Non-feedback equipped vehicles, mainly heavier HDVs, are
not expected to have as much difficulty meeting their exhaust
HC standard with higher RVP fuel. For one reason, the fuel
consumption of a HDV is relatively high, so the effect of purge
HC on air/fuel ratio will be relatively small. With LDVs and
LDTs, the changes in the air/fuel ratio are thought to result
in an oxygen deficiency in the catalytic converter, resulting
in increased HC emissions. However, since heavy HDVs do not
use catalytic converters, this increase can not happen. Also,
the HDV exhaust HC standard is less stringent, so they should
be able to compensate for any purge-related exhaust HC effect
with only calibration changes. However, since heavy HDVs only
represent a small portion of the gasoline-fueled HDV fleet,
they will not be treated separately in the following cost
section. This treatment of heavy HDV vehicles will result in a
worse case cost analysis.
II. Costs
A. Vehicle Hardware
This section will describe the method by which EPA has
estimated the costs associated with the changes in evaporative
control technology previously discussed. The costs described
are for the future certification fuel, which will range
somewhere between 8.0 and 11.5 RVP. The values for 9.0 RVP
only represent the estimated costs for the proposed change in
the certification test procedure, since the present
certification fuel is 9.0 RVP.
The costs of control are developed as they pass from the
vendor to the vehicle manufacturer to the dealer and ultimately
to the consumer. The Retail Price Equivalent (RPE) represents
the ultimate cost to the consumer and includes all of the
markup increases seen along the way. The potential affect of
this price increase on vehicle sales is addressed at the end of
this section. All prices are presented in 1986 dollars, with
adjustments from other years based upon the Consumer Price
Index (CPI) for new cars as published in the MVMA 1986 Motor
Vehicle Facts and Figures.[9]*
Some of the values used m this analysis came from 1983
studies. As indicated by MVMA, the increase in the CPI
between 1983 and 1985 is 6.2 percent. Based on this
value, the estimated increase in the CPI from 1985 to 1986
is 3.1 percent, for a net increase between 1983 and 1986
of 9.5 percent.
-------
4-23
The vendor costs, manufacturer costs, markups, and RPE are
summarized for each possible certification fuel (8.0 and 11.5
RVP) in Tables 4-5, 4-6, and 4-7 for LDVs, LDTs, and HDVs,
respectively. Since the development and certification costs
are assumed to be amortized over a 5 year period, Tables 4-5
through 4-7 show both the short-term costs (within five years),
and long-term costs (greater than five years).
1. Vendor Level
In this particular case, "vendor" refers to both the
canister manufacturer and the purge control valve manufacturer,
both which may be the vehicle manufacturer in some cases. For
an increase in certification fuel RVP, the need to build a
larger capacity canister and make adjustments to the purge
system will lead to certain cost increases. These costs can be
divided into: 1) a larger amount of charcoal, 2) larger
canister components, 3) retooling, 4) purge control hardware,
and 5) revision of the software for the onboard computer. For
a decrease in certification fuel RVP, reduction of the canister
capacity will incorporate savings for the charcoal and canister
components, but will still require a retooling cost. The
vendor will also include overhead (20 percent) and profit (20
percent) m the price that is passed on to the vehicle
manufacturer.[10]
As the canister size changes, the amount of carbon
required will also change with the volume of the canister. The
change required for a given certification fuel (8.0 to 11.5
RVP), was developed in section 4.1 and is also summarized there
in Table 4-3. The charcoal cost used is a vehicle-sales
weighted average of the cost of the various types of carbon
currently used by vehicle manufacturers. Tables 4-5 through
4-7 show the calculated costs for the increase/decrease in
charcoal for each possible certification fuel, (8.0 to 11.5
RVP). The cost reductions for 8.0 and 8.5 RVP are the
negatives of the cost increases for 10.0 and 9.5 RVP,
respectively, since the change in uncontrolled evaporative
emissions is nearly linear in this range. Since the costs
associated with all potential changes in RVP are extremely
small, the error in this assumption is negligible. All of
these costs include the vendor overhead and profit mentioned
previously. The details of these cost calculations, except for
the markups for overhead and profit, are given in Appendix 4-A
at the end of this chapter.
The canister components are assumed to increase in size in
proportion to the increase in the total area of the canister.
Therefore, they are treated separately from the carbon. The
prices for the relevant components: the body, the grid, the
filters, the caps, and the connectors have been taken from a
-------
4-24
Table 4-5
LDV - Vendor Cost, Manufacturer Costs and Retail
Price Equivalent Increases (1986 Dollars)
Short-terra Costs (5 years or Less)
Cost Breakdown Certification Fuel RVP
8.0
8 5
9 . 0
9 . 5
10 . 0
10 . 5
11 0
11 5
Vendor Costs:
Charcoal*
-.45
-.22
0
. 22
45
. 69
.96
1 . 25
Canister*
-. 19
-.09
0
09
. 19
. 29
.39
. 50
Tooling
. 08
. 08
0
. 08
.08
. 08
. 08
. 08
Valve
0
0
. 18
. 34
. 49
. 65
. 80
.96
ECU
0
0
.07
. 07
.07
.07
. 07
. 07
Manuf. Costs:
- . 56
-.23
.25
. 80
1 . 28
1 .78
3 .30
2 . 86
RD&T
. 12
. 12
. 12
. 13
14
.15
16
. 17
Cert.
. 71
.71
.71
. 71
. 71
.71
.71
. 71
Subtotal
. 27
. 60
1 . 08
1 . 64
2 13
2. 64
3 . 17
3 . 74
Profit* *
. 05
. 12
.21
.31
.41
. 51
.61
. 72
Dealer Costs:
. 32
. 72
1 . 29
1 . 95
2 . 54
3. 15
3 . 78
4 .46
Profit***
. 02
.04
. 07
. 11
14
. 18
22
.25
RPE
. 34
.76
1.36
2 . 06
2.68
3 .33
4 . 00
4 .71
Long-
-Term Costs (Greater
than 5
years)
Cost Breakdown
Certification Fuel RVP
8.0
8.5
9 . 0
9 . 5
10.0
10. 5
11.0
11 . 5
Vendor Costs:
Charcoal*
- . 45
-.22
0
. 22
. 45
. 69
. 96
1 . 25
Canister*
-.19
-.09
0
. 09
. 19
. 29
.39
. 50
Tooling
0
0
0
0
0
0
0
0
Valve
0
0
. 18
. 34
. 49
. 65
. 80
.96
ECU
0
0
0
0
0
0
0
0
Manuf. Costs:
-.64
-.31
. 18
. 65
1 . 13
1 . 63
2.15
2 . 71
RD&T
0
0
0
0
0
0
0
0
Cert.
0
0
0
0
0
0
0
0
Subtotal
-. 64
-.3i
. 18
. 65
1. 13
1 . 63
2. 15
2 . 71
Profit**
-.12
-.06
. 03
. 12
. 22
31
.41
. 52
Dealer Costs:
-. 76
-.37
.21
. 77
1.35
1 .94
2 . 56
3 . 23
Profit***
-.04
-.02
.01
. 04
08
.11
. 15
. 18
RPE
-.80
-.39
. 22
.81
1 . 43
2 . 05
2 .71
3.41
* Prices include 40 percent vendor mark-up for overhead and profit.
** A mark-up of 19.2 percent was used for corporate overhead and profit.
*** A mark-up of 5.7 percent was used for dealer profit.
-------
4-25
Table 4-6
LDT - Vendor Cost, Manufacturer Costs and Retail
Price Equivalent Increases (1986 Dollars)
Short-Term Costs (5 Years or Less)
Cost Breakdown Certification Fuel RVP
00
o
8 . 5
9 . 0
9.5
10 . 0
10 . 5
11.0
in
r-i
Vendor Costs:
Charcoal*
-.62
-.29
0
. 29
. 62
.95
1 . 32
1 . 72
Canister*
- 22
10
0
. 10
. 22
.34
. 46
. 57
Tooling
. 08
. 08
0
. 08
. 08
. 08
. 08
08
Valve
0
0
. 18
. 34
49
. 65
. 80
. 96
ECU" '
0
0
. 07
. 07
. 07
. 07
. 07
. 07
Manu. Costs:
-.76
-.31
. 25
88
1 .48
2 . 09
2 73
3 . 40
RD&T
. 22
. 22
.22
. 24
.25
.27
. 29
31
Cert.
. 89
. 89
89
. 89
. 89
. 89
89
89
Subtotal
35
.80
1 .36
2 . 01
2 . 62
3 . 25
3 . 91
4 . 60
Prof it**
. 07
. 15
. 26
. 39
. 50
. 62
. 75
.88
Dealer Costs:
. 42
. 95
1 . 62
2 . 40
3 12
3.87
4 . 66
5 . 48
Profit***
. 02
. 05
. 09
. 14
. 18
. 22
. 27
.31
RPE
. 44
1 . 00
1 . 71
2 . 54
3.30
4 . 09
4 93
5 . 79
Long-
¦Term Costs (Greater
Than 5
Years)
Cost Breakdown
Certification Fuel RVP
8.0
8.5
9.0
9 . 5
10 . 0
10.5
11.0
11 . 5
Vendor Costs:
Charcoal *
-.62
-.29
0
. 29
.62
.95
1. 32
1 . 72
Canister*
-.22
-.10
0
. 10
.22
.34
. 46
. 57
Tooling
0
0
0
0
0
0
0
0
Valve
0
0
. 18
.34
.49
. 65
. 80
. 96
ECU
0
0
0
0
0
0
0
0
Manu. Costs:
-.84
-.39
. 18
. 73
1 .33
1 .94
2 . 58
3 .25
RD&T
0
0
0
0
0
0
0
0
Cert.
0
0
0
0
0
0
0
0
Subtotal
- . 84
-.39
. 18
. 73
1.33
1 . 94
2 . 58
3.25
Prof it* *
- . 16
-.07
. 03
. 14
. 26
.37
. 50
. 62
Dealer Costs:
-1 . 00
-.46
.21
.87
1 . 59
2.31
3 . 08
3.87
Prof it* **
-.06
-.03
.01
. 05
. 09
. 13
. 18
.22
RPE
-1 . 06
- . 49
.22
.92
1 . 68
2 . 44
3 . 26
4 . 09
* Prices include 40 percent vendor mark-up for overhead and profit.
** A mark-up of 19.2 percent was used for corporate overhead and profit.
*** A mark-up of 5.7 percent was used for dealer profit.
-------
4-26
Table 4-7
HDV - Vendor Cost, Manufacturer Costs and Retail
Price Equivalent Increases (1986 Dollars)
Short-Term Costs (5 Years Or Less)
Cost Breakdown Certification Fuel RVP
8.0
8.5
9 . 0
9 . 5
10 . 0
10 . 5
11.0
11 . 5
Vendor Costs:
Charcoal*
- . 77
-.38
0
.38
. 77
1. 19
1.65
2. 16
Canister*
-.39
-.20
0
.20
.39
. 60
.81
1 . 04
Tooling
08
. 08
0
. 08
. 08
. 08
. 08
. 08
Valve
0
0
. 18
.34
. 49
. 65
.80
. 96
ECU
0
0
.07
. 07
. 07
. 07
. 07
. 07
Manu. Costs:
-1 . 08
-.50
.25
1 .07
1.80
2.59
3.41
4.31
RD&T
. 49
.49
. 49
. 5S
. 61
. 67
.73
. 79
Cert.
1 . 74
1 74
1 74
1 . 74
1 . 74
1 . 74
1 .74
1 . 74
Subtotal
1 . 15
1 . 73
2.48
3.36
4 . 15
5 . 00
5.88
6 . 84
Profit**
. 23
.34
. 49
.67
.82
. 99
1 . 16
1 .35
Dealer Costs:
1 . 38
2 07
2.93
4 . 03
4.97
5 .99
7 . 04
8 . 19
Prof it* * *
. 09
. 13
. 18
.25
.31
.37
.44
. 51
RPE
1 . 47
2 . 20
3.11
4.28
5 . 28
6.36
7 .48
8. 70
Long-
-Term (
-osts (Greater
Than 5
Years)
Cost Breakdown
Certification Fue1 RVP
8 . 0
8 . 5
9 . 0
9 . 5
10. 0
10 . 5
11 . 0
11 . 5
Vendor Costs:
Charcoal*
-. 77
-.38
0
.38
. 77
1. 19
1.65
2. 16
Canister*
-.39
-.20
0
. 20
.39
. 60
.81
1 . 04
Tooling
0
0
0
0
0
0
0
0
Valve
0
0
. 18
.34
. 49
. 65
.80
. 96
ECU
0
0
0
0
0
0
0
0
Manu. Costs:
-1. 16
-.58
. 18
.92
1 . 65
2.44
3.26
4 . 16
RD&T
0
0
0
0
0
0
0
0
Cert.
0
0
0
0
0
0
0
0
Subtotal
-1 . 16
- . 58
. 18
.92
1 . 65
2 . 44
3.26
4 . 16
Prof it**
-.23
-.11
. 04
. 18
.33
.48
. 65
. 82
Dealer Costs:
-1 .39
- . 69
.22
1.10
1.98
2.92
3.91
4 . 98
Profit***
-.09
-.04
. 01
. 07
. 12
. 18
.24
.31
RPE
-1 . 48
-.73
.23
1 . 17
2. 10
3. 10
4 . 15
5.29
* Prices include 40 percent vendor mark-up for overhead and profit.
** A mark-up of 19.8 percent was used for corporate overhead and profit.
*** A mark-up of 6.2 percent was used for dealer profit.
-------
4-27
1983 draft report entitled, "Manufacturing Costs and Retail
Price Equivalent of Onboard Vapor Recovery System for
Gasoline-Filling Vapors," prepared by LeRoy Lindgren under
contract to the American Petroleum Instltute.*[11] This report
has been reviewed in a previous document, and the costs have
been modified slightly to correct for some arithmetic errors
and discrepancies in markups for overhead and profit.[12] The
calculations to determine the increased cost for each
certification fuel RVP are detailed in Appendix 4-A, and the
results are presented in Tables 4-5 through 4-7. These include
labor costs, along with markups for vendor overhead and profit.
Retooling costs will also be incurred by the canister
manufacturer. Total tooling costs are specified by Lindgren as
$0.16 per canister.[ 11 ] Complete retooling will not be
required, though, since the current canister sizes used on
vehicles with relatively large emissions will still be
appropriate for many of the vehicles with relatively smaller
emissions. Thus, it is assumed that the tooling cost
associated with an increase m canister size will be
approximately half of this cost, or $0.08 per canister. Due to
amortization , the long-term retooling costs incurred by a
change m certification fuel will be $0. These costs are
summarized in Tables 4-5 through 4-7.
The costs for a purge control valve are shown m Table 4-5
through 4-7. A detailed derivation of the purge control valve
required for 11.5 RVP fuel is shown in Appendix 4-A. The valve
cost for 9.0 RVP fuel only represents the additional control
needed for the proposed change in the certification test
procedure. Due to the lack of precise fuel control, carbureted
vehicles are less able to quickly and accurately adjust to
changes in purged hydrocarbons. Therefore, it is predicted
that all of the future carbureted vehicles (an expected 12
percent of the 1990 fleet) will require the frequency modulated
solenoid valve to meet the emission standards with the new test
procedure on 9.0 RVP fuel. Fuel injected vehicles should not
require a valve change for this condition, due to their tighter
control of fuel metering. Using a $.80 cost for the existing
pneumatic valves typically found on carbureted vehicles,** this
This report was later updated as the "Revised Report of
API/LHI Cost Estimate of Onboard Vapor Recovery System and
Review of EPA Technical Report EPA-AA-SDSB-84-01 on
Feasibility, Cost, and Cost Effectiveness of Onboard Vapor
Control," September 28, 1984. Since the costs cited here
were not changed significantly, this analysis continued
with the original values.
A survey was made of the descriptions of emission control
systems m the 1985 Mitchell Manuals.[13]
-------
4-28
hardware change results m an additional cost of $1.50 (case
2.b in Table 4-A-10 in Appendix 4-A). This additional cost for
carbureted vehicles results in an average fleet cost of $.18
per vehicle.
The valve cost for certification fuels between 11.5 and
9.0 RVP was made by linear interpolation. Since the 8.0 or 8.5
RVP certification fuel scenarios would result in a reduction in
purge requirements, the use of an improved purge valve is not
expected. The valves used for light-duty vehicles should also
be adequate for light-duty trucks and heavy-duty vehicles.
Therefore, the valve cost for all three classifications are the
same.
The proposed purge solenoid valve will be operated by the
onboard computer (electronic control unit - ECU). To modify
the software for these existing units, a $.07 per-vehicle cost
was allocated. This cost is the same as that assumed necessary
to modify the software for an onboard refueling vapor control
system.[14] No ECU modifications are expected for 8.0 and 8.5
RVP certification fuel.
The sum of the above costs represent the total vendor
level costs, which are then passed on to the vehicle
manufacturer. However, where the canister manufacturer is the
same as the vehicle manufacturer, as it is in many cases here,
this is primarily a transfer cost. For an ll.5-psi RVP
certification fuel, the total short-term per-vehicle component
cost to the vehicle manufacturer is $2.86 for LDVs, $3.40 for
LDTs, and $4.31 for HDVs. The long-term cost increases for the
vehicle manufacturers only include the increase in hardware
costs (charcoal, canister, valve). For an 11.5-psi RVP
certification fuel, the total long-term per-vehicle component
cost to the vehicle manufacturer is $2.71 for LDVs, $3.25 for
LDTs, and 4.16 for HDVs.
2. Manufacturers Level
The vehicle manufacturer must purchase (or transfer) the
canister and control technology at the vendor level cost rate.
The manufacturer will also face research, development, testing,
and certification costs associated with the implementation of
the improved technology. There are also corporate overhead and
profit that are incorporated into the price that is passed on
to the dealer.
Research, development, and testing costs will be incurred
by the manufacturer to recalibrate the fuel metering and
emission control systems. The additional purge control
hardware was included at the vendor level.
-------
4-29
Estimates for these costs are difficult to determine. It
will be assumed here that, on average, for an 11.5-psi RVP
certification fuel standard, 25 vehicle tests, two months of
technician time and one month of engineering time will be
sufficient to recalibrate each engme-evaporative family
combination. Testing costs for LDVs and LDTs were obtained
from earlier EPA work on light-duty certification costs.[15]
For HDVs, the cost used here was that determined in the
previous HDE rulemaking which instituted HDV evaporative HC
controls.[ 16 ] It is further assumed that this cost is
proportional to the control required. A vehicle acquisition
cost, held constant for all levels of control, is also
included.[12] This methodology is detailed more in Appendix
4-A.
The research, development and testing costs (RD&T) for
LDVs, LDTs, and HDVs are summarized in Tables 4-5, 4-6, and
4-7, respectively. The costs for RD&T were adjusted for 2
years of development time at 10 percent interest, and then
amortized over 5 years to reflect the spreading out of payments
by the manufacturer. For an 11 5-psi RVP certification fuel,
this yields $.17/vehicle for LDVs, $.31/vehicle for LDTs, and
$. 79/vehicle for HDVs. Since all RD&T costs will be recovered
over a 5 year period, the RD&T costs will be $0 on a long-term
base.
Manufacturers of LDVs, LDTs, and HDVs will incur a cost to
certify their fleets with a new certification fuel. The
LDV/LDT costs were obtained from the EPA work previously
cited.[12] Because year to year carryover of engine families
is not 100 percent, only 90 percent of the LDV/LDT
recertification costs would be attributable to this change.
Including one year of opportunity cost at 10 percent and then
amortizating over 5 years results in a $.71/vehicle LDV and
$.89/vehicle LDT cost impact. The HDV certification costs are
based on a previous Regulatory Impact Analysis and a Regulatory
Support Document.[17,18] Following the same discounting and
amortization procedure, the resulting cost is $ 1.74/vehicle.
The development details for the LDV/LDT/HDV certification costs
are shown in Appendix 4-A. As with RD&T costs, there are no
long-term price increases due to certification costs.
Before passing on the cost to the dealer, the vehicle
manufacturer will add markups for corporate overhead and
profit. The total manufacturer markup according to the vehicle
type (LDV, LDT, HDV) were determined under a contract to Jack
Faucett Associates [19], and are respectively footnoted in
Tables 4-5 through 4-7. These markups are applied to the sum
of the total vendor level costs, the research and development
costs, and the certification costs.
-------
4-30
The total cost at the manufacturer level, then, is the sum
of the hardware, development, certification, and markup costs.
It is assumed that no significant retooling will be required at
the assembly level — only at the canister manufacturing level.
3. Dealer Level
The total increase in cost seen by the dealer is shown m
Tables 4-5 through 4-7. The dealer is expected to make a
reasonable profit from the sale of a vehicle. Therefore, a
markup which recovers incremental costs and yields a fair
return on incremental investment must be included in the total
cost This markup was also developed under the previously
mentioned contract to Jack Faucett Associates.[19] The dealer
markups are footnoted in each table and ranged from 24.9 to
26 0 percent.
4. Consumer Level (Retail Price Equivalent)
The bottom lines of the short-term and long-term sections
of Tables 4-5 through 4-7 show the increase in cost expected to
be seen by the consumer. The resulting short-term RPE increase
associated with a change to an 11.5-psi RVP certification fuel
is $4.71/vehicle for LDVs, $5.79/vehicle for LDTs, and
$8.70/vehicle for HDVs. The resulting long-term RPE increase
associated with a change to an 11.5-psi RVP certification fuel
is $3.4l/vehicle for LDVs, $4.09/vehicle for LDTs, and
$5.29/vehicle for HDVs.
5. Impact on Vehicle Sales
Substantial changes in vehicle price do affect vehicle
sales. A typical model to calculate the impact on vehicle
sales is the price elasticity of demand. However, a 0.05
percent change in price ($4.71 change for a $10,000 light-duty
vehicle) is too insignificant to be accurately handled by this
model. In any event, the EPA expects a $4.71 to $8.70 change
in price (LDV to HDV), to have a negligible effect on vehicle
sales. This price increase should especially be of little
consequence in relation to annual price increases occurring at
the time of a new model year introduction.
B. Effects of Increased Canister Size on Operating Costs
There will be a very slight reduction m fuel economy
associated with the increased weight of the canister in the
modified evaporative control system. This will affect a
vehicle's operating cost, and must therefore be included in the
total costs associated with the proposed changes to the
certification test procedure. Key values used in the
calculation of this weight penalty are summarized in Table 4-8.
-------
4-31
Table 4-8
Summary of Values Used in Calculation
LDV LPT HDV
Fuel Economy (mpg)* 23.19 17.72 9.208
Vehicle Weight (lb)** 3082 3832 9270
Weight Sensitivity 0.329 0.402 0.450
Factor (% change
in fuel economy per
% change in weight)
* * *
Discounted Lifetime 65,400 80,900 71,700
Mileage****
* 1985 values from "MOBILE3 Fuel Consumption Model," Mark A.
Wolcott, EPA, and Dennis F. Kahlbaum, CSC, September 1985.
** "Light-Duty Auto Fuel Economy...Trends Through 1985,"
Heavennch, Murrell, Cheng and Loos, SAE 850550.
*** "Analysis Memorandum: Design Factor Update," prepared by
Energy and Environmental Analysis, Inc., for EPA, October
1, 1982.
**** "Regulatory Impact Analysis, Oxides of Nitrogen Pollutant
Specific Study and Summary and Analysis of Comments,"
EPA/OAR/OMS, March 1985. (Discounted at 10 percent over
life of vehicle.) [Reference 17 of Chapter 4.]
-------
4-32
Appendix 4-A describes how the component costs for each
certification fuel RVP were calculated. The weight associated
with each component of an 850-ml canister is given in Table
4-A-5. By scaling up these component weights, as was done for
the costs (described in Appendix 4-A), the increased canister
weight associated with each certification fuel RVP can be
determined. These are provided in Table 4-9.
These weight increases can be expressed as a percentage of
the total weight of the vehicle using the estimates of total
vehicle weights shown in Table 4-8. Estimated weight
sensitivity factors, which relate a percentage increase in
weight to a percentage reduction in fuel economy, and
class-average fuel economies are also given in Table 4-8 and
are used along with the percentage weight increases to
determine the expected reductions in fuel economy.
In their comments to the November 1985 study, GM took
exception to some of the values which were used in the
calculation.[20] GM stated that a "weight compounding" factor
of 1.7 should be applied to the additional canister weight to
account for "additional weight in the structure and powertrain
in order to maintain equivalent structural integrity and
vehicle performance." While recognizing the validity of weight
compounding in general, it does not seem practical that the
added one to two pounds to the canister will have any farther
reaching affects than possibly to the bracketing which holds it
in place. [21] Thus, as can be seen in Table 4-9, a weight
compounding factor of 1.1 has been added to account for
possible small changes in bracketing.
In addition to the weight compounding factor, GM also took
exception to the 0.329 weight sensitivity factor used for light
duty vehicles. They instead promoted a factor of 8.5
gallons/10,000 mi/100 lbs. which is approximately equivalent to
a 0.608 weight sensitivity factor depending on the vehicle
weight and fuel economy chosen. The GM factor is made up of
three components: 1) a factor recognizing the differences in
vehicle inertia (about 55 percent of the total), 2) a factor to
account for increases in rolling resistance (about 15 percent),
and, 3) a factor to account for necessary changes in vehicle
performance (about 30 percent). The first portion attributable
to inertia effects and the 0.329 EPA sensitivity factor are
very close in magnitude, even though the EPA factor may also
have taken into account much of the rolling resistance which GM
added separately. The EPA value was based on a much larger and
broader data set (over 5,000 cars of 1975 through 1981 model
years from more than 20 different manufacturers), and therefore
should prove to be more reliable. In addition, since road load
may have been increased with inertia weight settings during
testing, the rolling resistance should not be factored in
separately.[21]
-------
4-33
Table 4-9
1985 Vehicle Weight Penalty Associated with Various Certification Fuels
LDT
HDT
Certification
Fuel
RVP's (psi)
8.0
8.5
9.0
9.5
10.0
10.5
11.0
11.5
LDV
Increased Weight (lb)
-0.35
-0.16
0
0.19
0.45
0.66
0.88
1.19
Adjusted weight (xl.l)
-0.385
-0.176
0
0.209
0. 495
0.726
0.979
1.309
Reduced Fuel Economy
(gal/mi x 10~6)
-1.766
-.807
0
0.958
2.270
3.329
4.490
6.003
Extra Fuel
(gal/useful life)
-0.115
-0.053
0
0.063
0.148
0.218
0.294
0.393
Cost ($/vehicle)
-0.094
-0.043
0
0.052
0.121
0.179
0.241
0.322
Increased weight {lb) -0.45 -0.22 0 0.26 0.58 0.88 1.21 1.56
Adjusted weight (x 1.1) -0.495 -0.242 0 0.286 0.638 0.968 1.331 1.716
Reduced Fuel Economy
(gal/mi x 10"6> -2.93 -1.433 0 1.693 3.777 5.731 7.881 10.160
Extra Fuel
(gal/useful life) -0.237 -0.116 0 0.137 0.306 0.464 0.638 0.822
Cost ($/vehicle) -0.194 -0.095 0 0.112 0.251 0.380 0.523 0.674
Increased weight (lb) -0.62 -0.32 0 0.41 0.81 1.25 1.70 2.20
Adjusted weight (x 1.1) -0.682 -0.352 0 0.451 0.891 1.375 1.870 2.420
Reduced Fuel Economy
(gal/mi x 10"B) -3.596 -1.856 0 2.378 4.697 7.249 9.859 12.758
Extra Fuel
(gal/useful life) -0.258 -0.133 0 0.170 0.337 0.520 0.707 0.915
Cost ($/vehicle) -0.212 -0.109 0 0.139 0.276 0.426 0.580 0.750
-------
4-34
The final portion of GM's sensitivity factor is a
significant amount for a "performance adjustment." Based on
EPA's own data, GM's performance adjustment, if it was
applicable, is only nearly half of what would be appropriate.
However, just as it seems unreasonable to allow for weight
changes to the vehicle structure and power train for at most a
1 to 2 pound increase in canister weight, it also seems
unreasonable to think that a manufacturer would change the size
of future engines to correct for such a small loss in
performance.[21] Thus it does not appear appropriate to change
the 0.329 weight sensitivity factor from the previous analysis.
To further quantify the weight penalty, the reductions in
fuel economy calculated using the weight sensitivity factors
are then coupled with estimates of lifetime vehicle mileages
(shown in Table 4-8) to calculate the extra gallons of fuel
used over the vehicle's life. These lifetime mileages are
discounted (at 10 percent per annum) to represent the fact that
a dollar saved in the tenth year of vehicle use is not the same
as one saved m the first year. The net cost to the consumer
is then determined using a fuel cost of $0.82 per gallon.
Table 4-9 summarizes the calculated values for reduction in
fuel economy, extra gallons of fuel, and cost.
The costs shown in Table 4-9 are accurate for the 1985
model year, as the calculations were performed on 1985 data.
Since vehicle fuel economy is expected to increase in the
future, these costs thus represent a worst case situation.
Accurate estimates of future costs can not be made since
although there are future fuel economy estimates, there are no
accurate future vehicle weight estimates. Since future vehicle
weight is expected to decrease while fuel economy increases,
assuming current vehicle weights for the future would severely
under estimate the weight penalty costs.
C. Body Modification Costs
Section I estimated that roughly a 60 percent increase in
canister volume may be required to achieve control of
evaporative emissions from vehicles designed for 11.5 RVP
fuel. The question has been raised as to whether current
engine compartments could accommodate canisters that were 60
percent larger than those presently being used. A number of
industry sources have claimed that extensive and costly vehicle
modifications would be required if manufacturers were forced to
certify using higher RVP fuel. However, upon further
investigation, it appears that the problem is neither as
widespread nor as difficult to solve as it has been pictured,
and that in the vast majority of cases, canisters large enough
to handle evaporative emissions from 11.5 RVP fuel (worse case)
-------
4-35
could be accommodated without vehicle modifications. For even
the most difficult situations, there are a number of options
open for the manufacturer to gain the necessary additional
capacity without incurring costs for modifying the vehicle.
First of all, not all vehicles will require larger
canisters. Some vehicles, particularly those with smaller fuel
tanks, may already have excess canister capacity. Most
manufacturers make only two or three canister sizes to service
their entire vehicle line, so that a canister with sufficient
capacity for a mid-size car with a V-6 engine may also be used
for a 4-cylinder vehicle with a much smaller fuel tank, Then
too, the growing trend toward fuel injection means that less
canister capacity will be required for most applications, due
to the typical decrease in hot soak emissions from
fuel-injected vehicles.
Those vehicles that do require additional capacity may not
need to physically increase canister size. For example, some
manufacturers could gain the additional adsorptive capability
needed for the higher RVP fuel by switching to charcoal with a
greater specific working capacity than they are currently
using. Of course, this option would not be available to the
industry as a whole, since some manufacturers already use
charcoal with a high specific working capacity. These
manufacturers may find it necessary to increase their current
canister volumes by up to 60 percent. However, as will be seen
below, there is ample room in most engine compartments for the
relatively modest increases required. The actual increases
required in canister dimensions are smaller than might be
expected. For example, a cylindrical canister of approximately
2 liters capacity (roughly 5" in diameter and 6 1/4" long)
would require only a 1 3/8" increase in diameter (28 percent)
to increase the volume to 3.2 liters. Similarly, a one liter
rectangular canister, such as might be used in a subcompact or
compact, with dimensions of 2 inch (") by (x) 5" x 6" would
only require a 60 percent increase in the smallest dimension
(i.e., from 2" to 3.2").
For those vehicles with crowded engine compartments where
increasing the size of the canister might be more difficult,
there are also a number of options available. Baffles could
possibly be used to change the effective length to diameter
ratio. Or, the actual shape of the canisters could be changed
to fit the available space. A change from a cylindrical to a
rectangular canister of the same dimensions, for example,
results in a 27 percent increase in volume with a slightly less
increase in capacity. Finally, for the relatively few
worst-case situations the canister could be relocated or a
multiple-canister approach could be used.
-------
4-36
As a rough gauge of whether current vehicle designs could
accommodate canisters that were 60 percent larger than those
now in use, a spot check was made of the space available for
canisters in a number of late-model vehicles. An attempt was
made to identify the configurations where available space was
likely to be most limited. Those included (1) subcompact or
compact vehicles, particularly those with the short,
close-quarter engine compartments associated with transverse
front-wheel-drive (FWD) powertrains, (2) so-called pony cars,
i.e., small sporty vehicles with large engines, (3) mid-size or
full-size vehicles equipped with a multitude of engine-driven
options, including turbochargers, air conditioning and fuel
injection pumps, as well as the more usual power steering,
power brakes, and air pumps.
Accordingly, a sample of nine vehicles was selected,
including four compacts (two of which were ]ust barely over the
subcompact dividing line), three full size or mid-size cars
with many options, including fuel injection and turbocharging,
and one pony car. No LDTs were inspected, since there is ample
room in most LDT engine compartments for a much larger increase
m canister volume than would be required for 11.5 RVP fuel.
However, one FWD mini-van with a forward-located engine was
checked, again representing a configuration with the most
limited space under the hood.
The vehicles selected consisted of three Ford, three
Chrysler and three GM models, six of the vehicles were FWD with
transverse engines. Four of the vehicles were carbureted and
five were fuel injected. One of the vehicles was
turbocharged. Five vehicles were powered by 4-cylinder engines
while four were equipped with V6 or V8 engines. On five of the
vehicles the evaporative canisters were mounted in the engine
compartment, while on the remainder (three Chrysler products
and one Ford) the canister was mounted in the front fender
ahead of the wheel well. The latter location made it somewhat
difficult to obtain exact canister measurements, however, it
was possible to get a rough approximation of the dimensions and
resultant volumes. It should also be noted that access to
canisters located inside the engine compartment was also
difficult in many cases, due to the canisters being located
under alternators, air conditioning compressors, etc. However,
difficulty of access did not necessarily equate to lack of room
for increasing canister size.
The results of the survey are shown in Table 4-10. In
general, there appears to be room for the required increases in
canister volume in all the vehicles surveyed. As might be
expected, the subcompact designs would present more of a
problem than the larger designs. However, it should be pointed
-------
4-37
Table 4-10
Results of Evaporative Canister/
Engine Compartment Survey
Approximate Approximate Space for
Vehicle/Engine Canister Size Capac1ty Notes* Larger Canister 7
Lynx 1.6L 4 2" x 5" x 5 1/2" .9L F,R yes, but tight
Sable 3.0L V-6 2" x 5" x 5 1/2" .9L E,I,R yes
LTD 5.0L V-8 3" x 7 1/2" x 10" 3.7L F,I,R yes
Sundance 2.2L 4 5" x 5 1/2" 1.8L C,F yes, but tight
LeBaron 2.2L 4 5" x 5 1/2" 1.8L C,F,T yes
Voyager 2.2L 4 5" x 5 1/2" 1.8L C,F yes
Nova 1.6L 4 5 1/2" x 3 1/2" 1.3L C,E yes, but tight**
Camaro 2.8L V-6 5" x 5 1/2" 1.8L C,E yes
Celebrity 2.8L V-6 5" x 5 1/2" 1.8L C,E,I yes
Explanation of Notes:
C - Cylindrical canister
E - Canister located in engine compartment
F - Canister located in fender ahead of wheel well
I - Fuel in]ected
R - Rectangular canister
T - Turbocharged
Would need small increase in both diameter and length for
60 percent volume increase.
-------
4-38
out that the canisters on these vehicles are smaller and would
require proportionally smaller size increases for a 60 percent
volume increase. The necessary increases in dimensions are
feasible for most of the vehicles surveyed; only the smallest
compacts might require a change in more than one major canister
dimension due to available space constraints. For example, the
Nova, which has a short cylindrical canister approximately 5
1/2" in diameter and 3 1/2" long, would require either a 1 1/2"
increase in diameter or a 2 1/8" increase in length to gain 60
percent more volume. Due to the cramped quarters where the
canister is currently located, either of those increases by
itself might present a problem. However, a 1" increase in
diameter and a 1/2" increase in length would result in 60
percent more volume. Alternatively, a rectangular canister of
5 1/2" x 5 1/2" x 4 1/2" would also provide the necessary
increase, since almost 27 percent of the additional 60 percent
could be obtained by switching from a cylindrical canister to a
rectangular canister of the same dimensions. The 1" increase
in length would provide the additional capacity required in
this particular application.
Thus, it appears that few if any vehicles will require
modifications to accommodate canisters suitable for 11.5 RVP
fuel. Some vehicles may not need to increase canister size at
all, either due to excess capacity in current designs or as a
result of switching to charcoal with higher specific adsorptive
capacity. If it should be necessary to increase canister size,
there is ample room in most engine compartments for the
relatively modest size increases that would be required. For
those relatively few cases where the required additional
capacity could not be achieved through a relatively small
increase in one or more canister dimensions, there are other
options such as shaping the canister to fit the available
space, canister relocation, or switching to a multiple-canister
configuration. The cost for body modifications is therefore
assumed to be zero.
III. Conclusions
A worse case or minimal R&D approach was taken in this
analysis. The methods that may be required to increase storage
capacity and purge control appear to be feasible with current
technology. No significant effects upon vehicle performance
other than reduced evaporative emissions are expected from
these changes. In the most extreme case of an 11.5-psi RVP
certification fuel, the maximum final cost to the consumer at
the time of vehicle purchase has been estimated as
$4.7l/vehicle for LDVs, $5.79/vehicle for LDTs, and
$8.70/vehicle for HDVs. By amortizing the development and
certification costs over a 5 year period, the final cost to the
consumer will be reduced to an estimated $3.41/vehicle,
$4.09/vehicle, and $5.29/vehicle for LDV, LD71, and HDV,
respectively.
-------
4-39
References (Chapter 4)
1. "Basic Concepts of Adsorption on Activated Carbon,"
Activated Carbon Division, Calgon Corporation.
2. "Control Characteristics of Carbon Beds for Gasoline
Vapor Emissions," EPA-600/2-77-057, Michael J. Manos and Warren
C. Kelly, Scott Environmental Technology, Inc. for EPA, ORD,
IERL, February 1977.
3. "Activated Carbon for Effective Control of
Evaporative Losses," Ronald S. Joyce, Paul D. Langston, George
R. Stoneburner, Charles B. Stunkarol, and George S. Tobias,
Pittsburgh Activated Carbon Division, Calgon Corp., SAE Paper
#690086, January 1969.
4. Meeting with Howard Shrut, Calgon Corporation, July
18, 1984.
5. Letter from John Urbanic. Calgon Carbon Corporation,
June 18, 1986.
6. Conversation with Bill Kornegay, Ph.D., Technical
Director, Chemical Division, Westvaco Corporation, November 5,
1984 .
7. "Characterization of Fuel/Vapor Handling Systems of
Heavy-Duty Gasoline Vehicles over 10,000 Pounds GVW," Draft
Report, prepared by Jack Faucett Associates for EPA, Contract
No. 68-03-3244, September 27, 1985.
8. "Sensors and Actuators for Automotive Control and
Diagnostics," SAE Seminar by William Wolber, September 25-26,
1986.
9. "MVMA Motor Vehicle Facts and Figures '86," Motor
Vehicle Manufacturers Association of the United States, Inc.,
1986.
10. "Cost Estimations for Emission Control Related
Components/Systems and Cost Methodology Description,"
EPA-460/3-78-002, EPA, March 1978.
11. "Manufacturing Costs and Retail Price Equivalent of
Onboard Vapor Recovery System for Gasoline Filling Vapors,"
LeRoy Lindgren under contract to API, June 1983. (Attachment
to MVMA Comments, "Evaluation of Air Pollution Regulatory
Strategy for Gasoline Marketing Industry," available in Public
Docket A-84-07.)
-------
4-40
12. "The Feasibility, Cost, and Cost Effectiveness of
Onboard Vapor Control," EPA-AA-SDSB-84-01, Glenn W. Passavant,
EPA, March 1984.
13. "Emission Control Service & Repair - Domestic Cars -
1985 Supplement," Mitchell Information Services, Inc 1985.
14. "Costs of Onboard Vapor Recovery Hardware," Mueller
Associates under sub-contract to Jack Faucett under contract to
EPA, February 14, 1985.
15. "Light-Duty Vehicle Certification Cost," memo to
Edmund J. Brune, Director, Certification and Surveillance
Division from Daniel P. Hardin, Jr., Certification and
Surveillance Division Staff, EPA, March 13, 1975.
16. "Regulatory Analysis
Final Emission Regulations for
Heavy-Duty Engines," EPA, OMSAPC,
Public Docket No. OMSAPC-78-4.)
and Environmental Impact of
1984 and Later Model Year
December 1979. (Available in
17. "Regulatory Impact Analysis, Oxides of Nitrogen
Pollutant Specific Study," EPA, March 1985. (Available in
Public Docket No. A-80-18)
18. "Regulatory Support Document for the Final
Evaporative Emission Regulation and Test Procedure for 1984 and
Later Model Year Gasoline-Fueled Heavy-Duty Vehicles," EPA,
January 1983. (Available in Public Docket No. OMSAPC-79-1)
19. "Update of EPA's Motor Vehicle Emission Control
Equipment Retail Price Equivalent (RPE) Calculation Formula,"
prepared by Jack Faucett Associates for EPA, Contract No.
68-03-3244, September 4, 1985. (Available in Public Docket NO.
A-84-07)
20. General Motors Corporation, Letter to EPA, May 13,
1986 .
21. Memorandum, "General Motors' Comments on Fuel
Economy Impact of Onboard Regulation," Chester J. France,
Chief, SDSB to Richard D. Wilson, Director, OMS, July 28, 1986.
22. "Ward's Automotive Yearbook 1986," Ward's
Communications Incorporated, 1986.
23. "Emission Control Technology and Strategy for
Light-Duty Vehicles 1982-1990. Final Report," prepared by
Energy and Environmental Analysis, Inc.
-------
Appendix 4-A
Detailed Derivation of Evaporative
ECS Component Costs
This appendix details the calculations used to derive the
evaporative emission control system (ECS) costs discussed in
Chapter 4. The first section will describe how the average
canister sizes for current evaporative control systems were
determined. This will be followed by sections discussing: l)
canister material costs, 2) purge control valve costs, 3)
research, development and testing costs, and 4) certification
costs associated with the changes in the certification test
procedure.
Average Canister Sizes in Current Vehicles
The average canister sizes for current vehicles were
calculated by using a sales-weighted average cf the canister
sizes currently in use. Only the major manufacturers — GM,
Ford, Chrysler, Toyota, and Nissan — were considered in this
calculation. These manufacturers represent 83 and 90 percent
of the total 1983 gasoline light-duty vehicle (LDV) and
gasoline light-duty truck (LDT) market, respectively
Table 4-A-l shows the canister sizes used by each
manufacturer for various engine families. Table 4-A-2 combines
these values into a single value for light-duty vehicles and
light-duty trucks for each manufacturer, assuming egual sales
per engine family. A forecasted 1990 normalized market share
was generated as described and shown in Table 4-A-3. Using
these market shares for the five manufacturers, a
sales-weighted average canister size for each vehicle class is
determined. These industry-wide average canister sizes are
summarized in Table 4-A-4. As indicated in the table, the
average canister size for heavy-duty gasoline-fueled vehicles
is taken as the current size for GM vehicles, as GM dominates
this market (i.e., two-thirds of sales).
Canister Material Costs
This section details the methodology used to determine the
costs associated with changing the canister size. These costs
will be determined for each potential certification fuel RVP.
Reference 11 in Chapter 4. A final version of the report
has since become available, but the changes were not
significant so the calculations were not redone.
-------
4-A-2
Table 4-A-l
Canister Distribution
Manufacturer
GM
GM
GM
Ford
Ford
Ford
Chrysler
Chrysler
Chrysler
Toyota
Toyota
Toyota
Toyota
Nissan
Nissan
Canister Size (ml)
1500
2500
2500 + 300
925
1400
1400 + 1400
1320
1790
1320 + 1320
835
845
1400
1400 + 645
580
1230
Number of Families
LDV'
29
7
3
1
4
3
2
1
1
6
2
LPT*
2
5
1
8
7
1
6
4
1983 Model Year.
-------
4-A-3
Table 4-A-2
Average Canister Size by Manufacturer
Avg. Canister Size
Vehicle Class Manufacturer for 9.0 RVP Fuel (ml)*
LDV
GM
1500
LDV
Ford
1225
LDV
Chrysler
1521
LDV
Toyota
979
LDV
Nissan
743
LDT
GM
2288
LDT
Ford
1147
LDT
Chrysler
2059
LDT
Toyota
1427
LDT
Nissan
580
Average of 1983 MY canisters assuming equal sales of evap.
f ami 1les.
-------
4-A-4
Table 4-A-3
Market Shares by Manufacturer
1983 Normalized* 1990 Normalized**
Vehicle Class Manufacturer Market Share (%) Market Share (%)
LDV
GM
45.74
46 . 81
LDV
Ford
17 . 78
18 . 20
LDV
Chrysler
9 . 56
9 . 78
LDV
Toyota
13 . 89
13 . 00
LDV
Nissan
13 . 03
12 . 21
LDT
GM
41 . 98
40 . 86
LDT
Ford
34 . 76
33 . 84
LDT
Chrysler
10 . 12
9 85
LDT
Toyota
7.29
8 . 57
LDT
Nissan
5.85
6 . 88
Normalized on the basis of 100 percent of the domestic LDV
and LDT market is shared by GM, Ford and Chrysler, and 100
percent of the imported LDV and LDT market is shared by
Toyota and Nissan.
Normalized as for 1983. Manufacturers' shares of domestic
or imported market is assumed to be the same as for 1983,
but the ratio of imported to domestic registrations
changed in accordance with projections made in Reference
23 of Chapter 4.
-------
4-A-5
Table 4-A-4
Industry-wide Average Canister Size*
Avg. Canister Size
Vehicle Class for 9.0 RVP Fuel (ml)
LDV 1292
LDT 1688
HDG 4 000**
Uses 1989 Market Shares.
Average canister size used by GM,
based on certification records for
the 1985 model year.
-------
4-A-6
Table 4-A-5 shows the canister material costs to vendors
taken from a draft report prepared for API by LeRoy Lindgren.*
The costs presented are for an 850 ml GM canister, which is
smaller than the average canister sizes for current LDV, LDT
and HDV classes shown m Table 4-A-4.
Table 4-A-6 shows these material costs after they have
been scaled up to the average canister sizes for LDV, LDT and
HDV families. The charcoal costs were increased in proportion
to the increase in volume of the canisters. The body, grid,
filter and cap costs were increased in proportion to the
increase in surface area of the canister (For cylinders which
have the same length-to-diameter ratio, the ratio of surface
areas is equal to the ratio of volumes to the two-thirds
power.) The connector cost is independent of the canister size
and therefore remains constant.
The charcoal costs per unit volume were calculated on a
1990 vehicle sales-weighted basis (see Table 4-A-3). GM and
Chrysler are assumed to continue to use wood-based charcoals
(price = $l/lb, density = .26-.30 g/ml) whereas Ford, Nissan,
and Toyota are assumed to continue to use coal-based charcoals
(price = $2/lb., density = .37-.38 g/ml). Tables 4-A-7, 4-A-8,
and 4-A-9 contain the canister material cost increases to meet
a higher RVP certification fuel for LDVs, LDTs and HDVs,
respectively. The increases in volume required for each higher
RVP certification fuel are listed in Table 4-3 and the
increases in canister surface area can be determined from the
volume ratios (as stated earlier). The increases in costs were
calculated by exactly the same method as was used above to
scale up the 850-ml GM canister to the average canister sizes
for each class of vehicle. The charcoal and canister savings
for 8.0 and 8.5 RVP fuels are assumed to equal the charcoal and
canister cost increases for 10.0 and 9.5 RVP fuels,
respectively.
Purge Control Valve Costs
The proposed purge control valves were described in
Chapter 4. In summary, the valves are to slowly initiate any
increases in purge rate and to vary the purge flow rate
according to engine conditions. To accomplish this type of
control, two types of valves were proposed: a pneumatic
diaphragm valve, and an electric solenoid valve with frequency
modulation.
Existing purge systems vary the purge rate according to
air flow to the engine. Most of these systems allow purging
only during cruise or acceleration conditions, when the engine
is above a minimum rpm, and after the engine has had time to
-------
4-A-7
Table 4-A-5
Canister Material Cost to Vendor (Dollars)*
Component
Description
Body
Grid
Int. Filter
Ext. Filter
Charcoal
Cap
Connectors
Material
DB437
G7
AY332
KZI-4
54448
DB437
DB437
Weight
(lbs.)
.30
. 10
10
. 20
. 50
. 10
. 05
Material
Costs ($)**
. 27
. 04
.11
. 23
. 55
. 09
. 04
TOTAL
1 . 33
Taken from Reference 11 of Chapter 4.
Material costs are those for an 850-ml GM canister and
were converted to 1986 dollars from 1983 dollars, using a
net increase of 9.5 percent.
-------
4-A-8
Table 4-A-6
Baseline* Canister Material Costs to Vendor (1986 Dollars)
Component
Description
Vehicle Class
LDV
LDT
HDV
Body
.35
41
. 74
Grid
. 05
. 06
. 12
Int. Filter
. 14
17
. 30
Ext Filter
.30
35
63
Charcoal* *
1 . 48
2 . 04
2 . 55
Cap
. 12
. 14
. 23
Connectors
. 04
. 04
. 04
TOTAL
2.48
3.21
4 . 61
Baseline refers to the average canister sizes for 9.0
RVP fuel shown in Table 4-A-4.
Charcoal cost is the 1990 sales-weighted average cost of
the different types of charcoals. Wood-based charcoal
($l/lb) is used by GM and Chrysler and coal-based
charcoal ($2/lb) is used by Ford, Nissan, and Toyota.
NOTE: Body, filter, grid, and cap costs increase in proportion
to the surface area and charcoal cost increases in
proportion to the volume. (Canister length-to-diameter
ratio is assumed to remain constant.) Connector cost is
independent of the canister volume, thus there is no
increase m material cost to the vendor for connectors
for larger canisters.
-------
4-A-9
LDV
Canister Material
Table 4
Cost Changes
-A-7
at Vendor
Level *
{1986 Dollars)
Component
Certification Fuel
RVP (psi
)
Description
8.0
8.5
9.5
10.0
10.5
11.0
11.5
Body
-.05
-.02
.02
.05
.07
. 10
. 13
Grid
-.01
.00
.00
.01
.01
.01
.02
Int. Filter
- .02
-.01
.01
.02
.03
.04
.05
Ext. Filter
-.04
-.02
.02
. 04
.06
. DQ
. 11
Charcoal
-.32
-.15
.15
. 32
. 50
.68
. 90
Cap
-.02
- .01
.01
.02
.03
.03
. 04
TOTAL
-.46
-.21
.21
. 4b
.70
.05
1.25
Does not include 40-percent vendor mark-up Cor overhead
and profit. No change is needed at 9.0 RVP.
-------
4-A-10
Table 4-A
-8
LDT
Canister
Material Cost
Increase at Vendor
Level*
(1986 Dollars)
Component
Certification Fuel
RVP (psi
)
Desc rlption
8.0
8 . 5
9.5
10.0
10. 5
11.0
11.5
Body
-.06
-.03
. 03
.06
.09
. 12
. 15
Grid
-.01
.00
.00
.01
.01
.02
.02
Int. FiIter
-.02
- .01
.01
.02
.04
.05
06
E t. F11te r
-.05
- .02
.02
.05
.07
. 10
. L 3
Charcoal
-.44
-.21
.21
.44
. o 8
.94
1.23
Cap
-.02
-.01
.01
.02
.03
.04
.05
TOTAL
-.60
-.28
. 28
.60
. 02
1.27
1. 54
Does not: include 40-pei.cent vendor mark-up Cor overhead
and profit. No change is needed at 9.0 RVP.
-------
4-A-ll
Table
4-A-
9
HDV
Canister
Material Cost
Increase at
Vendor Level*
<1986 Dollars)
Component
Certification
Fuel
RVP (psi)
Description
8.0
8 . 5
9.5
10.0
10.5
11.0
11.5
Body
-.10
-.05
.05
. 10
. 16
.21
.27
Gt id
-.02
-.01
.01
. 02
.03
.03
.04
Int. Filtei
-.04
- 02
.02
.04
.06
.09
. 11
Ext. F liter
-.09
- .04
.04
.09
. 13
. 18
. 23
Charcoal
-.55
- . 27
.27
.55
.85
1.18
1. 54
Cap
-.03
-.02
.02
.03
.05
.07
.09
TOTAL
-.83
-.41
.41
.83
1. 28
1.76
2 .28
Does not include 40-percent vendor mark-up for overhead
and profit. No change is needed at 9.0 RVP.
-------
4-A-12
warm-up. A few systems that purge to the air cleaner (which
provides a vacuum source proportional to engine air flow), do
not use control valves. Most systems purge to the throttle
housing or to the intake manifold. These locations provide
greater vacuum, but inversely proportional to air flow.
Therefore, all of these systems require some type of purge
control.
The existing purge control valves use pneumatic, electric,
or combined pneumatic/electric systems. The pneumatic systems
vary from simple on/off valves, to two-stage valves, to more
elaborate pneumatic purge control systems which will even
switch the vacuum source. The pneumatic control valves are
typically integral with the canister housing, but may be of
independent construction.
The electric systems are usually solenoids which may be
simple on/off valves, or two-stage orifice controls. Some
manufacturers are starting to use frequency modulated solenoid
valves, which is the same type as that assumed here for use
here on the majority of vehicles at 11.5 psi RVP. Their
present use indicates the feasibility of incorporating them
into a purge control system, and that the associated cost is
not inhibiting. The solenoid valves are controlled by the
onboard computer. Some hybrid pneumatic/electric purge systems
use an on/off solenoid m the pneumatic signal line to prevent
purging during unwanted conditions.
Table 4-A-10 shows the cost for existing purge control
systems, the cost for the corresponding proposed system, and
the cost differences. Using certification records and the
Mitchell Manual[13] the various existing systems were grouped
into one of four general categories or cases: l) no control
valve; 2) pneumatic valve; 3) solenoid valve; 4) pneumatic/
solenoid valves. The number of vehicles with modulating
solenoid valves were few and therefore were included in the
on/off solenoid category.
The cases in Table 4-A-10 with either no control valve or
a pneumatic valve are matched with both proposed systems to
allow cost comparison. The proposed modulating solenoid valve
is assumed to be the logical replacement for the two cases
already incorporating solenoid valves. This assumption
incorporates the existing development inertia, and the improved
purge control inherent with computerized systems.
The costs shown in Table 4-A-10 are based on
over-the-phone estimates of the vendor sale price from
Borg-Warner Automotive. Due to the large number of similar
systems amongst engine families, a large volume production (1
million units) was assumed. The $0 80 pneumatic valve price
-------
4-A -13
Table 4-A-10
Proposed Purge Control Systems
Case
Existing System
(Cost)*
Proposed System (Cost)*
Add'1 System
l.a
No Control Valve
Proportional Valve (4,1.60)
Delay Valve** $1.00)
Subtotal
SO
Subtotal (2.60)
$2 . 60
l.b
No Control Valve
Modulatmq Solenoid ($2.30)
Subtotal
$0
Subtotal $2.30
$2 . 30
2 . a
Pneu. Valve
< SO.80 }
Proportional Valve ($l.o0)
Delay Valve** ($1.00)
Subtotal
SO. 80
Subtotal $2.6 0
$1.30
2 .b
Pneu. Valve
($0.30)
Modulatmq Solenoid ($2.30)
Subtotal
SO.SO
Subtotal $2.30
SI . 50
3 .
Solenoid
($2.00)
Modulatmq Solenoid ($2.30)
Subtotal
$2 .00
Subtotal $2.30
SO. 30
4 .
Pneu. Valve
Solenoid
($0.80)
($2.00)
Modulatmq Solenoid ($2.30)
Subtotal
$2 .80
Subtotal $2.30
-$0.50***
* Valve costs are for high production volumes (1 million).
Values based on phone conversations with engineers at
Borg-Warner Automotive, August 1-4, 1986.
** Delay valve retards the vacuum signal that opens the
proportional valve.
*** The automotive industry automatically takes advantage of
cost saving changes on their own. Therefore, rather than
incorporating in a cost credit for a modification that
will happen, the EPA will consider the additional system
cost for case 4 to be SO.
-------
4-A-14
was estimated by the EPA from the reported $1.35 price for an
independently constructed two-stage valve. The reduced price
should be more representative of the common construction of
integrating the body of the purge valve in with the canister
housing. The $.30 difference in price between existing
solenoid valves and the proposed frequency modulated solenoid
valves incorporates a dampening capability for pulse control.
The cost for a delay valve was expressed as being under $1.00,
but $1.00 was used to be conservative. The price for the
proportional valve was a Borg-Warner estimate for a valve that
was developed, but never put into production.
The additional system cost for the modulating solenoid
valve is $.30 less than the proposed proportional/delay valve
system. Due to this economic advantage, along with the
improved purge control available, this study will assume that
frequency modulated solenoid valves will become common
throughout the fleet for high volatility purge control.
Table 4-A-ll shows an estimate of the average fleet costs
for changing to the proposed purge control system. The
vehicles shown are categorized according to the cases presented
in Table 4-A-10. The fleet weighting is based on 1985 sales of
these Chrysler, General Motor, and Ford vehicles, as reported
in the 1986 Ward's Automotive Yearbook.[22] While these
vehicles only represent 54 percent of the 1985 gasoline LDV
fleet, any error associated with calculating the average fleet
cost will be small, considering a $2.30 range of options for
the cost of the proposed valve change. The average fleet
hardware cost for the proposed change to frequency modulated
solenoid valves for purge control was therefore estimated at
$0.96 per vehicle.
Research, Development and Testing Costs
Table 4-A-12 contains a summary of the research,
development and testing (RD&T) costs for a change in
certification fuel to an RVP of 11.5 psi. The various costs of
RD&T per engine family are broken down in Table 4-A-10 and
costs for each component were estimated using engineering
judgment. The total cost was calculated using the total number
of engine families from 1984 certification records. The total
cost was adjusted for 2 years of interest at 10 percent, and
then amortized five years. This amount was then used to
determine the cost per vehicle based upon projected 1989 sales.
To obtain the RD&T costs for a change in the certification
fuel to an RVP between 9.0 psi and 11.5 psi, the vehicle
acquisition costs were held constant and the salary and testing
costs were varied linearly. The RD&T costs for 8.0 and 8.5 RVP
-------
4-A-l5
Table 4-A-l1
Estimate of Fleet Cost for Changing Purge Control System
Manufacturer/ 1985 Total
Engine Size Case* Cost ($) Sales** Cost ($)
Chrysler
2.2(1 3 0.30 894,394 268,318.20
5.20. 4 0 168,238 0
GM
2.54 2.b 1.50 689,530 1,034,295.00
2.84 4 0 615,328 0
3.8)1 3 0.30 683,578 205,073.40
5.Oil 2 . b 1.50 1,111,823 1,667,734.50
Ford
1.99. 2.b 1.50 209,200 313,800 00
2.30. l.b 2.30 478,225 1,099,917.50
3.84 3 0.30 435,519 130,665.70
5.00 2.b 1.50 617,490 926,235-00
Total 5,903,325*** $5,646,039.30
Average Fleet Cost = $5,646,039.30/5,903,325 = $0.96/vehicle
* These cases refer to the case presented in Table 4-A-10.
** 1985 Sales obtained from the 1986 Ward Manual [Reference
22 of Chapter 4.]
*** This value represents 54 percent of the 10,947,224
gasoline light-duty vehicle sales for 1985.
-------
4-A-16
Table 4-A-12
2
Research, Development and Testing (RD&T) Costs Summary
( 1986 Dollars)
LDV/LDT: ($/family)
Vehicle Acquisition 17,000'
Engineer Salary (1 mo. @ $53K/yr) 4,400~
Technician Salary (2 mo. @ $37K/yr) 6,200
25 Tests (@ $650/test)' 16,250
43,850
HDV: {$/family)
Vehicle Acquisition 21,300z
Engineer Salary (1 mo. @ $53K/yr) 4,4002
Technician Salary (2 mo. @ $37K/yr) 6,200z
25 Tests (@ $2200/test)3 55,000
86,900
LDV LDT HDV
Cost/Family $43,850 $43,850 $86,900
Number of Families 1374 81"* 115
Total Cost $6,007,000 $3,552,000 $955,900
W/ 2 yrs int.(@ 10%) $7,269,000 $4,298,000 $1,157,000
Amort. (5 yrs @ 10%) $1,918,000/yr $1,134,000/yr $305,100/yr
Projected 1989 Sales 11,000,0005 3,640,000s 386,0001
RD&T Cost/Vehicle ($) 0.17 0.31 0.79
Inflation-adjusted values from EPA memo, "Light-Duty
Vehicle Certification Cost," from Daniel P. Hardin, March
13, 1975. [Reference 15 of Chapter 4.]
Estimated.
"Regulatory Impact Analysis, Oxides of Nitrogen Pollutant
Specific Study and Summary and Analysis of Comments,"
EPA, March 1985. [Reference 17 of Chapter 4.]
EPA's "Control of Air Pollution from New Motor Vehicles
and New Motor Vehicle Engines; Federal Certification Test
Results for 1984 Model Year."
Based on DRI "trendlong" projections from the Fall 1984
Long-Term Review, Data Resources Inc., 1984.
-------
4-A-17
fuels are assumed to equal the RD&T cost for 9.0 RVP fuel. A
summary of the RD&T costs for the different certification fuels
are listed in Table 4-5 through 4-7.
Certification Costs
Table 4-A-13 contains a summary of the certification costs
associated with a new certification fuel for LDVs and LDTs.
The costs per vehicle tested are summarised for both emissions
data and durability vehicles. The total costs are based on the
number of LDVs and LDTs certified in 1984. However, only those
engine families which are carried over from the previous year
are relevant, since those which are recertified anew would
experience certification costs with or without a new
certification fuel. Carryover is estimated to be 90 percent;
therefore, the total cost was reduced to 90 percent of its
original value. The reduced total was adjusted for 1 year of
interest at 10 percent, and then amortized for five years.
This amount was then used to determine the cost per vehicle
based upon 1989 sales projections. The certification cost for
LDVs and LDTs is therefore estimated to be $.7 1 and $.89,
respectively.
Heavy-Duty vehicles have a wider range of vehicle sizes
and applications than LDVs/LDTs, thereby making the
certification cost more complicated to estimate. Therefore,
the EPA has broken the certification cost into exhaust emission
and evaporative emission components. Exhaust emission
certification would be necessary due to the new requirement
that the canister be connected to the engine during testing,
and is estimated to be $200,000 per family.[17] This estimate
includes both durability assessment and the use of three
emission data engines per family, In 1986, HDV manufacturers
certified eight families using 23 emission data engines. Total
exhaust emission certification costs including durability
assessment are therefore estimated to be $1.6 million dollars
Evaporative emission certification would be necessary due
to the proposed test procedure changes. This is estimated to
be $31,000 per family ($620 per test * 50 tests per
family).[18] In 1986, HDV manufacturers certified 23
family/system combinations. Using these figures, total
evaporative emission certification costs are estimated to be
$713,000.
Table 4-A-14 shows a summary of these HDV certification
costs. Still assuming a one year opportunity cost at 10
percent interest and amortizing the result at 10 percent, the
yearly fleet cost will be $671,200. Using a projected sale of
386,000 heavy-duty gasoline vehicles m 1989 [18], the average
certification cost is $1.74 per vehicle.
-------
4-A-18
Table 4-A-13
LDV/LDT Certification Cost Summary (1986 Dollars)
Emission Data Vehicle Costs:* ($/Vehicle Tested)
Vehicle Acquisition
17,000
Mileage and Maintenance
11,060
(@ $2.77/mile, 4000 miles)
Testing Cost
1,300
(2 Tests/Vehicle, $650/test)
29,360
Durability Vehicle Costs:* ($/Vehicle Tested)
Vehicle Acquisition
17,000
Mileage and Maintenance
164,500
(@ $3.29/mile, 50,000 miles)
Testing Cost
8 ,450
(13 Tests/Vehicle, $650/test)
189,950
Total Vehicles Tested and Costs: ($)
LDV
LDT
Emission Data (307 LDV, 133 LDT)* * 9,013,520
3,904,880
Durability (109 LDV, 45 LDT)** 20,704,550
8,547,750
Total Cost 29,718,070
12,452,630
90% Carryover 26,746,000
11,207,000
W/ 1 yr of interest (@ 10%) 29,421,000
12,328,000
Amortization (5 yrs. @ 10%) 7,761,000
3,252,000
1989 Sales*** 11,000,000
3 , 640 , 000
Certification Cost/Vehicle ($) 0.71
0.89
* Inf1 at ion-ad]usted values from EPA memo.
"Light Duty
Vehicle Certification Cost," from Daniel P.
Hardin, March
13, 1975. [Reference 15 of Chapter 4.]
** EPA's "Control of Air Pollution from New
Motor Vehicles
and New Motor Vehicle Engines; Federal Certification Test
Results for 1984 Model Year."
*** Based on DRI "trendlong" projections from the Fall 1984
Long-Term Review, Data Resources Inc., 1984.
-------
4-A-19
Table 4-A-14
HDV Certification Cost Summary (1986 Dollars)*
Category
Exhaust Emission Component*
($200,000/family){8 exhaust families)
Cost ($)
1,600,000
Evaporative Emission Component**
($620/test)(50 test/family)(23 evap families)
713,000
Total Cost
2,313,000
W/ 1 yr of interest (@ 10%)
2,544,000
Amortization (5 years @ 10%)
671,200
Projected 1989 sales** = 386,000
Ave. vehicle cost = $671,200/386,000 = $1.74
"Regulatory Impact Analysis, Oxides of Nitrogen Pollutant
Specific Study and Summary and Analysis of Comments," EPA
March 1985. [Reference 17 of Chapter 4.]
Values from EPA report, "Feasibility, Cost and Cost
Effectiveness of Onboard Vapor Control for Heavy-Duty
Gasoline Vehicles".
-------
Chapter 5
Technological Feasibility and Cost of In-Use
Gasoline Volatility Control
As described in previous Chapters, one of the two possible
methods of eliminating the current evaporative emission excess
is to control the volatility of commercial gasoline Estimates
of refinery costs and fuel economy benefits of reducing m-use
gasoline volatility are presented in this chapter. As
described in Chapter 2, the two fuel parameters most relevant
to evaporative emissions are RVP and %16q. In-use control of
both of these parameters is being considered. However, the
great majority of the refinery modeling has focused on the
control of RVP, as the effect of this parameter on evaporative
emissions is most well known.
This chapter begins with a general discussion of gasoline
volatility and the types of HC compounds which most affect it
(Section I). The next section (Section II) presents a general
description of the main sources of information for che cost of
m-use gasoline RVP control and %i60 control, two studies
conducted by Bonner and Moore Management Science for EPA.[1,2]
These studies use Bonner and Moore's proprietary Refinery and
Petrochemical Modeling System (RPMS), which is a linear
programming (LP) computer model Section III presents the
results of the Bonner and Moore studies, primarily the refinery
cost of reducing gasoline RVP one to two psi below ASTM limits
under various situations, but also includes the cost of
reducing %i60 to 25 or 30 percent. This section also
includes a discussion of revisions that were made to the costs
in light of comments received from external review of the
reports. Section IV examines the effect of RVP control (and
the likely butane excess generated by it) on the wider butane
market. Section V examines the effect of RVP control on the
energy content of gasoline and estimates the resultant effect
on vehicular fuel economy as well as the additional economic
benefit of recovering or preventing evaporative emissions (via
both fuel- and vehicle-related controls) . Section VI discusses
how control of fuel volatility affects the dnveability of
vehicles. Section VII examines the cost of enforcing in-use
control regulations.
It should be noted that the refinery costs used m this
Regulatory Impact Analysis are a revised edition of those used
in EPA's November 1985 study of gasoline volatility. EPA has
contracted Bonner and Moore to do additional modeling to ensure
that the adjustments made here fully reflect the complexities
of U.S. refineries. When completed, the report will be placed
in public docket A-85-21 at EPA Headquarters for review
-------
5-2
I. Refinery Control of Gasoline Volatility
Oil refineries currently refine crude oil to supply
several petroleum products, one of which is gasoline for motor
vehicle consumption. The volatility of the gasoline produced
depends on the properties of the crude oil and the natural gas
liquids (NGLs), which include butanes and natural gasoline,
used in the refining process and on the actual refining
processes themselves. There are ASTM specifications for
gasoline volatility characteristics which serve as guidelines
for the refineries to follow. These ASTM gasoline volatility
specifications and in-use volatility trends were already
described in Chapter 2, Section II.
As already mentioned, RVP and %iSo are the two most
important gasoline volatility properties affecting evaporative
HC emissions. The major HC compounds contributing to high RVP
are n-butane and iso-butane, although n-pentane and iso-pentane
also affect RVP. The RVP blending values for these parafins
are discussed in Section II of Chapter 2. Generally, the
easiest way to reduce gasoline RVP is to reduce the blending of
straight butane into gasoline. The next easiest approach is to
remove butane presently contained in other gasoline stocks.
This requires modification of existing separation facilities
and/or the installation of new facilities to remove contained
butanes. In addition, butane can be converted, via processes
such as alkylation, in higher boiling compounds. As a further
step, extreme RVP limits could require removal of some Cs
components which, in most cases, would require new separation
facilities The approach that is actually used depends on the
economics involved, which are complex and highly interactive.
Both butanes and pentanes strongly affect %i6o, but so
do many other gasoline components, and the specific
relationship between each of these components and %iso is
beyond the scope of this study. Suffice it to say that the
more short-chain hydrocarbons (with four to five carbon atoms)
in gasoline, the higher the RVP and %iso.
Butane has a high-octane quality. Thus, RVP control which
reduced the use of butane would also require additional
refinery processing to compensate for the octane quality which
the butane would otherwise provide.
The use of alcohols in gasoline complicates matters, since
both methanol and ethanol affect RVP as well as other
volatility parameters and octane. Adding 5%/2.5% methanol/TBA
to ll.O-psi RVP gasoline increases RVP by 2.2 psi. Adding 10
percent ethanol to gasoline increases RVP up to 1.0 psi. The
RVP blending values of methanol and ethanol were discussed in
Section II of Chapter 2. Thus, their addition to gasoline must
be accompanied by an even greater reduction in butane/pentane
-------
5-3
content if RVP is to be held constant, though the alcohols'
high-octane more than compensates for that of the lost butane
in this case.
II. The Bonner and Moore Studies
Bonner and Moore Management Science conducted two studies
on the refinery cost impact of vapor pressure control under a
number of subcontracts with Southwest Research Institute (SwRl)
and Jack Faucett Associates
-------
5-4
were 65 percent unleaded regular, 18 percent unleaded premium,
and 17 percent leaded regular. Leaded gasoline is allowed to
contain only 0.1 gram per leaded gallon of lead, so the effects
of EPA's lead phasedown program are fully factored in.
There are also some additional assumptions that had to be
made in modeling the alcohol-fuel blends. First, gasoline
production was set at the same level used in the alcohol-free
cases. No fuel economy changes that might apply to
alcohol-containing fuels were taken into account. Furthermore,
100 percent of each grade of gasoline, unleaded premium,
unleaded regular and leaded regular, was assumed to contain the
specified concentration of alcohol. While this is not
realistic, it was the most convenient way to include alcohols
in the RPMS and the results appear to be reasonable. (As will
be discussed below, this same approach was used to model
ethanol blends and the results were clearly unreasonable).
Also, a lower RVP case study of three psi below the baseline
vapor pressure was evaluated for the methanol blend because of
the possibility that commingling of the blend with alcohol-free
gasoline would increase RVP beyond that of either fuel due to
the azeotropic behavior of alcohol and hydrocarbon mixtures
(discussed further in Section II of Chapter 2). The 1-psi
further reduction in vapor pressure would tend to partially
offset any increase in vapor pressure due to such commingling
effects, resulting in evaporative emissions equal to those for
alcohol-free fuels.
The estimated costs in all cases for each PADD were
developed by the RPMS using a single "super refinery" to
represent all of the refining capabilities of that PADD (i.e.,
the average refinery). This super refinery was required to
produce all of the gasoline projected to be produced by all of
the individual refineries in that PADD in the timeframe of the
study, which was 1990. Individual refineries would be expected
to experience costs both above and below that estimated by the
RPMS, but on average, the actual costs should be close to that
projected by the model. The complexities involved with
modeling individual refineries make such an approach
economically infeasible and make the use of a single refinery a
necessary limitation of this study.
The gasoline costs estimated by the RPMS are, by design,
incremental in nature and do not attempt to represent the full
cost of refining gasoline. This avoids a number of complex
issues associated with valuing capital equipment already m
place. As the desired output is the effect of RVP on refining
costs, a difference between an uncontrolled and controlled
scenario, this is fully satisfactory for this study.
To accomplish this, a base 1990 case is run to determine
optimal process requirements and refinery costs associated with
producing the 1990 product slate, considering process
-------
5-5
capacities known to be available in 1984, and, thus, not
requiring capital investment. The controlled case is run in an
analogous fashion (i.e., a fresh optimization from 1984
capacities), only with a lower RVP or %i6o product.
Conceptually, this approach assumes that investment occurring
between now and 1990 in the uncontrolled case can be redirected
toward more productive use, if economically desirable, in the
controlled case. This may or may not be the case, depending on
the timing of any volatility controls. Information from the
past model runs however, show that very little of the
investment occurring in the base case does not also occur in
the controlled case. Thus, little redirection of 1984-1990
investment appears to be occurring and the effect of allowing
this in the modeling runs appears to be small
For those readers investigating such details, it should be
noted that the RPMS runs tend to project sizable capital
investments between 1984 and 1990 for the base cases even
though the refinery industry as a whole is expected to invest
little for gasoline capacity aside from environmental control
(i.e., lead phasedown).[2] This occurs because the current
capacity of many peripheral processes (e g., cooling towers) is
not known and was presumed to be zero in 1984 for modeling
purposes. Thus, the required 1990 base capacity for these
processes is considered to be entirely incremental, though m
all likelihood, the vast majority of it is currently in place.
These sizable capital investments have no direct effect on the
estimated RVP control costs, nor the estimated capital
investment required for RVP control, since these are
incremental costs involving the subtraction of base case costs
from the controlled case costs, both of which contain these
costs. It simply means that the capital investment shown for
either the base or controlled cases cannot be used to estimate
the total capital investment required by the refining industry
between 1984 and 1990. However, as the current capacities of
some of these peripheral processes may be in excess of that
needed in 1990, the model may be overestimating the additional
capacity needed for RVP control. The degree to which this may
be occurring is unknown and is not easily estimated.
The original RPMS runs also assume that all capital
investment is amortized over year-round production. This is
appropriate in the base case, since most of this equipment is
of the kind that is used year round. However, equipment
purchased expressly for RVP control might only be used during
the specified five-month RVP control season and are thus only
amortized over five months. The supplemental report by Bonner
and Moore showed the facilities purchased to control RVP had
little value m the non-control period, and should be amortized
only over the control season. Amortizing equipment only over 5
months raises the refining costs and this increase will be
discussed in the results section.
-------
5-6
A no-investment scenario was modeled primarily to simulate
the situation where the leadtime granted to refiners prior to
RVP controls was insufficient to design and build new capital,
but it applies as well to the extreme situation in which the
model itself avoids all incremental capital investment because
capital investment associated with RVP control has no value
outside of the RVP control period. While in theory the cases
were to be strictly no new investment, this stipulation had to
be relaxed in practice, again due to the unknown current
capacities of many peripheral processes. It did not seem
reasonable to limit such capacities to those required in the
base case, because historic gasoline production has been much
higher than projected 1990 levels and much excess capacity
could exist. At the same time, the degree of this excess is
unknown. Thus, the no-investment costs may be underestimated,
since the benefit of some capital investment may be included.
The extent to which this is true is not known and is not easily
estimated. Thus, these costs represent the best estimates
available under such conditions.
The degree of RVP control that can be achieved in a
no-investment scenario can theoretically be determined by the
modelling results. In the Bonner and Moore modelling, however,
this is not the case due to the lack of knowledge of the
capacities of peripheral processes as discussed above. This
issue of lead-time, how low RVP can be reduced without capital
investment, was examined by Sobotka and Company.[3] The
results of the analysis will be discussed in the next section.
Another important aspect of modeling refinery RVP control
is the treatment of natural gas liquid (primarily butane)
supply and demand. Currently the butane market varies
dramatically between summer and winter. With gasoline RVP
levels 2 psi higher in the winter (corresponding to a butane
composition increase of 4 percent by volume), butane supplies
tend to be short and prices high. In the summer, the opposite
is true. With RVP control, even more butane will be available
in the summer and prices could decrease further. This
potential price decrease is dependent on the entire
butane/petrochemical market and not just on petroleum refinery
operations. Thus, a model such as RPMS, which only models
petroleum refining, cannot project the price drop. In fact,
the price of butane must be input to the model.* However, two
types of situations were modeled using RPMS to simulate the
effect of the potential price drop.
Current modelling efforts will model petroleum refining
and ethylene manufacture. A separate analysis will
determine butane market impacts by the volumes of butane
not consumed by these sectors.
-------
5-7
The first situation assumed that butane could be purchased
or not purchased at its current price, which varies with PADD,
depending on its economic usefulness. This situation is
referred to as the "open" NGL purchase scenario. Butane
availability was limited to that purchased by refineries in
recent times. The second situation forced the refinery to
purchase all of the available butane at the current summer
price. This situation is referred to as the "fixed" NGL
purchase scenario. The first situation was intended to place a
lower limit on RVP control costs by allowing refineries to sell
any excess butane generated by RVP control at current market
prices In reality, lower butane prices would probably occur,
thereby reducing the profitability of doing this and also
increasing the cost of RVP control. The second situation was
designed to place an upper limit on RVP control costs by
requiring the refinery segment of the butane market to use all
of the excess butane at current market prices. In reality,
prices would drop and other segments at the market would
utilize at least part of the butane excess and result m lower
RVP control costs. In addition, analysis of the impact of RVP
control on the entire butane market was performed Its results
are described in Section IV below.
Until more detailed information is available, the midpoint
between costs under the "fixed" and "open" NGL purchase
scenarios will be used for the cost of RVP control. Further
discussion on the decision to choose the midpoint between these
two scenarios is presented in Section IV below.
The final aspect of the RPMS deserving discussion here is
the way the model simulates ASTM gasoline specifications. RVP
levels for the uncontrolled RVP base case were set at maximum
ASTM D-439 RVP specifications. PADD average maximum RVPs were
estimated by volumetncally weighting the RVPs of all the
gasoline produced by refineries in a particular PADD. National
average maximum RVPs were estimated by volumetrically weighting
the PADD-specific RVPs by the volume of gasoline produced in
each PADD. RVP of PADDs 4 and 5 was assumed to be the average
of the RVPs for PADDs 2 and 3. The PADD-specif ic RVPs and
national RVPs are presented in Table 5-1. The volumetric
production weighting factors are also presented at the bottom
of Table 5-2. Current (1986) national average gasoline RVP, as
determined from the MVMA fuels survey, is 11.06 psi, which is
near the maximum ASTM RVP specification of 11.27 psi. It is
assumed that, by 1990, the national average RVP will equal the
maximum ASTM RVP specification. This was discussed in detail
Section II of in Chapter 2.
Besides RVP, ASTM addresses fuel volatility by specifying
minimum and maximum temperatures at which specified fractions
of the fuel are evaporated via distillation (i.e., TJ0 and
Tso) as discussed in Section IV of Chapter 2. Such
-------
5-8
Table 5-1
National and PADD-Speci f ic RVPs and aoiMiS
Resulting from Bonner & Moore RPMS Cases
Level of RVP Control
NGL Purchase
PADD Scenario
Baseline (B) 1 psi Reduction 2 psi Reduction
RVP °0IS„ RVP °o,bH RVP "o, ,H
I Open
I Fixed
II Open
II Flxed
III Open
III Fixed
Nat. Avg Open
Nat. Avg Fixed
11.5 '',4.07 8
11.5 34.978
11.46 33.028
11.46 35.0
11.12 33.144
11.12 33.144
11.27 33.27
11.27 33.92
10.5 33.7
10.5* 33.778*
10.46 32.487
10.4-5 33.284
10.12 30.753
10.12* 31.184*
10.27 31.59
10.27* 32.11*
q 5 32.1
9.5 32.0
9.46 31.808
9.46 30.73Q
9.12 28.325
9.12 28.28
0.27 2 9.32
9.27 29.42
Found by interpolation between (B) and (B-2) cases for the 11h condition
RVP = midpoint between baseline and the 2-psi control case
°<>is8 = (baseline) - 0.403 [(baseline) - (baseline-2)]
-------
5-9
Table 5-2
Cost of RVP Control
($ per Barrel of Gasoline)
2
PADD
RVP 4+5 Total U. S
Reduction, psi 1 2 3 (ex. CA) 1 (ex CA)
Alcohol-Free Gasoline
With Investment, Open Butane Purchases
1 0.184 0.365 0.211 0 288 0.259
2 0.547 0.748 0.501 0.625 0 587
With Investment, Fixed Butane Purchases
1 0.1893 0.396 0.2123 0.304 0 271
2 0.561 0.857 0.504 0.681 0.626
With Investment, Average of Fixed and Open Butane Purchases
1 0.186 0.380 0.212 0.296 0.265
2 0.554 0 802 0.502 0 653 0 606
No Investment, Fixed Butane Purchases
1 0 .2144 0 . 5014 0.3004 0 401 0 358
2 0 641 1.084 0.713 0.899 0.826
No Investment, Estimated Average of Fixed and Open Butane
Purchase
1 0.211 0.480 0.300 0.390 0.350s
2 0.633s 1.014s 0.710s 0.862s 0.800s
2.5% MeOH +2.5% TBA Blend, with Investment, Open Butane Purchases
1 0.369s 0 .4 0 76 0.314 0.360 0.35
2 0.815 0.898 0.692 0.795 0.77
3 1.3707 1.5097 1.163 1.336 1.29
5.0% MeOH +2.5% TBA Blend, with Investment, Open Butane Purchases
1 0,195s 0.387" 0.22 48 0.298 0.27
2 0.581s 0.794s 0.532 0.676 0.62
PADD-Specific Fraction of Total Gasoline Volume (%)
8.2 23.3 48.0 20.5 100.00
Footnotes on following page.
-------
5-10
Table 5-2 (cont'd)
1 Estimated as average of cost for PADDs 2 and 3.
2 Total U.S. (ex. CA) costs were estimated by volumetrically
weighting of PADD-specific costs for PADDs 1, 2, 3, and 4 and
5 (excluding California) using PADD-specific gasoline
production.
3 Actual case not run, cost estimated from open butane purchase
for l-psi RVP reduction using the percentage increase for
2-psia reduction costs between open and fixed NGL purchases.
4 Actual case not run, cost estimated from fixed butane purchase
costs with investment for l-psi RVP reduction using the
percentage increase between 2-psia reduction costs with and
without investment.
5 These costs were estimated from the No-Investment costs for
fixed butane purchases presented in the previous row using the
ratio of midpoint (open-flxed)/2 vs. fixed butane purchases
costs for the cases with investment.
6 Actual case not run, cost estimated by applying ratio of l-psi
reduction cost to 2-psi reduction cost for PADD 3 control to
the 2-psi reduction cost for PADD in question.
7 Actual case not run, cost estimated by applying ratio of 3-psi
reduction cost to 2-psi reduction cost for PADD 3 control to
the 2-psi reduction cost for PADD in question.
8 Actual case not run, cost estimated using the ratio of
corresponding costs from open butane purchase cases for
alcohol-free gasoline with investment applied to PADD 3 cost
for 5%/2.5% MeOH/TBA from B&M.
-------
5-11
specifications do not lend themselves to linear programming
since temperatures at which certain fuel volumes are evaporated
cannot be easily manipulated when two fuel streams are merged.
It is much easier to work in the reverse mode, the percent fuel
evaporated at specific temperatures (i.e., %iGo), since these
can be volumetrically averaged when two streams are blended
together. This is the mode in which the RPMS works. It is
possible to convert from one mode to the other, but only
approximately. Thus, at the present time, it is not clear
precisely how the RPMS limits (other than RVP) approximate
those of ASTM. However, the RPMS requirements specified by B&M
for maximum percent of gasoline evaporated at a given
temperature appear to be within the ASTM requirements
specifying minimum and maximum temperatures corresponding to a
given percent of gasoline evaporated. ASTM D-439 requirements
for minimum and maximum temperatures corresponding to a given
percent of gasoline evaporated are not tight specifications,
and current fuels are not being restricted by these
requirements.
The point of major concern on the gasoline distillation
curve is the %i6o, as discussed in Section II of Chapter 2.
Bonner and Moore limited the maximum %iGo to 35 percent for
the RPMS cases in their original study. This resulted in the
values of %iGo presented m Table 5-1, for the specific PADDs
and for the national on a whole. Current (1986) national
average gasoline volatility properties are 11.06-psi RVP and
32.6 %iso. These values are those for volumetrically
weighting all gasoline grades (65 percent unleaded regular, 18
percent unleaded premium, and 17 percent leaded regular), but
are also the same as those for unleaded regular gasoline as
presented in Table 2-6 of Chapter 2. The average %i6o of
current gasolines is approaching 35 percent, indicating that
the B&M restriction of 35 %i6o is not too lax. Therefore,
the B&M costs for RVP control with %1G0 restricted to 35
percent appear reasonable from this point of view.
B&M modeled some additional cases to evaluate the cost of
further controlling %16o, both independently and in addition
to RVP control. The cases evaluated and the results of running
these cases with the RPMS are presented in the following
sect ion.
III. Refinery Costs of Volatility Control
A. RVP Control
The projected costs of 1- and 2-psi RVP reductions below
the maximum ASTM-specifled RVP for various gasoline types and
other scenarios are shown in Table 5-2. Before discussing the
results of Table 5-2, it should be noted that the %(6o of
these fuels are indirectly reduced through this RVP control,
because removing compounds which contribute to high RVP also
-------
5-12
lowers %ico, to a certain extent. The reductions in %iSo
associated with these 1- and 2-psi RVP reductions for the
"open" NGL purchase scenario are 1.7 percent and 3.5 percent,
respectively, as indicated by the national average levels for
%160 presented in Table 5-1. The %iso is reduced 1.8 and
4.5 percent for the corresponding "fixed" NGL purchase scenario.
The cost of controlling the RVP of alcohol-free gasoline
is of primary concern, since it represents nearly 93 percent of
all gasoline sold in the U.S. (with gasohol making up the other
7 percent of the nations gasoline sales.) As can be seen in
Table 5-2, the nationwide-average cost of reducing RVP by 1 psi
is 0.62-0.95 cents per gallon ($0,259-0.358 per barrel), while
the cost of a 2-psi reduction is greater than two times that,
or 1.40-1.97 cents per gallon ($0,587-0.826 per barrel),
depending on how capital investment and the butane market are
treated. These values are the results of the modeling being
performed with a crude oil price of approximately $30 per
barrel.
The least expensive RVP control scenarios are those m
which the refineries have the ability to invest in capital
equipment to optimize their refinery processes. This situation
occurs when the controls are to be in place long enough to
justify the capital investment and after refiners have had
sufficient time to design and implement new equipment. This
definitely applies to the long-term control scenario (i.e.,
2010) and may apply to short-term controls if the above two
conditions are met.
In addition, two extreme scenarios were evaluated with
capital investment permitted, one with "open" NGL purchases and
one with "fixed" NGL purchases, as described earlier. RVP
control for the "open" NGL purchase scenario was slightly less
expensive than for the "fixed" case ($0,587 vs. $0,626 per
barrel for a 2-psi RVP reduction, respectively).
For the no investment situation, only fixed NGL cases were
modeled. Open NGL costs were estimated from the runs "with
investment" to determine the best estimate midpoint
"no-investment" cost, as shown in Table 5-2. These "no
investment" costs simulate short-term control if refineries are
not given sufficient time to invest in capital prior to control
or if the period of control is too short to justify capital
investment (generally thought to be 2-4 years). Although the
costs for controlling RVP in a no-investment scenario are shown
for both a l-psi and 2-psi reduction, recent analysis found
only a l-psi reduction is feasible in the short run.[3] In the
EPA study performed by Sobotka and Company, a one-psi reduction
in RVP was found to be the upper limit of volatility reductions
due to current gasoline production capabilities in the U.S.
Lead phasedown regulations and higher gasoline demand due too
low crude oil prices prohibit the refining industry from
-------
5-13
reducing RVP by more than 1-psi without further capital
investment. The 1-psi reduction was found to be feasible for
all the ASTM classes in the short term.
The cost of controlling the RVP of alcohol-blends follows
the same general PADD-to-PADD trend as that for alcohol-free
gasoline. Overall, the cost of vapor pressure control is
greater for gasoline containing the 2.5/2.5 percent
methanol/TBA blend, than for alcohol-free gasoline. The cost
of a 3-psi reduction for PADD 3 (the only PADD evaluated) is
increased 68 percent over that of ]ust a 2-psi reduction. This
increase may differ slightly between PADDs, but was assumed to
be constant m Table 5-2.
The cost of reducing RVP for the 5.0/2.5 percent
methanol/TBA blend, in contrast, is less than that for the
2.5/2.5 percent methanol/TBA blend and is 6-7 percent greater
than that for alcohol-free gasoline, on average. Bonner and
Moore credits the smaller cost to the non-linear vapor pressure
blending behavior of the additional methanol. The additional
2.5 percent of methanol incurs a small additional vapor
pressure penalty but contributes a larger octane benefit. As
this is expected to be more popular than the 2.5/2.5 percent
methanol/TBA blend in the future, and its RVP control costs are
so close to that for alcohol-free gasoline, only the cost for
alcohol-free gasoline will be used hereafter.
As indicated above, attempts were made to model 10 percent
ethanol blends, in the same manner as methanol blends.
However, in all cases, the base (uncontrolled) RVP was already
very near or below the 2-psi reduction level. This was an
artificial result of restricting %lSo to 35 percent. This is
unrealistic for current ethanol blends, as ethanol dramatically
affects %iso and current levels of %iso for ethanol blends
at 11.5-psi RVP are around 42 percent. Thus, the forced
lowering of %i60 likely forced most of the butane out of the
fuel and lowered RVP dramatically prior to control. This
situation is being corrected in additional modeling currently
being conducted by B&M. Results are not available yet, but
will be incorporated in the study of controlling evaporative
emissions as soon as the RPMS case studies are complete.
Thus, little can presently be said based on actual
refinery modelling about the cost of controlling the RVP of
ethanol blends. Qualitatively, ethanol has a smaller RVP
effect and greater octane effect than methanol/TBA mixture.
Thus, one would expect its presence to impact RVP control costs
less than the methanol/TBA mixture when %i60 is not
controlled. Since the methanol/TBA blend increased cost by
approximately 7 percent over control cost for alcohol-free
gasoline, any additional cost to control ethanol blends is
assumed to be negligible. This holds for the case discussed in
-------
5-14
Chapter l where alcohol fuels are given a one-psi allowance
over alcohol-free gasoline. If alcohol blends are required to
meet the same RVP specifications of alcohol-free gasoline, then
the control cost is equal to the cost of controlling the RVP of
alcohol-free gasoline plus the cost to control the next one-psi
increment. This results from the fact that ethanol will
contribute one-psi RVP to the base gasoline. This additional
cost to control RVP of ethanol blends will be quantified
below. The impact on the ethanol blending industry of
controlling the quality of the finished blend (i.e.,
eliminating splash blending and requiring coordinated blending)
will be discussed in Chapter 7.
The costs in Table 5-2 were used in EPA's 1985 study of
gasoline volatility and hence have been extensively reviewed.
Modeling efforts are currently underway to fully address the
comments made to the original study. The comments, which will
have the greatest impact on the costs, can be incorporated into
the estimates of Table 5-2. The details of the revisions are
described below.
The most important revision to make to the cost is to
reflect the drop in crude oil prices which occurred after the
study was published. The 1985 Bonner and Moore costs are based
on crude oil prices of approximately $30 per barrel. The
revised cost estimates will be developed at three crude oil
price scenarios, $15, $20, and $25 per barrel with $20 per
barrel being the base case assumption. The revisions to the
Bonner and Moore costs can be made by modifying the costs of
the swing crude purchases in each PADD. The difference in the
assumed crude oil price and the actual price in the model was
determined and then multiplied by the swing crude purchases in
the RVP control cases. This savings in crude purchases was
subtracted from the total cost in each case modeled by Bonner
and Moore. For the first psi reduction in RVP, the nationwide
costs for the $25, $20, and $15 per barrel cases were 16, 32,
and 46 percent less than the original costs, respectively.
This methodology for changing costs due to crude oil
prices is considered to be the conservative approach. If the
computer model was reoptimized for the lower crude cost cases,
the model would opt for raw material purchases over investment
to a greater extent and possibly further reduce RVP control
costs. It is likely, however, that as crude cost decreases
consumer demand would increase and the refinery would lose some
flexibility. These issues are currently being evaluated by
Bonner and Moore and will be incorporated into the final RIA
when completed.
The second revision is an adjustment for the lower
purchase price of butane in the control cases. In the Bonner
and Moore study the butane purchase price was held constant at
approximately $23.00 per barrel. As will be discussed in the
-------
5-15
next section, as RVP is controlled and butane backed out of
gasoline, the price of butane will drop until it reaches a
value at which it is competitive as a petrochemical feedstock.
This price was calculated at approximately $20.00 per barrel.
This lower butane price will affect the price of all butane on
the market (not only those backed out of gasoline) since, as
the Gas Proccessors Association noted in their comments on
EPA's gasoline volatility study, a two-tier price structure for
butane could not exist.[5] This lower butane price in the
control cases is a savings to the refining sector over the
higher price in the control cases of the Bonner and Moore study.
The above-mentioned figures were derived for a $30 per
barrel crude oil situation. The price of butane and hence the
increment was assumed to change proportionately with crude oil
price. These savings were applied to each barrel of butane
purchased in the control cases. The effect of the lower value
of butane ranged from no change in the PADD 1 cost to a $0,046
per barrel decrease in cost m PADD 3. The effect of the lower
value of butane was only calculated for the open butane
purchase scenario, since the fixed scenario represents the case
where there is no impact on the NGL industry (i.e., refiners
continue to purchase the same amount of butane at the high
price).
The original Bonner and Moore costs assumed that all
capital investment be amortized over year-round productions.
This is appropriate m the base case, since most of this
equipment is of the kind that is used year round. However,
equipment purchased expressly for RVP control in the control
cases will only be used during the specified five-month RVP
control season and, thus should only be amortized over five
months. The supplemental report by Bonner and Moore showed the
facilities purchased to control RVP had little value in the
non-control period, and should be amortized only over the
control season.[2] The effect of a 3-month period of recovery
of control-facilities is an increase in the control costs of 9
percent. For a 5-month control period the cost increase was
assumed to be 7 percent. The effect of the crude oil price,
the lower value of butane, and the shorter control period on
the refinery costs (average of open and fixed) is shown in
Table 5-3.
The values of Table 5-3 have been extrapolated to the
third psi RVP reduction step using the data availability in
Table 5-3. The incremental cost of reducing RVP from the one
psi reduction to the second psi reduction was compared to the
cost of reducing the RVP by the first psi. For the
alcohol-free gasoline, the second increment of control
increased over the first increment by an average of 30
percent. It was assumed that the third increment of control
would increase by a similar percentage over the second
increment. This assumption was confirmed by the results of the
-------
5-16
Table 5-3
Revised Cost of RVP Control
($ per Barrel of Gasoline)
RVP Crude Oil Price ($/bbl)
Reduction, psi 30 25 20 15
1 0.278 0.228 0.181 (.232)* 0 137
2 0.636 0.533 0.443 0.354
3 1.101 0.930 0.784 0.636
No investment.
-------
5-17
2.5 percent methanol, 2.5 percent TBA case in Table 5-3, the
only case in which a 3 psi reduction was investigated. In that
case the third increment increased over the second increment by
a percentage that was similar to increase of the second
increment to the first increment. Therefore, an increase in
the second increment of control of 30 percent was applied to
the revised cost of Table 5-3 and then added to the second psi
control step to determine the cost of control for the third
psi. This assumption was also confirmed by the results of a
study by the Bonner and Moore which was prepared for the
California Resources Board.[6] This study was particularly
interesting since it investigated RVP control cost from a
baseline of 9.0 psi. The cost results are not as relevant
today, since the study was completed in 1983, however, the
change in the increment of cost is relevant. The study found
the cost of reducing RVP from 8 psi to 7 psi to only be 16
percent greater than the cost of controlling RVP from 9 psi to
8 psi. Additional modeling efforts are investigating the cost
to control RVP by 3 psi as well as cost of controlling RVP in
California. The results will be incorporated into this
analysis when completed.
As discussed above the cost to control ethanol blends is
dependent on the control strategy chosen. If blends are given
a l-psi allowance then cost to control the blends will be the
same as non-alcohol gasoline. If blends must meet the same RVP
specification as non-alcohol gasoline, then the base gasoline
must be l-psi lower in RVP. As indicated in Table 5-3, the
extra increment of control on alcohol blends would increase
control cost by $0.18-$0.34 per barrel, depending on the level
of RVP control.
B. Cost of Volatility Control of %i60 of In-Use Fuels
In addition to the control of in-use fuel RVP, the
mid-range volatility, specifically the %i6o, has received
attention as another possible controllable parameter which can
lead to reductions in evaporative emissions. The effect of
mid-range volatility on evaporative emissions was presented in
Chapter 2. This section is intended to present the refinery
costs of controlling the %iGO of in-use fuels (gasohol,
methanol blends, and straight gasolines). The costs presented
in this section will be combined with the evaporative emission
reductions from Chapter 2 to predict the cost effectiveness of
reducing the %i6o of m-use fuels in Chapter 6.
For the original EPA volatility study, Bonner and Moore
Management Science, under sub-contract to Southwest Research
Institute, used a computer refinery model to determine the cost
of gasoline and alcohol blend volatility control.[1,2 ] Table
5-4 contains the Bonner and Moore results for reducing the
%ibo point under different fuel scenarios adjusted down by a
ratio of 0.74. This adjustment accounts for changes in the
-------
5-18
Table 5-4
Refinery Cost of °oi ui Reduction
Case
RVP
(ps i)
Change
in
M i) (i
Change in
°o \ t. i) of
Pool Base
Gasol1nes'
PADD 2
Cost
(S/Barrel
in pool)
Nationwide
Cost
($/barrel
in pool/
A",i (,n in
pool base
gasoline)
All alcohol-free
gasoline
All gasoline
containing 5.0/2.5
MeOK/TBA
9.4b 30.74-30 30.7-30.0
0.46 35-30
23.0-18.2
0.033
0.459
0.032
0 . C 6 1
7°o of pool is
gasohol (Only
gasohol %tho
reduced)
11.46 45-35
34.4-33.7
0 .017
0.018
7% of pool is
gasohol (All fuel
"oim) reduced)
35-25
33.7-23.6
0. 559
0 .040
* The bo blending value used for 10°o ethanol in gasoline was 220, for
7.5% MeOH/TBA (2:1) the blending value used was 175.
-------
5-19
price of crude oil and other changes in modelling approaches as
explained earlier in this chapter which have occurred since the
publication of the original Bonner and Moore reports.
In this %i6o analysis, the cost of mid-range volatility
control is based on the properties of the base gasolines in the
entire pool. This approach is taken because the base gasoline
is the component of alcohol/gasoline blends which will have to
be modified to counter the increased volatility effects of the
alcohol. The volatility properties of the alcohol are fixed
and cannot be adjusted.
The change in the l16o value of the base gasolines in
the entire pool is calculated based on the change in %,6o
value of the fuels affected and alcohol blending values. The
PADD2 costs are worst case results, and a nationwide (excluding
California) cost per barrel of gasoline in the pool is
calculated by a ratio of the nationwide (excluding California)
costs to PADD2 costs determined in the first Bonner and Moore
report for RVP control. The numbers used for the ratio are
found in Table 5-5 of this report and the average of the fixed
and open butane purchase ratios is used.
The final cost determined is the nationwide (excluding
California) cost per barrel of gasoline per percent change in
the pool %16o value. Figure 5-1 is a plot of the cost per
barrel per %i6o reduction versus the base %iSo value of the
pool. It is used to determine the final cost of %iSo
reduction.
It should be noted that there are two separate estimates
of costs over the %i6o range of 33.7 to 23.6, one cost based
on a scenario in which seven percent of the fuel supply is
gasohol, and the other cost based on an alcohol-free fuel
scenario. For the gasohol scenario, the change in the %i6o
of the pool base gasoline was 33.7 to 23.6, but for the
alcohol-free scenario, the actual change in the pool base
gasoline %i6o was only 30.74 to 30.0. The cost of %i6o
reduction per percent change in the %J6o for the alcohol-free
scenario is assumed to hold over the range from 33.7 to 23.6
and two lines of different slopes are drawn in Figure 5-1 to
represent the data. Over the %iG0 range from 23.6 to 18.2,
two lines of the same slope are drawn, representing the same
cost per percent change in pool %ieo, but starting at
different points.
These separate cost curves result in a range of the cost
estimate for several °$uo reduction fuel scenarios. The
final total cost of reducing the %IGo value under the various
fuel scenarios is listed in Table 5-6 and is based on annual
fuel consumption projections from MOBILE3. As previously
mentioned, the cost effectiveness of reducing the %iS0 of
m-use fuels will be presented in Chapter 6.
-------
5-20
Table 5-5
Cost of RVP Control*
($ per barrel of gasoline)
PADD
RVP
Reduction (psi)
1
2
3
4+5 Total US
(ex. CA) (ex. CA)
1
2
0.186 0.380 0.212
0.554 0.802 0.502
0 . 296
0 . 653
0 . 265
0 . 606
These are the costs for alcohol-free gasoline, with time
allowed for investment in new equipment, and the averages
of fixed and open butane purchases. Taken from Bonner &
Moore report, "Estimated Refining Cost Impact of Reduced
Gasoline Vapor Pressure."
-------
Figure 5-1
COST OF REDUCING BASE GAS POOL *160 F
15 17 13 21 1Z 25 27 29 il J 5
*1 6v F VALUE OF BASE GASOLINES IN POOL
Alo^nol Bl#na —. Straight Gasolin*
-------
5-22
Fuel
Gasohol
Methanol
All
Fuels
Table 5-6
Cost of Reducing %i6o of In-Use Fuels
Final
o
^160
45
35
45
35
35
30
25
Pool
34 4
33 7
35 0
34 . 6
33 3
28 . 2
23 . 2
Cost per
Barrel of
Fuel In
Pool ($)
*
0 .013
*
0 . 007
0.163-
0 . 204
0 . 333-
0 . 412
Total
1990
Cost1
(10 6$/yr)
*
23 . 7
•k
13 . 5
308 . 7-
387 . 7
634 . 0-
784 . 1
Total
2010
Cost2
(108 $/yr)
*
21 . 2
*
12 . 1
275 . 7-
346 . 4
566 . 5-
700 . 6
Base from which costs, emission reductions, and cost
effectiveness results are determined.
1 Based on MOBILE3 fuel
gasoline sales of 79.901
2 Based on MOBILE3 fuel
gasoline sales of 71.395
consumption model prediction
x 109 gallons m 1990
consumption model prediction of
x 109 gallons in 2010.
-------
5-23
IV. Effect of RVP Control on the Butane Market
The major method of RVP control is to remove butane from
gasoline, as described in Section II, and to replace the butane
with heavier components. The butane that is no longer used in
gasoline is made available to the market, including
refineries, and this excess supply could and likely will
decrease the market price of butane and economically impact
suppliers and purchasers. While in aggregate, this economic
impact should be zero (i.e., the benefits to purchasers of
cheaper butane and/or the products made using butane should
equal the cost to suppliers), there may be economic impacts on
isolated segments of the butane market. It is likely that most
if not all of the NGL industry impact will be a transfer of
revenues from one sector of the economy to another. Lost
revenues to the NGL industry would largely be made up by
increased revenues to current butane users or by reduced cost
to consumers of products which involve butane in their
manufacture.
The amount of excess butane that will result from a
one-psi reduction in RVP was made by the Gas Processors
Association (GPA) in their comments on EPA's gasoline
volatility study.[5] They assumed that a one-psi reduction in
RVP would result in a 2 percent reduction in gasoline
production, with virtually all of the reduction being butane.
Applying this 2 percent to the 2 billion barrels per year of
motor gasoline production, the GPA estimated an excess butane
volume of 40 million barrels per year. Over a 5-month control
program this would amount to 17 million barrels of excess
butane. This is a simplistic approach since it does not
recognize any alternative use of butane, either within the
refinery or outside the refinery. A more accurate approach
would be to examine the results from the Bonner and Moore
study. For the open butane purchase scenario, in PADD's l and
3, the modelling found the refinery purchased all the butane in
the one-psi reduction case that was purchased in the base
case. The results of PADD 2 showed 28,531 barrels per day were
not purchased in the control case, versus the base case.
Extrapolating these results to a nationwide, 5-month control
period, the model predicts 5.5 million barrels of excess butane
will be generated. For a 2-psi reduction, the model predicts
9.4 million barrels of butane will not be purchased by
refiners. These estimates are considered to be conservative
since, in the refinery model, the refiners were allowed to
purchase butanes only at the high butane price. If excess
butanes reached the market and a price decrease was realized,
the refiners may be able to utilize the low-priced butane to a
greater degree.
Jack Faucett Associates (JFA) was initially contracted to
evaluate the effect of reducing the RVP of gasoline on butane
prices and usage. The results of their study are presented in
-------
5-24
a report entitled, "The Butane Industry: An Overview and
Analysis of the Effects of Gasoline Volatility Control on
Prices and Demand".[7] JFA concluded that excess butane supply
from any level of RVP reduction evaluated would be large
compared to actual butane demand as a fuel or as a
petrochemical feedstock. Thus, they concluded that any
significant RVP reduction would result in enough excess butane
to cause butane prices to fall to the level of the
petrochemical floor price. However, the demand for these other
feedstocks is large enough that even extreme reduction of RVP
would not result m providing enough additional butane to drive
the price below the petrochemical floor price
The JFA study, dated May 30, 1985, needed to be updated to
reflect the current butane market situation, the current crude
oil price as well as to address other comments received from
EPA's gasoline volatility study. Bonner and Moore will be
investigating the butane market impacts and releasing the
results along with the results of the updated refinery computer
model runs. Bonner and Moore will address the issue of the
possible increased value of low price butanes to refiners,
alternative uses for butanes, and the total impact not only on
butane but for the full range of natural gas liquids. Until
more detailed information is available, the average of the
fixed and open butane purchase scenarios will be used for the
refining cost. The open butane purchase scenario represents
the case where all losses to the NGL industry result in
increased revenues to other sectors. The fixed scenario
represents the other extreme where no transfer of revenues
occur. The results of the Bonner and Moore study will be
incorporated into the final RIA when completed.
V. Increased Energy Energy Density and Evaporative Emissions
Recovery
This section presents an analysis of the effect of
reducing the volatility of gasoline on fuel economy. It is
hypothesized that, if gasoline volatility is reduced by
removing butane from gasoline and replacing it with other fuel
components, the heat of combustion of the gasoline will
increase. Furthermore, vehicular fuel economy should increase
with an increase in fuel heat of combustion. Thus, there
should be a fuel economy benefit resulting from reducing the
volatility of m-use gasoline. As a result of the analysis, it
is estimated that reducing RVP by 0.5 and 3.5 psi will increase
fuel economy by from 0.090 to 0.105 and 0.890 to 1.031 percent,
respectively, for feedback and non-feedback-equipped vehicles.
The remainder of this section explains how these estimated
increases in fuel economy were determined. It is divided into
three parts: 1) the relationship between gasoline volatility
and fuel heat of combustion, 2) the relationship between the
heat of combustion and fuel economy, 3) the overall
-------
5-25
relationship between gasoline volatility and fuel economy, and
4) the economic credit associated with recovered evaporative
emissions.
A. Volatility and Heat of Combustion
Quantifying the relationship between RVP and fuel
properties is difficult because of: 1) the complex refinery
operations involved in lowering RVP and maintaining octane and
other requirements, and 2) the relatively wide range of
commercial fuel heats of combustion occurring at any given
RVP. Relevant information from two different sources was
available to derive independent estimates of the effect of RVP
on the heat of combustion. Also, both API and MVMA were
requested to submit any relevant information they might have.
The two independent analyses and the two submittals are
described in the following paragraphs. Table 5-7 details the
quantitative results from each for direct comparison.
The first independent analysis used the output of Bonner
and Moore's linear programming model, which is being used to
estimate the cost of controlling RVP.[l] In addition to RVP
and octane, the model generates estimates of other fuel
properties, including API gravity, aromatic content, and
50-percent distillation temperature, which can then be used to
estimate fuel heat of combustion using a well-accepted
relationship defined in ASTM D3338-74. There was significant
variation from region to region in the effect of RVP reduction
on energy density, resulting from the varying regional
compositions of gasoline at different RVPs. The heat of
combustion increased 0.22-0.30 percent for a 1-psi RVP
reduction and 0.33-0.69 percent for a 2-psi RVP reduction.
Weighting the regional effects by gasoline production volume
resulted in weighted heat of combustion increases of 0.25
percent and 0.56 percent for a 1- and 2-psi RVP reduction,
respectively.
The second independent analysis examined MVMA fuel survey
data from January and July gasoline samples taken from 1979-83
(these were available on tape and could be accessed en masse).
The heat of combustion of each fuel sample in the surveys was
again estimated from the properties of the fuels using the
relationship from ASTM D3338-74, as described above. A linear
regression was then applied to relate the RVP and the heat of
combustion (BTUs/gallon) for the nearly 2,000 summer fuel
samples and also for the 4,400 summer and winter fuel samples.
For the summer gasolines, a 1-psi reduction m RVP from 11.5 to
10.5 psi resulted in a 0.25 percent increase in the heat of
combustion, with a range of 0.22 to 0.28 percent at 90 percent
confidence. For summer and winter fuels combined, a 1-psi
reduction in RVP from 11.5 to 10.5 psi resulted in a 0.33
percent increase in the heat of combustion, with a range of
0.32 to 0.34 percent at 90 percent confidence. The R was
-------
5-26
Table 5-7
Effect of Change in RVP on Change In Energy Density
Source of Information
1. Bonner & Moore
PADD 1
PADD 2
PADD 3
Volumetric Wtd. Average
2. MVMA Fuel Sampling Data
Summer Fuels
Summer & Winter Fuels
3. MVMA Submittal
Calculated Effect
4. API Submittal
Percent of Increase in Heat
Of Combustion (Btu/gal)
RVP 1-psi
30
22
26
25
0 . 25
0 . 33
RVP 2-psi
. 54
. 33
. 69
56
0 . 50
0 . 66
0.32 0.64
No calculated results
-------
5-27
only 0.09 for summer fuels and 0.30 for summer and winter fuels
combined, but due to the large number of samples, the
relationship is quite certain, as evidenced by the tight
90-percent confidence limits. The regression of summer fuels
is probably the most appropriate for use here. The winter
fuels were included to provide a wider range of RVPs and to
test the sensitivity of the results to range of RVP.
The MVMA submittal in this area outlined a first-order
analysis of the effect of reduced RVP on fuel heat of
combustion.[ 8 ] They stated that they were not aware of any
test data that would provide a direct relationship between RVP
reduction through butane control and vehicle fuel economy;
however, they did make some calculations to estimate the effect
of butane content on the heat of combustion. They too used the
method described in ASTM D3338-74 to estimate gasoline heat of
combustion from predicted average properties of the gasoline.
They estimated the percent decrease in fuel butane content
associated with a 1- and 2-psi change in RVP to be 1.8 and 3.7
percent, respectively. Then, assuming that the composition of
the non-butane portion of the fuel would remain constant, they
estimated the change in the heat of combustion using the
relative heats of combustion of gasoline, and while MVMA did
this using three baseline gasolines of various RVP, their
analysis ignores the fact that reduced butane content will
reduce octane. This octane can be replaced by further
processing of the gasoline feed stocks, which will likely
reduce the heat of combustion, or by increasing aromatic
content, which will likely increase the heat of combustion.
MVMA estimated that 1-psi and 2-psi reductions would increase
energy density by 0.32 and 0.64 percent, respectively. As
these figures are somewhat larger than those estimated using
the Bonner and Moore model and the regression of summer fuel
RVPs, it appears that increased processing to replace lost
butane dominates somewhat and reduces the net heat of
combustion increase by about 0.07 percent per psi RVP.
API, in their submittal, stated that there is no
predictable relationship between gasoline vapor pressure and
gasoline heat of combustion.[9] They state that a number of
compositional changes occur in reducing RVP to ensure that the
other properties of the gasoline remain in accord with ASTM
specifications and that production volume is maintained. As
evidence, they cite the fact that the scatter m the heat of
combustion of surveyed fuels at a specified RVP is greater than
the difference in the heat of combustion between RVPs, and make
the determination that any relationship between RVP and the
heat of combustion impossible to ascertain. Therefore, they
did not submit any conclusions on the net effect of all the
factors affecting the heat of combustion accompanying a
reduction in RVP, other than to state that the relationship is
unpredictable.
-------
5-28
As our own assessment of the heats of combustion of
MVMA-surveyed fuels indicated, there is a wide variation in the
heat of combustion at any given RVP and this variation is
larger than the effect of RVP. However, the 90 percent
confidence limits on the predicted slope take this variation
into account and still predict a range of only 0.25+0.03
percent per psi for summer fuel. Thus, while other factors can
overwhelm the RVP effect for any given fuel, on average the RVP
effect is quite certain and quantifiable. The fact that the
Bonner and Moore estimates fall within this small range, and
yet were quite independent, is further support for their
accuracy.
Given that the analysis of the survey data and the B&M
study yield essentially the same results, the B&M estimates
will be used here as it is a forward looking study of lower RVP
fuels across the board rather than backward looking survey
results. The result is that the heat of combustion is
projected to increase 0.25 percent with a 1-psi reduction in
RVP and 0.56 percent with a 2-psi reduction.
B. Heat of Combustion and Fuel Economy
As we will use it here, the R-value refers to the percent
change in fuel economy for a percent change in fuel heat of
combustion allowing for vehicle optimization. While much data
on the fuel economy of all types of vehicles exist in the
literature, few studies relate fuel economy to fuel heat of
combustion and there are none for which the vehicles were
optimized for both fuels. Thus, while the effect of heat of
combustion on fuel economy should be more consistent and
discernible than the effect of RVP on the heat of combustion
(once measurement variation is eliminated), few data exist from
which to determine accurate estimates. Compounding this is the
fact that the random variation in fuel economy measurements is
large (e.g., 3-5 percent) relative to the expected change in
fuel economy (less than one percent). Two sources of
information were used in analyzing this relationship:
empirical test data, and theoretical analysis of factors which
have been proposed either by industry or ourselves as possibly
influencing the value of R. These sources are all discussed
in the following subsections.
1. Empirical Test Data
In a letter to EPA dated August 15, 1984, General Motors
cited data on the relationship between gasoline heat of
combustion and fuel economy (see Table 5-8) which was presented
in a Chevron Research Co. SAE paper in 1974, and added that GM
testing on more recent systems supported the results presented
in the SAE paper.[10,11] The vehicles tested by Chevron were
from the 1970 and 1972 model years, and the SAE paper points
out that these tests can only give an indication of 1970 and
-------
5-29
Table 5-8
SAE 740522 Data on the R Valuetll]
Test Vehicle*
Chevron Test Cycle**
1970 NFB
1970 NFB
1970 NFB
1972 NFB
1972 NFB
1972 NFB
Avg
.455
.750
. 809
. 590
.380
.417
. 567
* Each of the three vehicles in each MY is from a different
manufacturer.
** Test cycle information.
Mode
Number Operating Conditions
1 Light acceleration to 35 mph, cruise at 35 mph
2 Medium acceleration 35 to 50 mph, cruise at 50 mph
3 Medium acceleration 50 to 65 mph, cruise at 65 mph, stop
4 Light acceleration to 35 mph, cruise at 35 mph
5 Medium acceleration 35 to 50 mph, cruise at 50 mph
6 Medium acceleration 50 to 65 mph, cruise at 65 mph, stop
7 Light acceleration to 35 mph, cruise at 35 mph
8 Medium acceleration 35 to 50 mph, cruise at 50 mph
9 Medium acceleration 50 to 65 mph, cruise at 65 mph, stop
10 Idle
NOTE: A timer is used to allow 30 s for each mode.
Accelerations are made holding manifold vacuum constant.
Manifold vacuums are determined for each car to provide light
and medium accelerations. Ambient temperature is 72°F.
-------
5-30
1972 car performance in general. Their results were based on
testing six vehicles with six fuels, repeating each
vehicle/fuel-specific test at least eight times. Fuel economy
was measured by weighing the fuel consumed and measuring the
distance traveled during each test cycle. The heat of
combustion of the fuels was calculated using a widely accepted
correlation involving API gravity, percent aromatics, and
volatility of the gasoline from ASTM-3338. The 1970 vehicles
showed an average R defined as the ratio of the percent change
in vehicular fuel economy to the percent change in gasoline
heat of combustion of 0.67, while the 1972 vehicles averaged
only 0.45 for a combined average value of 0.57. The GM
submittal did not supply or refer to any data supporting their
average value for R resulting from testing on more recent
systems, but they did state that the testing yielded R values
ranging from 0.1 to 0.9 with an average of 0.5. In their
submittal they suggested an overall average R of 0.6.
This value of 0.6 for R was later recommended by GM, Ford,
and MVMA in response to questions asked by EPA's Certification
Division in the EPA memorandum referenced in the supplemental
NPRM on the CAFE adjustment rulemaking.[ 12,13,14] This ratio
of increased fuel economy to increased heat of combustion of
0.6 was recommended for all vehicles (without differentiating
between feedback and non-feedback equipped vehicles). The
major conclusions of these manufacturers as stated in their
January 22, 1985 letters to EPA as well as more recent
information are discussed below.
In their letter to EPA, GM summarized data from five
vehicles (two with throttle-body injection (TBI) and three
carbureted) of model years 1981 and 1984, which are presented
in Table 5-9.[12] They include R factors for FTP and highway
tests. The method which GM used to measure or calculate the
fuel economies of these five vehicles was not stated in the GM
letter. The results for FTP testing were an average R of 0.62
for five vehicles, with a range of 0.34-0.89. For the highway
test procedure the average ratio was 0.53, with a range of
0.41-0.72. The 0.89 and 0.72 R values were both for a 1984
Pontiac J2000 with TBI.
GM did not explain why R is greater for the FTP tests than
for the highway tests. This is not the expected result, since
the FTP contains cold operation, where the highway test does
not. Typically the cold operation signifies both rich
operation and open loop feedback control operation. This
combination as will be explained later tends to demand extra
fuel thereby decreasing fuel economy and therefore R.
Although the main difference between vehicles abilities to
obtain high R values is whether or not they utilize electronic
feedback control (EFC), both the TBI vehicles and the
carbureted vehicles were probably equipped with EFC operating
-------
5-31
Table 5-9
GM Data on the R value[12]
R Factor
Test Vehicle FTP HWY
Carbureted Vehicles
1984 Olds Delta 0.35 0.51
(5.0L, 4bbl.)
1981 Olds Cutlass 0.39 0.41
(4.3L, 2bbl )
1981 Chevette 0.80 0.62
(1.6L. 2bbl.)
Average 0.51 0.51
TBI Vehicles
1984 Chevrolet Citation 0.65 0.41
(2.5L, TBI)
1984 Pontiac J2000 (1.8L TBI) 0.89 0.72
Average 0 77 0.56
Composite Average 0.62 0.53
(Carbureted and TBI)
Testing was not performed all at once, but at three
different dates corresponding to the year of each vehicle.
-------
5-32
over most of the test cycle, as all GM vehicles of model years
1981 and later used EFC. Therefore, it is possible from this
data to distinguish between TBI and carbureted vehicles. If
the data from the five vehicles are divided into two groups, as
illustrated in Table 5-9, the TBI-equipped vehicles showed
higher R values (0.77 vs. 0.51 for the FTP and 0.56 vs. 0.51
for the highway test) than the carbureted vehicles. The
composite data have an average R of 0.58, which may be rounded
off to 0.6. However, the fact that the TBI vehicles showed
higher R values than the carbureted vehicles and the statement
by GM that R can approach 1.0 for some operating conditions
(i.e., steady state) indicate that it may be appropriate to
assume an R higher than 0.6 for fuel-in]ected (FI) vehicles.
The better performance of the FI vehicles may be due to the
tighter control over fuel metering and shorter feedback delay
time which fuel injection offers.
In a direct response to EPA's question, "Is it appropriate
and/or possible to account for the effect of fuel energy
content on the vehicle's energy efficiency? If so, how should
this be done?" GM supplied no other data than that in Table
5-9 and a reference to SAE paper no. 740522 and recommended an
R of 0.6.
Ford also recommended that an R = 0.6 be adopted to
represent the 1980-85 model year vehicles.[13 ] The Ford letter
of 1/22/85 goes on to state that future model year vehicles
could respond differently and, thus, should be evaluated
separately if warranted by future fuel specification changes.
Ford based their conclusion on repeated test results from four
different vehicles with different engines and control systems
using two different fuels. Twelve LA-4(Hot) and 12 HWFET tests
were conducted on each vehicle and with each fuel for a total
of 192 tests. The method used to determine the fuel economies
of these vehicles (volumetric measurement or carbon balance
calculation) was not stated in the Ford letter. These results
are presented in Table 5-10.
One of these four vehicles was equipped with electronic
fuel injection (EFI), always accompanied by EFC, while the rest
were designated NFB for non-feedback vehicles. The R value for
the EFI vehicle for the hot-start test was 0.75 For the
hot-start portion of the urban driving cycle LA-4(Hot) fuel
evaluation and the highway fuel economy test (HWFET) fuel
evaluation analyses, the R values were 0.71 and 0.84,
respectively. The other three vehicles were NFB with average R
values of 0.35 for the LA-4 (hot) analysis, and 0.67 for the
HWFET fuel evaluation analysis. These R values are
significantly lower than those resulting from testing the EFI
vehicle. MOBILE 3 projections predict that by 1990 nearly 90
percent of gasolme-fueled vehicles m-use will have EFI or
will be feedback-equipped carbureted vehicles. Thus, based on
-------
5-33
Table 5-10
Ford Data on Ratio of Percent Change in
Fuel Economy to Percent Change in Energy Density[l3 ]
Test Vehicle M-H HS* CVS-H* HWFET* CVS-C/H*
w/o Feedback Controls
3.8-216 0.5974 0.21b8 1.1837
1.6-602 0.5178 0.5574 0.7173
5.0-807 0.200 5 0.2537 0. llfcl 0.84 6
Average 0.4749 0.3476 0.b724 0.84b
W/Feedback Controls
1.6-343 EFI 0.7502 0.7082 0.8403 1.065
Composite Average 0.5437 0.4378 0.7143 0.955
(Feedback and Non-Feedback)
M-H Hot-Start = City (55°o) and HWFET (45Jo) combined to
yield an overall average value.
CVS-H = City test cycle - hot-start.
HWFET = Highway test cycle.
CVS-CH - City cycle test - cold-start.
-------
5-34
the Ford data, it is reasonable to assume an R value higher
than 0.6 for vehicles in use in 1990, the majority of which
will use EFC.
The MVMA letter of January 22, 1985 analyzes the data
submitted by GM and Ford and reaches the same conclusion; 0.6
is a reasonable value for R.[14] The MVMA letter does not
propose a higher R value for feedback equipped vehicles vs.
non-feedback equipped vehicles.
Ford also mentions in its March 27, 1986 submittal[ 15 ]
some additional data on four more late model year vehicles
which suggested a higher value for R than 0.6 may be more
appropriate. This information obtained from SAE paper
860533[16] is presented in Table 5-11. The four 1984 and 1985
model year prototype vehicles were tested in triplicate over
the entire FTP test cycle. Three of the four vehicles utilized
closed loop feedback, control while the other one did not. The
combined average R value was an extremely high 1.850. The
paper did not actually calculate the R value, therefore it gave
no explanation for the high value, nor did Ford in its
submittal provide any additional comments.
API, in their April, 1986 submittal[17] presented single
run test data collected by Chevron on 12 1981-1983 vehicles of
varying control strategy shown in Table 5-12. The vehicles
were tested over the FTP cycle, however, ambient temperatures
were adjusted to 43°F and 55°F in an attempt to simulate actual
morning startup temperatures in the LA basin. Although there
was no standard FTP data taken to compare to, the R values
determined seemed to be unusually low due to the cold
temperatures. But rather unexpectedly, the R values at 55°F
were much lower for many of the vehicles than at 43°F,
particularly the carbureted vehicles. This fact remains
unexplained. As data were not taken at standard conditions, it
is of limited use when comparing it to other data. However, it
may be of some use in explaining the causes for lower R
values. Besides the temperature affect, there appears to be a
strong benefit to closed loop control when looking at the 55°F
FTP data. The 43°F data also shows an improvement under closed
loop control.
In the same submittal, API presented another earlier set
of data of tests done on eight open-loop 1973-1976 model year
vehicles. The testing was once again performed by Chevron (see
Table 5-13). This time the vehicles were tested over the FTP
with ambient temperatures of 75°F and 55°F. The 75° FTP seemed
to match within reason other standard condition FTP data, while
the 55°F FTP data suggests that the previous data at that
temperature for open loop vehicles may have been too low. This
data once again supports the argument for possibly lower R
values due to cold temperatures, with a greater effect on open
loop vehicles than on closed loop vehicles.
-------
5-35
Test Vehicle
City
HWY
1984 1.6L NFB
1 . 132
1 .345
1985 2.3L FB
4 .419
1 .698
1985 3.8L FB
-.637
- . 489
1985 5.0L FB
2 861
4 .350
AVG,
1 944
1 . 726
Table 5-11
Ford Data on the R Value[16]
Harmonically Weighted
1.211
3 . 434
- . 570
3 .323
1 . 850
-------
5-36
Table 5-12
API Data on the R Value[17]
Test Vehicle 43°F FTP
Open Loop
81 Aries .243
81 Datsun 1.116
82 Escort .198
83 Accord -530
83 SR5 .009
Avg 419
Closed Loop
82 Cavalier
. 034
82 Ciera
. 430
82 Corolla*
A A
83 Cavalier
951
83 Maxima
. 032
83 Cutlass
1 . 022
83 GLC*
1.117
83 Corolla
. 002
83 Lincoln*
* A
Avg
.513
Composite Avg. .474
55°F FTP
-1 . 707
- .210
-1.115
- .269
. 113
- . 638
.383
1 . 887
. 376
* *
.335
.451
. 642
A A
1 . 408
. 783
. 191
* Level of feedback control assumed to be closed loop.
** Not tested.
-------
5-37
Table 5-13
API Data on the R Value[17]
Test Vehicle
FTP @ 75°F
FTP @ 55 °F
73 Plymouth
. 124
. 810
73 Pontiac
1.114
. 957
73 Pontiac*
. 511
-.005
74 Volkswagen
. 544
. 341
7 5 Chevy
.614
. 194
75 Ford
1 550
615
75 Pinto
. 930
-1.189
76 Olds
.793
- .458
Avg. =
. 773
158
* Same vehicle
as the one
above, but with a
new carburetor.
-------
5-38
When all of the data are combined and weighted for the
year 1990 in which 90 percent of the vehicles in-use are
estimated to be closed-loop feed back controlled (see Table
5-14), then the data predicts an R value of 1.075. This is
actually greater than one, however, it is based on a combined
average of the FTP and Metro-Highway (a combination of the FTP
and HWFET cycles) data. While the R value for the
Metro-Highway cycle would be expected to be higher than for
the FTP due to lower cold start interference, the GM data in
Table 5-9 did not support this fact. Thus, combining the data
to obtain one value is not thought to introduce significant
errors. Even at the 55°F FTP test cycle condition, the
predicted R value is 0.69; greater than the 0.60 estimate given
by all of the industry sources as a good estimate. Only when
the temperature for the FTP test cycle gets to 43°F does a
lower value of 0.504 appear. However, there is such a broad
range in the data, that it is impossible to rely on these
values to any great extent.
2. Error in Testing
The broad range in the test data for R value suggests that
there may be sources of error in the testing. Some are
possibly procedural problems, while others are simply the
uncertainty of measured values. First, m determining the
effect of a reduction in gasoline RVP on vehicular fuel
economy, it is necessary to evaluate the vehicle that will be
operating on this lower RVP fuel, and the design optimization
of that vehicle. Automobile manufacturers are concerned with
obtaining the highest fuel economy possible to meet CAFE
regTjirements and advertise high fuel economies to attract
consumers. Vehicular fuel economy figures are the result of
testing over the EPA FTP and HFET cycles, in which the vehicles
are operated on Indolene, a 9-psi RVP gasoline. Therefore, the
manufacturers presumedly optimize the vehicle fuel systems to
operate on Indolene, to maximize the fuel economy figures
resulting from the FTP test cycle. As a result, when vehicles
are operated on a different fuel to determine the affect this
has on its fuel economy, the resulting value will be
erroneously low since the vehicle was not re-calibrated to
optimize for the new fuel. Thus the R value resulting from
this testing will also reflect this underestimated fuel-economy.
A second source of error in this testing is that the R
value is defined as the percent change in fuel economy to the
percent change in fuel heat of combustion. But vehicle fuel
economy is a function of many more fuel properties than just
the fuel heat of combustion. This alone can explain many of
the excursions in the test data. Two studies could use fuels
with the same change in fuel heat of combustion but relatively
large differences in other fuel properties such as specific
gravity, hydrogen/carbon ratio, and RVP. Thus, the resulting R
values determined will vary dramatically. For example, the 0
-------
5-39
Table 5-14
Summary of All Data on R Value
Open Loop Vehicles
Cycle
Std FTP/M-H
43°F FTP
55°F FTP
Chevron Cycle
No. Vehicles
12
5
13
6
R Value
.735 + .411
.419 + .432
. 148 + .803
, 567 + .180
Closed Loop Vehicles
Cycle No. Vehicles R Value
Std FTP/M-H
9
1 . 113
+ 1.356
43°F FTP
7
-513
+ . 507
55 °F FTP
7
. 783
+ . 615
Composite for 1990 (90% Closed Loop)
Cycle R Value
Std FTP/M-H 1.075 + 1.262
43°F FTP .504 + .496
55°F FTP .690 + .634
-------
5-40
to 3.5 psi change in test fuel RVP that exists in the studies
presented thus far resulted in a change in the fuel heat of
combustion of from 0.4 to 1.6 percent. However, this same
range in RVP's may result in an increase in evaporative
emissions to the canister of from 0 to 65 percent. Since the
vehicle evaporative recovery system is not modified to handle
this large increase, the resulting effect on the fuel economy
measurements could overwhelm the effect of the heat of
combustion.
Not only are there intrinsic sources of error in the test
procedure, but also in the estimation of fuel parameters used
in the testing. According to API, method D3338 which is
commonly used for calculating the heat of combustion of
gasoline, has an error range of + 0.74 percent. In addition,
method D1198 for calculating the specific gravity of the fuel
has an error range of + .0012 between labs (roughly + 0.2
percent). When typical values are assumed for the heat of
combustion and specific gravity, the two sources combine into
an error of + 0.90 percent for the heat of combustion per
gallon of fuel. This compares to the change in the heat of
combustion of only approximately 0 percent to 1.6 percent due
to RVP differences in the test fuels of 0 to 3.5 psi
respectively. These two values translate into a difference in
the calculated heat of combustion between the two fuels tested
of at best 1.6 + 1.8 percent. This range of error can have
serious consequences when trying to determine the value of R
for such a small change in the heat of combustion.
In addition, the fuel economy change of the vehicle using
the two different fuels is only on the average 1.6 percent if
we assume a value of R equal to 1 (a smaller value for R would
decrease this value still further). Thus, the fuel economy
measurement must be determined to a high degree of accuracy to
distinguish between the two fuels. Since the accuracy of any
one fuel economy measurement is at best within + 1 percent,
then it is once again extremely difficult to accurately
determine the change in fuel economy and therefore the value of
R by actual test methods even with replicate testing. For this
reason theoretical analysis remains as what we believe to be
the only viable alternative. However, the test data described
previously may still be useful in corroborating the results of
theoretical analysis.
3. Theoretical Analysis of the Value of R
There are many parameters which must be taken into account
when analyzing the fuel economy of a vehicle. But in this
instance we are only interested in those things which are
affected by a change in the fuel properties. Figure 5-2 shows
an approximate distribution of energy consumption for a typical
vehicle. In the ideal case, these fractions will remain
constant; independent of changes in the fuel properties. This
-------
Figure 5-2 [2U
VEHICLE ENERGY DISTRIBUTION
-------
5-42
will result in identical engine efficiency, and therefore, the
energy required to drive the vehicle for one mile will remain
constant, resulting in an R-value of 1.0. Unfortunately, the
real situation is such that some of these fractions will vary
due to changes in the fuel properties. This in turn will cause
the value of R to vary.
Throughout their comments on the subject of R-value,
industry supplied several reasons in support of their assumed
R-value of 0.6. These are listed in Table 5-15 in order of
what is thought to be their relative importance.
The first point, higher losses (lower efficiency) m the
engine cylinders due to higher flame temperatures, requires
in-depth quantitative analysis to evaluate to what degree it
actually affects the value of R. A description of this
analysis follows, however, it is based on the change in heat of
combustion rather than on the change in maximum flame
temperatures. This is believed to better represent the average
losses in the cylinder.
A model derived for the ratio of the heat loss in the
cylinder of an engine to the heat of combustion of the fuel was
used to determine the change in heat losses from the cylinder
of the engine due to the change in fuel properties with RVP
control.r18J Table 5-16 shows the fuel properties used in this
analysis. Values for the 9.0 RVP fuel and butane were assumed,
and properties for the other gasoline fuels were weighted based
on the fuel butane fraction. Table 5-17 shows the model which
was used to determine the heat losses, and describes the
variables which are used in the model as well as providing the
values assumed for this analysis. The results indicate that
the fraction of heat losses to the cylinder walls increased by
approximately 0.0247 percent from 19.6309 percent to 19.6358
percent of the total energy input with the lower RVP, higher
heat of combustion fuel.
This increase can then be applied to the value of 29
percent previously assumed for the losses to the engine
cylinder from Figure 5-2. If we also assume that this same
increase applies to the heat losses to the exhaust (33
percent), then the energy required with the new fuel to move
the vehicle one mile can be determined from the following
equation:
1) Q2=Q2(.62) (Qevll) + (.38)(Qi)
(Qc y I I )
Where: Q= Energy required to drive vehicle one mile
1= Original base case fuel
2= New fuel
Qcyi= Calculated fraction of energy lost to
the cylinder
-------
5-43
Table 5-15
Reasons Given to Support a Low R Value
1. Increase in flame temperature due to increase m
aromatics content causing higher losses (low efficiency)
in the cylinder. (Closed loop affect)
2. Cold F/A enrichment and open loop operation of feedback
systems at cold start (Open loop affect)
3. Test cycle transients. (Open loop affect)
4. F/A ratio shifts slightly richer. (Closed Loop affect)
5 Even late model vehicles do not fully use a change in
fuel heat of combusion under conditions where they
operate rich, including idle, acceleration, and cold
start and warm up. (Open loop affect)
6. Higher density fuels have higher concentrations of
aromatics which increase viscosity, surface tension and
latent heat of vaporization. These tend to decrease
evaporation rates with a resulting decrease in fuel
mixture homogeneity and quality of cylinder to cylinder
distribution. (Closed loop affect)
7. High density low H/C fuels may run at higher equivalence
ratios resulting in lower efficiency and lower R. (Closed
loop affect)
8. Increased friction. (Closed loop affect)
-------
5-44
Fuel Parameter Butane
Heat of Combust ion(Qc) 19643
in BTU/lb
Specific Gravity 0.5836
(Density in lb/gal) (4.87)
Stoichiometric F/A 0.64516
Liquid Viscosity (20°C) 0.18 **
in centipoise
Specific Heat of Charge 0.2847
in BTU/lbm °R
Table 5-16
Fuel Property Change Summary*
11.0 RVP
Gasoline
18532
0.7349
(6.13)
0.06884
0.342
0.28266
9.0 RVP
Gasoline
18500
0.7405
(6.18)
0.06897
0.348
0.2826
11.5 RVP
Gasoline
18540.9
0.7335
(6.12)
0.06881
0.340 cp
0.28267
13.9 RVP
Gasoline
18580
.7267
(6.06)
0.068665
0.333 cp
0.28274
Propane Methanol
19770 8580
.49973
(4.1705)
0.064222
.796
(6.64294)
0.155552
0.2856
0.2979
Changes for gasoline based strictly on replacement
with gasoline.
Since butane is normally a vapor at 20°C, this is an
of 3.6 vol. percent (2.8 v/eight percent) butane
extrapolation from 0°C.
-------
5-45
Table 5-17
Modeling Equation:
QCJ1 = Ke (Tq-Te)(1+F) G; 25
FQc
Term
Qc
y i
Ke
Kg
b
mgD
Ta
Tc
Description:
Heat lost from the gases
in the cylinder
Divided by the
Heat of combustion
of the fuel
(used in the cylinder)
(10.4 Kg/b) (b/mg0) 75
Gas conductivity at T9
Cylinder bore
Gas viscosity at Tg
Assumed Value:
0 . 4073
8.5x10 ~ 6 BTU/Sec ft°F
2.5"
22x10"* lbm/sec ft
Mean effective gas
temperature
72 + ( 693 *QC z *F z *C„,*{1+Fl))
Qci*Fi*CP z *(1+F2)
765°F for base case
for new case
Mean effective
coolant temperature
Fuel-air ratio
17 0 °F
Stoichiometric (See Table
5-16)
Qc Heat of combustion
per unit mass of fuel
G9 Gas flow per unit time
per unit piston area =
P(1+F) 2545
3600 (#cyl)(Ap)(F)(Qc)(Ni)
P Engine horse power
Nt Thermal efficiency of
the engine
Ap Piston area = Tr(b/2)2
#Cyl Number of cylinders
Cp Charge Specific Heat
1 Base Case
(See Table 5-16)
1.14305 1bm/sec f2
@ low RVP
1.14439 @ high RVP
100 HP
0.38
0.034088 ft2
4
(See Table 5-16)
?
-------
5-46
The last term remains constant over fuels, since the
magnitude of the individual work necessary to move the vehicle
one mile does not change at this poinr in the analysis.
Solving the equation for the ratio of energy required to drive
the vehicle in the new case as opposed to the base case (i.e.
Q2/Q1) yields a value of 1.000403. Since this is greater
than 1.0, the energy required to move the vehicle with RVP
control has increased, which is to say that the value of R has
decreased below 1.0. The fuel economy of the vehicle with the
new fuel can now be determined from the following equation:
2) FE2 = FEi x Qt x QC;
Qc, x Q2
Where: FE = Fuel economy
1 = original base case
2 = new fuel case
Qc = fuel heat of combustion per gallon of
fuel
Q = as before
The value for the original fuel economy for new vehicles
in 1988 is assumed to be 28.35 miles per gallon based on Mobile
3. The new value for the fuel economy from the above equation
is then 28.506 miles per gallon Once this number is
determined, the value of R for this situation can be determined
from the following equation:
3) R = FE; t FE, - 1
Qc 2 Qc 1 — l
The resulting value for R is 0.931.
The second point listed in Table 5-15, cold F/A enrichment
during open loop feedback control also deserves some
consideration. Extra fuel is required during a cold start in
order to get enough fuel evaporated in the cylinder to enable
combustion. If you now have a fuel with lower volatility, you
must use still more fuel in order to get the minimum amount
vaporized. This greater amount of fuel has a higher heat of
combustion causing still a larger use of fuel energy.
Unfortunately, data were not available on a real time
basis for the fuel consumption during a cold start over the
open loop portion of the test for high and low RVP fuels. It
was thought that a comparison of bag 1 versus bag 3 of the FTP
could give an overall estimate of the cold start open loop
effect. However, this would also include the effects of cold
engine temperature during closed loop operation which as
discussed later, should not affect the value of R. Therefore,
it would be a severe over-estimation of the cold start open
loop effect. In addition, estimation of the fuel volatility
-------
5-47
effect proved to be very difficult due to lack of knowledge of
actual required F/A ratios over the period of cold start, open
loop performance.
Thus, although this cold start, open loop operation effect
on the value of R may be real, it is not possible to determine
its magnitude from available data. In addition this effect
might be compensated for by the low RVP, higher energy content
fuel raising engine temperatures faster than the higher RVP
fuel. This in turn would decrease the time period required for
open-loop operation during the cold start, saving fuel and
therefore raising R
The third point presented m Table 5-15 which may reduce
the value of R, is test cycle transients. There can be no
doubt that test cycle transients affect the efficiency of the
engine. However, the value of R is a ratio of the efficiency
of an engine with one fuel to the efficiency of the engine
using a base fuel. As the test cycle transients are not a fuel
property, they by themselves cannot affect the value of R.
However, since the fuel with RVP control has a higher
volumetric heat of combustion, and transients cause F/A ratio
enrichment which "wastes" fuel, the first thought is that you
are now "wasting" higher energy content fuel, therefore
decreasing your R value. But even though you are "wasting"
higher energy content fuel, you are also getting more work from
the higher energy content fuel which is burned. For this
reason the efficiency of the engine during transients, although
decreased relative to steady state, should still be constant
across fuel parameters resulting m no change in the value of
R. This may be supported somewhat by the conflicting data from
Ford and GM. Ford showed a decrease in R over a cycle with
more transients while GM showed an increase (see Tables 5-9 and
5-10).
The fourth point states that the F/A ratio shifts slightly
richer. Although the variability between fuels is large, the
primary effect of reduced butane content supports this
statement. The changes in fuel properties as seen in Table
5-16 reflect weighted averages of replacement of 3.6 volume
percent butane with gasoline. The specific gravity increase of
0.76 percent will cause 0.76 percent more mass of fuel to be
metered since it is metered volumetrically. However, not all
of this added fuel enrichens the F/A mixture since the F/A
ratio required for stoichiometry has also increased by 0.19
percent. Assuming that most vehicles by 1990 will have closed
loop feedback control utilizing oxygen sensors, the vehicles
entering the market should be able to adjust back to
stoichiometry for this slight deviation during operation in the
closed loop mode causing no detrimental effect on the value of
R. In addition, the open loop control set point of new
vehicles can be adjusted to account for the change in fuel
properties resulting in no added losses. Current in-use
-------
5-48
vehicles both closed loop and open loop, however, will not be
able to compensate for the change in fuel properties when they
operate in the open loop mode. In addition, as will be seen
later, the sensitivity of our modeling to the F/A ratio is such
that the heat losses to the cylinder walls are affected by the
F/A ratio in a manner such that with the fuel composition
change we are expecting to see, the value of R should actually
increase instead of decrease,
The fifth point, which emphasizes the inability of even
late model vehicles to utilize the change in fuel heat of
combustion when they operate rich, is actually the sum of
previous discussions focusing on transients and cold start.
Thus, no further discussion of it is needed.
Warm-up as it pertains to open-loop operation was
discussed above under open loop operation due to cold start.
However, warm-up can also be used to describe the condition
after closed loop operation begins, but when engine temperature
has not yet stabilized. In this case, the higher combustion
temperatures due to an increased heat of combustion content
will cause greater losses per unit time to the still cool
engine in the same manner as discussed under point number one.
However, these higher combustion temperatures will also
decrease the time for vehicle warm up and the time to reach
equilibrium For this reason, the effect of closed-loop
warm-up is not predicted to have any detrimental effect on the
value of R.
For the point made of idle operation equation 1 can be
modified to account for the added loss associated with the
increase in fuel heat of combustion for the periods when the
vehicle is operating at zero percent efficiency. From Figure
5-2, these periods (coast and idle) represent roughly 4 percent
of the energy consumption of the vehicle with the base case
fuel. This should increase proportionately to the volumetric
increase in fuel heat of combustion since the fuel is metered
volumetrically. Therefore, equation 1 becomes:
4) Q2 = Qz ( •62)(Qcvlz) + Q, (.04)(Qc2.) + -34Q,
Qc y 1 1 Qc 1
Once again, solving for Qz/Qi and substituting into
equations 2 and 3 yields a new value for R of 0.826.
This, however, is not precisely accurate, for there are
two different conditions, cold idle and hot idle. The case can
be made that for the hot idle condition the manufacturer can
change the set point of open loop idle operation to adjust the
flow of fuel in proportion to the change in energy content.
Therefore, no additional losses will occur. However, as has
been discussed, for a cold start, additional fuel is required
over and above the normal amount due to the decreased
-------
5-49
volatility of the fuel. So, m the case of cold start idle,
not only can the amount of fuel not be decreased in proportion
to the increase in energy content, but it must be increased
Thus, cold idle could have a detrimental effect on the value of
R. This detrimental effect will depend directly on the length
of time at which the vehicle is allowed to idle. As before,
however, a benefit may be derived from the higher engine
temperatures due to the higher energy content fuel. These
increased temperatures should decrease the time required for
open-loop operation following idle, thereby reducing the total
losses due to the cold vehicle operation.
The above equation probably over-estimates the effect of
hot idle, and may under-estimate the effect of cold idle
operation. Although the hot portion is probably much greater
than the cold portion, no truly accurate estimate can be made.
For this reason equation 4 remains as the best available
estimate of the effect of idle.
The sixth point covers a broad range of topics including
the estimated changes m fuel density, viscosity, surface
tension, and latent heat of vaporization; the effects that
these changes have on fuel vaporization rates; and the effect
that this has on fuel mixture homogeneity and cylinder to
cylinder distribution. Tables 5-16 and 5-18 show the average
expected changes in fuel properties due to the primary cause of
a reduction in butane with RVP control. Although other
secondary causes can possibly dominate the effect of the change
in butane content, this is the only one which is easily
quantifiable and it may accurately represent the average
changes in the fuel properties due to RVP control.
The fuel properties in question such as surface tension
and latent heat of vaporization not listed in Table 5-16 are
not easily quantifiable as they relate more to properties of
the fuel/air mixture. In any case, the changes in these
properties will be slight in comparison to liquid fuel changes
since the air properties will tend to dominate. The percent
vaporized at 330°F is a value that should lump the effects of
these fuel/air properties together.
The change in percent vaporized at 330°F is so slight
(0.055 percent) as can be seen m Table 5-18 that any affect of
the change in fuel evaporation rates on fuel mixture
homogeneity or cylinder-to-cylinder distribution would be
unnoticeable. However, the change in liquid fuel viscosity
although still not large, is significant with an increase of
1.7 percent. But its effect is only seen in the fuel
distribution system of the vehicle and is related to efficiency
(and, therefore, R value) through the increased load on the
fuel pump. However, since most fuel pumps for fuel in]ected
vehicles run at constant load, pumping more fuel than is needed
to the fuel injectors and recirculating the rest back to the
-------
5-50
Table 5-18
Change in Percent Vaporized at 330°F[l]
Base
Low RVP (-2)
Diff.
PADD 1
94 . 685
94 . 942
+ .257
PADD 2
93.748
93.920
+ .172
PADD 3
92 . 543
92.314
-.229
Weighted Avg.
93 . 131
93.076
-.055
NOTE l: Values are for pool average.
NOTE 2: Due to the marketing of butane, we would not expect
butane to be used in the refineries. Therefore, open
NGL was assumed most representative.
NOTE 3: PADD Volumetric Distribution
1) 0.0883
2) 0.2925
3) 0.5399
4) 0.0793 (excluding Calif.)
-------
5-51
fuel tank, the increase in viscosity will not cause any
increased load on the vehicle. It will merely cause a slight
decrease in the amount of fuel which is recirculated. This
small decrease should not have any detrimental effect on the
engine operation. A second place where viscosity can effect
vehicle operation is m the fuel injector, where an increase m
viscosity could dampen the flow of fuel out of the fuel
injector during transient operation. This would have the
effect of increasing vehicle efficiency, and therefore raising
the value of R. However, since much of the fuel flow through an
injector is in the turbulent flow region, it is for the most
part independent of the fuel viscosity. Therefore, it too has
a negligible effect, and the effect of fuel viscosity on the
value of R can be ignored.
The seventh point states that higher equivalence ratios
may cause lower R values. While it is true that the
stoichiometric F/A ratio increases as has been discussed above,
it is not true that the equivalence ratio increases. The
equivalence ratio is the ratio of the F/A ratio to the
stoichiometric F/A ratio. Since the stoichiometric F/A ratio
is increasing, the equivalence ratio should actually decrease
with RVP control. The only exception is when the actual F/A
ratio changes proportionately (which is what we are assuming
for modern vehicles) in which case the equivalence ratio
remains constant. Therefore, the presumption that equivalence
ratio is increasing is erroneous and this is not a valid reason
for a low R value.
The final point, increased friction, as it relates to the
vehicle, is independent of the fuel being used since the
friction per mile of travel remains constant. Therefore, it is
irrelevant m relation to the R value. The engine friction
which represents 13 percent of base case energy consumption,
however, does not necessarily remain constant. The friction
per mile is proportional to the number of cylinder strokes, and
this m turn is a function of the power output per stroke. For
this reason, the engine friction should change in proportion to
the amount of work available from the fuel in the cylinder
while maintaining the necessary stoichiometry. This ratio is a
function of the fuel-air ratio, the heat of combustion of the
fuel per unit mass, and the change in the fraction of energy
lost to heat. Therefore, including the effect of fuel
parameters on engine friction per mile driven into equation 4
yields:
5) Qz = Qz ( -62)(Qcvl ,) + Qi ( . 04) +
Qcy I 1 Qc I
Qi(•13) * (Qr, * F, * Qc vi z) + ¦21Qi
Q f 2 * F 2 * Qcyll
-------
5-52
Where: Qr=the fuel heat of combustion per unit mass
F =the fuel-air ratio
(Other parameters as before)
Solving this equation for Q2/Qi and substituting into
equations 2 and 3 yields a value for R of 0.82.
This represents a maximum value that the value of R can
take on due to the added heat losses associated with the
increase in the heat of combustion of the fuel and the change
in other fuel properties associated with a 2 psi reduction in
RVP. It should be noted, however, that this is the maximum
only for the fuel property values used in this analysis which
are averages predicted for a high and low RVP fuel. Actual
fuel data used in a similar manner can far exceed the 0.82
value cited here.
An analysis performed on the model to test for its
sensitivity to the various fuel parameters involved shows that
holding everything else constant, changing the heat of
combustion of the fuel alone has little or no effect on the
value of R. The change in cylinder temperature associated with
the change in heat of combustion seems to offset it nearly
equally resulting in a constant value of approximately 0.2 for
R. Thus the difference between this value and the previous
value of 0.82 must be due to other fuel parameters. The change
in fuel heat of combustion alone cannot be used to accurately
define the value of R.
The change in specific gravity has a large effect on the
value of R as can be seen in Figure 5-3. In the range of
realistic changes in the heat of combustion and the specific
gravity the R value can be expected to increase, thus
explaining part of the increase in the value of R from 0.2 to
0.82.
The fuel-air ratio (F/A), which is used as an input a
number of times into the heat loss modeling equation and its
constituents, also has a large effect on the value of R as can
be seen in Figure 5-4. The effect of F/A is highly dependent
on the change in the heat of combustion of the fuel, but in the
range that we would expect for the changes in fuel properties
with RVP control, the effect of F/A would be to greatly
increase the value of R.
Figure 5-5 shows the overall effect of changes in the heat
losses in the engine due the changes in fuel properties. The
response of the R value to the change in heat losses appears to
be linear. The rather flat slope of the line indicates that
even if the value which was calculated for the change in heat
losses used in the previous analysis was incorrect, a gross
error in this estimate would not significantly effect the
resulting value of R.
-------
Figure 5-3
Sensitivity to Sg
for vorying heat of combu3tion/lb. fuel
Change in Specific Gravity
-------
11
10
9
8
7
6
5
4
3
2
1
0
Figure 5-4
Sensitivity to F/A Ratio
decreasing heot of combustion /lb. fuel
= -100 Btu
!
0.0005
Change in F/A ratio
D01
0.0
-------
Figure 5-5
Sensitivity of R to Heat Loss Estimate
% Change in Heat Loss from Cylinder
-------
5-56
In summary, out of all of the possible reasons provided
for a reduction in the value of R, only f ive of them are
expected to have any effect at all, and of these, one can be
expected to actually increase the value of R. The first point,
higher cylinder temperatures due to the fuel property changes
associated with RVP control was shown to result in a value of R
of approximately 0.931. This value, however, was decreased by
inclusion into the calculations of idle operation and engine
friction resulting in a final R value estimate of 0.82. The
fourth point, an increase in the F/A ratio tends to increase
the value of R and has already been included throughout the
calculations. The only point not included in the calculation
was point 2; cold start open-loop operation. This can be
expected to decrease the value of R, but since the effect of
idle has probably already under estimated the value of R, the
best assumption with the current information is to let the R
value remain at the calculated level of 0.82.
To qualify this R value, it should be noted that it is
based on a change in fuel properties which results from a
change in fuel RVP. There are many other ways in which the
fuel can be modified, with the expected changes in fuel
properties different m each situation. For example, a change
in the aromatic concentration of the fuel with little change in
the butane concentration may result in vastly different
specific gravity or F/A ratio changes for the resulting change
in heat of combustion. As a result, the R values calculated
may also prove different. The numbers shown above are only for
the case of RVP control.
In addition to those already discussed, there is one more
point which may increase the value of R. Since manufacturers
currently optimize their engines on indolene test fuel rather
than the actual in-use commercial fuel, in order to maximize
the estimated corporate average fuel economy, there is gain to
be made by having the in-use fuel be closer to the fuel for
which the vehicles are optimized. However, again, no
quantitative estimate can be made for this effect.
In an effort to support the modeling performed, the
results from using the model on propane were compared with
those from in-use information. However, the use of propane in
the model requires an additional slight modification. Since
the heat of combustion of propane per gallon of fuel is so much
less than that of gasoline, it was assumed that a larger energy
load on the fuel pump inversely proportional to the decrease in
heat of combustion would be required to deliver an adequate
amount of fuel to the engine. The fuel pump requires at most
0.25 HP or approximately 0 25 percent of the power output of a
typical engine which is approximately .0625 percent of the
energy input to the engine. Therefore, equation 5 becomes
-------
5-57
6) Q2= Q2(.62)(Qcv12) + Q,(¦13)(Qri+F1+Qcv>2)
Qcyll QfZ+FZ+Qcyll
+ Q.(.04)(2ci) + Q, ( .000625)(g^)+ 0.209375 Q,
Qci Qci
The effect of this change to the model is small even with
such a large change in fuel heat of combustion. Using values
for propane versus 9.0 RVP gasoline in the model now results in
an R value of 0.862. From actual vehicle data on propane
fuel[l9] the average increase in vehicle energy efficiency
relative to gasoline is 4.82 percent. This increase in
efficiency, when put into the following equation with the
corresponding fuel properties results in an R value of 0.865.
7) R = % A W
•6 A Q c
where: W = Qc x eff
eff= energy efficiency of the engine
Although the possible errors in both the model and the data are
significant, the remarkable correlation does serve to support
the use of the model in other applications.
Like propane, methanol is also a fuel for which there is a
relatively large amount of m-use data. When methanol is
burned in a gasoline engine without making any significant
design changes, the efficiency of the engine tends to improve
by approximately five percent. This efficiency increase when
combined with the fuel properties as shown in Table 5-16
results in an R value from equation 7 of 0.95 when compared
with 9.0 RVP gasoline.
Since methanol is not a petroleum fuel, its application to
volatility vcontrol may be questionable. However, since it is a
pure compound, its properties are well known In addition,
since a comparison between gasoline and methanol involves large
differences in fuel properties, the magnitude of the
differences is statistically significant. Therefore, an
R-value of 0.95 should represent a reasonable maximum value
while the theoretically calculated 0.82 value is kept as a
reasonable minimum value. Until such time as better fuels data
is available, or better test data is available, the best
estimate for the R value should be approximately 0.82 to 0.95.
4. R Value as it Relates to Evaporative Purge
In addition to the R value associated with the change in
fuel properties. There is an R value associated with the
efficiency at which a vehicle can burn the fuel-air mixture
which results during purging of the evaporative canister. From
data projected for the difference between controlled and
-------
5-58
uncontrolled evaporative emissions using 11.5 RVP fuel for the
vehicle fleet in 1990, the amount of hydrocarbon trapped by the
average evaporative canister is 47 685 grams per gallon of fuel
during normal operation Over the LA-4, it is estimated in
Section I of Chapter 4 that the vehicle is able to purge the
canister for slightly more than half of its driving time If
we assume then that the 47.685 grams of hydrocarbon is now only
purged over half the time, and further, for a somewhat worse
case, assume that it is all butane, then 4.317 liquid volume
percent of the hydrocarbon entering the engine during purge is
butane Using a rule of thumb that 1 8 volume percent butane
represents a change of 1 psi in RVP, during purge the engine is
essentially burning a fuel with an RVP of 13.9, and
approximately an 11 RVP fuel the rest of the time. The fuel
properties for the 11 RVP and 13 9 RVP purge case are shown in
Table 5-16. By applying these fuel parameters to the same
models, equations, and conditions as previously yields an R
value for this case of 0.831. In this instance, this means
that the 0.71 percent drop in volumetric heat of combustion of
the fuel during purge decreases fuel economy by only 83 1
percent of this amount This of course assumes that the
vehicle is able to maintain stoichiometry during purge, which
was established in Section I of Chapter 4.
The purged hydrocarbon is entirely a benefit to fuel
economy. Previously, the vapors merely escaped into the
atmosphere and remained unburned. With the canister, the
vapors are now consumed, but due to the carbon balance
measurement of fuel economy, the benefit to fuel economy
measured over the EPA test procedure goes unnoticed. However,
using the R value calculated above, the relative efficiency of
the burning of this purged hydrocarbon in the engine can be
determined Assuming as before that it is all butane, and that
stoichiometry is maintained during purge, the relative
efficiency of butane to gasoline is 1.0334. Thus, not only is
there a benefit to actual fuel economy from the purged vapors,
but there is the added benefit that they actually burn more
efficiently than the gasoline coming from the fuel tank.
Applying the estimated range for the value of R results in a
corresponding range of 1.00976 at R = 0.95 to 1.03492 at R =
0.82 for the relative combustion efficiency of purged butane to
that of gasoline
C Overall Relationship Between Gasoline Volatility and
Fuel Economy
Combining the relationship between fuel volatility and
energy density (from the Bonner and Moore study) with our best
estimate for the relationship between energy density and fuel
economy yields the overall relationship between fuel volatility
and fuel economy. Energy density is projected to increase 0.25
percent with a 1-psi reduction in RVP and 0.56 percent with a
2-psi reduction m RVP. Vehicles should take advantage of this
-------
5-59
increase in energy density to achieve a resultant increase in
fuel economy. Thus, for both feedback and non-feedback
equipped vehicles, the increase in fuel economy for 1- and
2-psi reductions in RVP would be 0.205 to 0.238 and 0.459 to
0.532 percent, respectively. The fuel economy effects for
other RVP reductions were derived by fitting a curve through
these fuel economy increase values for 1- and 2-psi RVP
reductions and are shown in Table 5-19.
These fuel economy increases were used to evaluate a
dollar credit resulting from gasoline RVP control. This credit
was determined by multiplying the percent increase in fuel
economy by the total number of gallons of gasoline consumed by
motor vehicles, and then valuing that gasoline at $0.82 per
gallon. This value was determined by subtracting a
consumption-weighted state fuel tax of $0.13 per gallon and the
Federal tax of $0.09 per gallon from the 1984 national-average
retail gasoline price of $1.04 per gallon.[20] In other words,
the consumer would be able to travel additional miles on the
high energy, low RVP gasoline, and is thus credited the dollar
value of the gasoline he would otherwise have had to purchase
in order to travel those extra miles made available by the
resulting higher vehicular fuel economy.
D. Economic Credit From Evaporative HC Recovery/
Prevention
There are three major sources of evaporative hydrocarbon
emissions that are associated with gasoline RVP. They are
losses from the gasoline distribution system (such as bulk
storage terminal breathing losses, bulk transfer losses, and
service station losses from transfer and underground tank
breathing), refueling emissions, and motor vehicle evaporative
emissions. The emission reductions for each of these sources
associated with RVP control are detailed in Chapter 3 of this
study. However, in addition to representing an environmental
benefit, these emission reductions also represent an economic
benefit in that these HC emissions are now available to be
consumed as fuel by the motor vehicle (i.e., current emissions
due to high volatility fuel represent a cost to the economy) .
The same is true for certification fuel RVP control, although
only evaporative emission reductions from motor vehicles are
relevant there. The methodology used to evaluate this cost
credit resulting from the recovery and prevention of
evaporative HC emissions via both fuel and vehicle control is
outlined below.
The reductions in evaporative HC emissions from the
distribution system, refueling, and motor vehicles (as
described in Chapter 5) are used directly to determine the mass
of HC now usable that would otherwise be lost if excess
evaporative emissions were not controlled. This tonnage of
hydrocarbons is converted to an equivalent volume using the
-------
5-60
Table 5-19
Fuel__Economy Effect of RVP Control
RVP Reduction Percent Increase in
_ (PSi) Fuel Economy
Best Estimate
R = 0 82 R-0.95
0.5 0 090 0 105
1 0 0 205 0.238
1.5 0 328 0 380
2 0 0 459 0 532
2.5 0 599 0 694
3 0 0 741 0 859
3.5 0 890 1 031
-------
5-61
density (lb/gal) of the hydrocarbons. Because the lighter
hydrocarbons evaporate first, the specific gravity and energy
densities (Btu/gal) of those hydrocarbons no longer lost to
evaporation are significantly less than those of gasoline. As
a first-order estimate, the evaporative hydrocarbons were all
assumed to be butanes. The equivalent volume (gallons) of
butane saved by RVP reduction is converted to energy using
butane's energy content. This energy is then converted to
equivalent gallons of gasoline, using a representative gasoline
energy content figure (Btu/gallon). This volume of gasoline is
then converted to a dollar amount using a value of $0.82 per
gallon of gasoline for the $20/bbl case. Overall, the value
of a ton of evaporative emissions (butane) controlled or
prevented is $290.32. The value of a gallon of gasoline needed
to be determined as a function of crude oil price. Data was
collected from Oil and Gas Journal from September of 1985 to
September of 1986 in order to draw a correlation between
gasoline price and crude oil price. The retail price of
gasoline (minus taxes) was found to be strongly correlated to
crude oil price if a 2 week lag period between the crude oil
price and the pump price was assumed (R=0.94) The results of
the correlation for the range of crude prices examined in this
study, as well as the estimates for densities, energy
densities, and gasoline value are summarized in Table 5-20.
VI. Effect of RVP Control on Dnveability
This section presents an analysis of the effects of
reducing the volatility of gasoline on vehicle driveability.
In the period from 1974 to 1985 the volatility of unleaded
regular gasoline as measured by the Reid Vapor Pressure (RVP)
increased by 10 percent to 20 percent depending on the area of
the country (see Table 5-21, Figure 5-6). In 1985 for the
first time the average level of volatility surpassed the
average ASTM recommended limit with a level of 10.8 psi RVP and
maximum levels as high as 13.6 psi RVP. The increase for
alcohol-containing unleaded regular gasoline has been even
greater with average levels reaching 11.5 psi RVP and maximums
as high as 14.0 psi RVP. Although during the same period
vehicles have been designed to run satisfactorily on higher
volatility fuels, there is evidence which suggests that the
current fuel volatility has increased beyond the design point
of many vehicles in service. Some vehicles experience varying
levels of vapor lock and fuel foaming, both of which are caused
by high volatility, and both of which cause vehicle surge,
hesitation, power sags, engine stalling, and difficulty or
failure to start. This poor engine performance in turn results
in a severe increase in vehicle exhaust emissions. For these
reasons there is an immediate benefit to the auto manufacturer,
the consumer and to the environment to be realized through
better engine performance by reducing the volatility of the
fuel.
-------
5-62
Table 5-20
Estimates for Evaluating the Evaporative Recovery/
Prevention Credit Resulting from RVP Control
Description
Composition of Evap. Emissions
Density of Butane
Energy Density of Butane
Energy Density of Gasoline
Value of Gasoline
Value of Controlled/Prevented
Evap. Emissions (Butane)
Units Estimate
10 0% Butane
lb/gal 4.77
Btu/lb 19,500
Btu/gal 93,100
Btu/lb 18,500
Btu/gal 114,000
$/gal 0.90 ($25/bbl crude)
$/gal 0.82 ($20/bbl crude)
$/gal 0.73 ($15/bbl crude)
$/ton 290.32 ($20/bbl crude)
-------
5-63
Table 5-21
NIPER Survey Results[22]: Summer Gasoline RVP Trends by Region*
Region
Years
1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984
% Increase Avg ASTM
1985 over 11 yrs. Limit
iicast
9.8
10.0
10.2
10.7
10.5
10.5
10.8
10.8
10.9
10.5
10.7
11.2
14
11.5
Atlantic Coast
9.3
10.1
10.1
10.4
10.1
10.2
10.3
10.6
10.8
10.7
10.8
11.1
19
11.5
li._ast
9.6
9.6
9.7
9.5
9.4
9.6
9.8
9.9
10.1
10.3
10.2
10.6
10
11.0
. iachian
10.5
10.6
10.5
10.4
10.1
10.6
10.5
10.9
11.1
11.4
11.1
11.5
10
11.5
i. i -jan
10.5
10.4
10.6
11.0
11.2
10.9
11.3
10.9
11.2
11.6
11.5
11.8
12
11.5
i h<=rn Illinois
9.5
10.5
10.3
11.0
10.8
10.9
10.9
11.1
10.8
11.7
11.3
11.2
18
11.5
ial Mississippi
10.1
10.1
9.9
9.9
10.1
10.4
10.3
9.7
10.5
11.0
10. 9
11.1
10
11.0
i Mississippi
9.4
9.6
9.6
9.5
9.8
9.5
9.7
9.3
10.1
10.1
10.0
10.5
12
10.5
i hi-rn Plains
—
9.6
9.9
—
—
9.2
9.8
—
11.0
11.0
10.5
11.3
18
10.5
1ial Plains
—
9.1
9.2
9.0
8.8
9.2
9.2
—
10.2
10.0
10.0
10.3
13
10.0
hc-rn Plains
9.2
9.3
9.2
9.3
9.1
9.5
9.2
9.7
10.0
9.8
9.8
10.1
10
10.0
' tv.rn Texas
9.1
9.5
9.4
9.6
9.5
9.4
9.2
9.4
10.3
10.2
10.3
10.1
11
9.5
hurn Mountain
8.4
8.9
8.7
8.8
8.9
8.7
8.9
8.4
8.8
9.1
8.9
9.4
12
9.5
I horn Mountain
8.9
10.1
9.5
9.9
9.9
9.6
9.5
9.2
10.4
10.4
9.7
10.4
17
10.0
i fic Northwest
9.5
9.9
10.6
10.4
10.0
10.3
10.8
11.0
10.8
11.2
10. 8
11.4
20
11.0
Lonal Average**
9.5
9.8
9.8
10.0
9.9
9.2
10.0
10.1
10.5
10.6
10.4
10.8
14
10.7
luding California)
Unleaded regular gasoline only (R + M/2 less than 90).
Calculated as a straight arithmetic average of the 15 regional averages listed.
-------
Figure 5-6
NIPER Survey Ruffians and
ASTW's July Volatility Classes
A " 9.0 pel
B - 10.0 pel
C " 11.5 pel
LFl
I
CTl
-------
5-65
A. Studies Performed on the Subject of Driveability
Vehicle driveability as discussed here is a function of
vehicle design, fuel characteristics, ambient conditions, and
characteristics of operation. Thus, there is a broad range of
possible situations during which a vehicle will experience
driveability problems. For our purposes these many different
situations are reduced to some extent by the fact that
volatility will only be controlled during the summer months, so
cold temperature driveability problems are of less concern.
But of interest is the entire range of vehicles on the road
today. The fact that current model year vehicles may be able
to run on high volatility fuel does not allow volatility to be
increased to a level where past model year vehicles encounter
difficulties.
One study on this subject performed by the Coordinated
Research Council (CRC)[23] consisted of tests using incremental
RVP blends of two fuels on twelve 1971 model year vehicles.
This testing was conducted at 95°F ambient temperatures and
relatively low altitude. The results show that on the average
the vehicles began to experience vapor lock limited performance
at approximately 9.5 psi RVP. In addition, the average RVP at
which the vehicles began to experience significant driveaway
driveability demerit levels (Demerits are given for any
driveability problems such as stumble, hesitation, or surge
during vehicle operation by trained observers. The amount of
demerits given is consistent with the magnitude of the
problem.) was 9.8 psi. Vehicle to vehicle variability was very
high as would be expected with vehicles of different design,
but the general trend holds true, that driveability worsened at
increasing volatility levels.
In a second CRC study[24] more recent vehicles were used
to once again determine driveability and vapor lock limited
performance. This time 47 1977-81 model year vehicles were
tested, and 32 of these experienced vapor lock limited
performance over the range of fuel volatilities tested at the
100°F ambient test condition. The average fuel RVP at which
these vehicles experienced vapor lock symptoms was
approximately 9.35 psi, and of those which did not encounter
vapor lock, the average maximum fuel RVP level tested was 11.7
psi.
As another measure of driveability, a hot start and
driveaway evaluation of driveability on the 47 vehicles
adjusted for rater severity and for Tv/L z zo adjusted for an
ambient temperature of 100°F was performed. This showed that
the total weighted demerits (TWD) associated with driveability
problems increased using a geometric mean from 24.4 TWD with
the low 7.3 RVP fuel to 38.9 TWD with the midrange 10.6 RVP
fuel to 53.8 TWD with the high 14.2 RVP fuel.
-------
5-66
In still another measure of driveability in this study,
the customer responses for acceptability and overall
performance of their vehicle with the various fuels provided
showed that for fuels with RVP greater than 13.7, 23 percent of
the total number of days driven were reported as being
unacceptable. This total drops to 9.9 percent for fuels with
approximately an 11.5 RVP, 6.2 percent at 10.0 psi RVP, 3.2
percent at 8 psi RVP, and down to zero at approximately 7.0 psi
RVP. However, at 6.2 psi the percentage of days reported as
unacceptable increases to 16.7 percent probably due to cold
start problems in early morning. Ambient information
throughout the period of these evaluations showed an average
daily low temperature of 83.4°F and an average daily high
temperature of 106°F. These averages correspond to a
temperature range of 70°F to 114°F during the months of June
through August in Phoenix, Arizona.
Data on vapor lock limited performance from the two CRC
studies just discussed as well as two other CRC studies are
shown in Figure 5-6. It shows that for 1971, 1975, 1977-81,
and 1982 model year vehicles to achieve a 90 percent customer
driveability performance satisfaction level, fuel volatility
levels of 5.0, 7.1, 7.6, and 10.0 psi RVP are required
respectively. Thus, these data seem to support the contention
that as fuel volatility has been increasing, the ability of
vehicles to perform on higher volatility fuels has also been
increasing. But in the last few years the fuel volatility
seems to have surpassed that for which all vehicles can
function properly.
Fuel-injected (FI) vehicles seem to be better able to
handle higher volatility fuels whereas carbureted vehicles have
more difficulty. This can be seen in data collected from a
study of three carbureted and three FI 1983-4 vehicles.[25]
Testing conducted at 80 to 90 °F showed that the carbureted
vehicles experience approximately twice as many driveability
demerits as the FI vehicles. In addition, the carbureted
vehicles were much more sensitive to fuel volatility than the
FI vehicles. The driveability demerits for the FI vehicles
remained relatively constant between ASTM class D (13.6 RVP
avg) and class C (11.6 RVP avg) fuels tested whereas the
demerits for the carbureted vehicle increased on the average by
50 percent.
One explanation for the poorer performance of carbureted
vehicles over and above that of FI vehicles is the emergence of
an additional driveability problem, this being fuel foaming in
the carburetor. Fuel foaming occurs when foam forms in the
carburetor bowl causing the float to fall below its normal
level and allowing fuel to continue to flow. Although the
symptoms such as stumble, surge, poor idling and acceleration
and even stalling are the same as for vapor lock, they are
caused by ennchening of the A/F mixture rather than the
leaning of it. In addition, one of the best solutions for
-------
5-67
vapor lock; putting an electric fuel pump in the fuel tank
which pressurizes the entire fuel system; actually causes more
fuel foaming in carbureted vehicles. One study[26] done on
just a single 1981 carbureted vehicle showed that ' at 68°F
ambient temperatures, a fuel with an RVP of 17.5 psi was
required before fuel foaming became significant enough to limit
vehicle performance. However, at 86°F the limiting RVP was
reduced to 13.5 psi, and at 104°F ambient temperatures, a fuel
with an RVP of 11.4 psi provided critical performance. Thus,
based on this data alone, it is possible that vehicles on the
road at today's fuel volatilities could experience fuel foaming
limited performance.
All of the information seen thus far seems to suggest that
the current level of fuel volatility is too high for the broad
range of vehicle designs on the market to operate acceptably.
This in turn suggests that there may be some benefit in
reducing the volatility of the fuel from a driveability point
of view. However, care must be taken not to decrease the
volatility too far, for just as there are driveability problems
with fuel of too high a volatility, there has also been data
shown to suggest driveability problems even in the Desert
Southwest when volatility falls to the extremes of 6.2 psi RVP.
For the most part, the level of volatility control
proposed in this study will not interfere with cold temperature
driveability. Only May through September volatility control is
being proposed, so in most cases the temperatures and proposed
volatility levels are high enough to circumvent any problems.
Under the 9 psi Class C RVP scenario, a 22 percent reduction
from current ASTM classifications results in a fuel RVP of 7.0,
8.0 and 9.0 psi for Class A, B and C areas, respectively. The
fact that this volatility level is adequate for driveability is
substantiated by the fact that California, which has had
volatility control to 9.0 psi during the months of April
through October in effect since 1971, has not had any low
temperature driveability problems recorded.[27] This despite
the fact that the South Coast air basin is ASTM Class C in
April and B/C in October. Even with the 9 lb. control, a check
of fuel in in-use vehicles found that the RVP of the gasoline
is typically 8.5 psi, and often as low as 8.0 psi, though
weathering may account for some of the difference. In
addition, often vehicles fueled on this low volatility fuel
travel into the mountains where 10th percentile minimum
temperatures are in the low to mid 30's. These temperatures
correspond to the lowest of the 10th percentile minimum
temperatures for the majority of the nation during the summer
months. Therefore, reducing the nations fuel volatility by 22
percent would not appear to cause any problems based on the
real world information from California.
The 8 psi Class C RVP scenario, in which there are fuel
RVP reductions of 30 percent, would reduce fuel RVP in Class A
-------
5-68
areas to 6.3 psi. Control to this extreme as mentioned earlier
with respect to a CRC study [24] may be too low for some
vehicles to operate satisfactorily.
Another study presented in the comments by API[28] also
suggests that some concern may be warranted with 30, or
possibly even 22 percent control. In this study Chevron tested
8 1973-76 model year vehicles with fuels of 8.5 and 6.5 RVP at
75 °F and 55°F ambient temperatures. 55°F was chosen to be
representative of typical morning temperatures m the LA Basin
based on April through October data of average daily lows.
Results of the testing indicate that due to the reduction in
volatility from 8.5 psi to 6.5 psi, driveability demerits
increased by 123 percent and 252 percent at 75°F and 55°F,
respectively. Also, driveability demerits increased by 19 and
87 percent at 8.5 psi and 6.5 psi, respectively, due to the
reduction in temperature from 75°F to 55°F. Thus, this study
although on older vehicles suggests that the 30 percent and
possibly 22 percent reduction in volatility is too severe,
especially as it relates to class A regions.
Another study by CRC on customer evaluations of cold
temperature driveability found similar, but not quite so
drastic results.[29] They tested 107 1973-78 vehicles with
fuels of 7.6, 10.1, and 12.7 psi RVP in San Antonio Texas from
January 23 to April 3. The 10th percentile minimum temperature
for this period in San Antonio is approximately 35°F
corresponding to the minimum for the B/A ASTM classification
regions in May (see Figure 5-7) or the Class B regions in June
(see Figure 3). The results indicated that although the 10.1
psi fuel achieved nearly the same level of satisfaction as the
12.7 psi fuel (87 and 93 percent respectively) the 7.6 psi fuel
merely achieved a 62 percent level of satisfaction. This
suggests that based on temperature and gasoline volatility data
for the country, [30] as seen in Figures 5-7 through 5-9, even
with the 22 percent RVP reduction (7.8 psi in Class B), there
are certain areas of the country in early May, early June, or
late September such as the high mountain areas of the west, or
extreme northern plain states where some vehicles may encounter
cold temperature driveability problems.
One must be careful, however, when drawing any definite
conclusions from these studies since many of the areas with the
lowest temperatures are also the areas with the highest
altitude. The required fuel volatility for starting and
operating a vehicle under cold conditions is dependent upon
achieving a minimum combustible air-fuel ratio. As one
increases in altitude from sea level to 13,000 feet, the air
density decreases by 2.9 to 2.2 percent per thousand feet.
Thus, it becomes easier and easier to achieve this air-fuel
ratio. This in turn means that as one increases in altitude, a
lower and lower volatility fuel (at constant temperature) is
acceptable for cold temperature operation. Since the altitude
-------
Figure 5-7
Nationwide 10th Percentile Minimum "Danpuratures
for May with Corresponding ASTM Classifications
40
-------
Fidfure 5-8
Nationwide 10th Percentile Minimum Tenperatures
For June with Corresponding ASTM ClassificaLions
60 65
U1
i
o
-------
Fiqure 5-9
Nationwide 10th Percentile Minimum TempernLures
For September with Corresponding ASTM Classifications
en
i
-------
5-72
of San Antonio, Texas is only 701 feet, the driveability
results with the 10.1 psi fuel tested would correspond with
those from testing an 8.8 psi fuel in Denver, Colorado or a 9.4
psi fuel in Great Falls, Montana. The results with the 7.6 psi
fuel in turn, would be similar to those from testing a 6.6 psi
fuel in Denver, or a 7.0 psi fuel in Great Falls. Thus,
although the first glance might suggest the contrary, from a
closer look, a 22 percent reduction from current ASTM may not
cause any significant driveability problems m the United
States. In addition, all of the data available on cold
temperature low volatility driveability was for vehicles of
1978 or earlier model years Thus, the improvements in recent
years for vehicles to operate better on a wider range of A/F
ratios is not expressed in any of the information with the
exception of the comments from California which suggest no
concern may be warranted for a fuel with a 22 percent reduction
in volatility from current ASTM classifications.
As one final point, although the 10th percentile minimum
temperatures for these regions may be low enough to be of
concern, lacking more recent and extensive information, the
same regions have 90th percentile maximum temperatures during
the same periods which become of concern under high temperature
driveability situations (see Figures 5-10 through 5-12). Thus,
although there may be some possibility of low temperature
driveability problems during the volatility control season, the
reduction in high temperature driveability problems should far
outweigh them.
B. Industry Data on Current Volatility Related Problems
In their comments to us, GM and Chrysler provided
information on vehicles which have experienced warranty
problems due to fuel volatility increases. Ford, although
insisting that they too were experiencing many of the same
problems and costs associated with fuel volatility, declined to
provide any information. Nissan was the only other
manufacturer to make any comment on this topic, and although
not providing any information on warranty problems, did say
that rising volatility had forced them to adapt their vehicles
to avoid driveability problems by designing them with injector
blowers, fuel temperature sensors, and fuel pressure control
systems.[31]
In the case of GM, there are three different vehicle lines
of 1982 and 1985 model year which experience vapor lock
problems during hot summer conditions throughout the
country.[32] GM estimates that, of the 570,000 of these
vehicles built, more than 10,000 annually experience this
problem nationwide. The simplest and most effective solution
is to install a small fan for improved cooling of the
carburetor. Another effective solution, but one which is
considerably more involved, is to replace the existing fuel
pump with an improved electric, m-tank fuel pump.
-------
Figuire 5-10
Nationwide 90th Percentile Maximum *I\3nperat.uros
For May with Corresponding AS1M Ciassi fications
-------
Figure 5-11
Nationwide 90th Peroentile Maximum Tenperatures
For June with Correspc*iding ASTM Classifications
-------
Figure 5-12
Nationwide 90th Percentile Maxurum Temperatures
For Sep tenter with Corresponding ASTM Classifications
-------
5-76
In addition to those three vehicle lines, there are an
additional 10 1982-5 model year vehicle lines which experience
vapor lock problems under high altitude conditions. In this
case GM estimates approximately 1500 vehicles annually
experience this problem, and for these situations, the more
tedious and costly fuel pump replacement seems to be the best
solution.
One additional engine produced by GM has also been
experiencing increased complaints. Unlike all the other
vehicles which were carbureted, this is a 1981 6.01
throttle-body fuel-injected vehicle (TBI). The solution was to
insulate the fuel lines, and to replace the gasket under the
TBI unit with one with lower thermal conductivity.
Chrysler, as opposed to GM, seems to have a larger problem
with driveability at today's volatility levels. They have at
least 38 vehicle/engine combinations (25 different vehicle
lines 13 of which have two different engine sizes) of 1981-1986
model year for which they have published Technical Service
Bulletins.[33 J Chrysler sites fuel foaming, vapor lock, or
both as the primary source of the problem in all instances.
The most common solution to the problem is the installation of
a new in-tank electric fuel pump, a pressure regulator control
module and a fuel reservoir filter. In some instances an
electric fan placed on the engine to cool the carburetor after
the engine is turned off provided an acceptable solution.
Unfortunately Chysler did not state a total vehicle population
which has or is projected to be affected by these problems.
C. Costs Associated With High Volatility Driveability
Problems
In order to accurately quantify the cost associated with
high volatility driveability problems, there are many different
second order costs to be taken into account due mainly to the
costs incurred when vehicles breakdown. Unfortunately most of
these costs are not easily quantifiable. However, two
first-order costs to the manufacturer; warranty costs, and
required vehicle design modification costs, can be quantified
to some extent.
In the case of Chrysler, they have estimated that through
approximately February 1986 they have paid out over $14 million
in volatility-related warranty costs. These costs have been
incurred mainly since 1981 when volatility began to reach
levels of concern in some areas of the country. In addition,
the only available information suggests that only 1981 and
later Chrysler vehicles have routinely received warranty fixes.
The most common warranty fix in the case of Chrysler is
the replacement of the current fuel pump with an m-tank
electric fuel pump, and also the installation of a pressure
-------
5-77
regulator in the fuel line. These pacts and associated
hardware as well as labor, cost Chrysler an estimated $400 per
warranty fix. Thus, over the five year period since 1981,
Chrysler has had approximately 7000 vehicles per year undergo
such a warranty fix. When the warranty costs are averaged out
over Chrysler's total U.S. new vehicle registration of
4,311,918 from 1981 through 1985, the average cost to Chrysler
is $3.25 per vehicle registered in the U.S. due to volatility
related problems.
In the case of GM, they estimate that approximately 11,500
vehicles per year undergo volatility related warranty fixes
Of these only approximately 1500 undergo the expensive fuel
pump replacement whereas the other 10,000 undergo the less
expensive installation of an after-run fan. Due to this, the
estimated cost to GM per vehicle fix under warranty has only
been approximately $130. This adds up to a cost to GM of
approximately $7.5 million over the 1981-1985 period. This in
turn breaks down to a per U.S. registered GM vehicle cost of
approximately 40£ This is much smaller than the cost to
Chrysler due mainly to a much smaller fraction of vehicles with
volatility related problems
If we combine the warranty costs and total vehicle
registration of both GM and Chrysler, the average cost per
vehicle is 88£. Since no other data was presented by industry,
applying this cost to the total U.S. new vehicle registration
over the last five years, the total warranty costs add up to
$40.5 million, or an average cost per year of $8.1 million
Chrysler was also the only manufacturer to provide any
information on costs associated with vehicle design
modification to correct problems associated with fuel
volatility. Chrysler estimated that up until about February of
1986 they have spent upwards of $58 million m design changes.
If once again we can assume that these costs have been incurred
over the 1981 through 1985 model years, the average cost for
such vehicle modifications is $13.5 per vehicle, which in turn
adds up to a total nationwide cost of $124.1 million per year.
Although it would be impossible to recover all of these costs
if fuel volatility were reduced, Chrysler did state that a fuel
temperature measurement device, and a fuel tank pressure relief
valve could be removed from the vehicle at a retail cost
savings of $11.29 per vehicle. Based on a contracted cost
study conducted for EPA, the retail cost of these parts on a
production vehicle, corrected for overhead and inflation, is
approximately $4.76 per vehicle.[34] This is actually higher
than might have been predicted by using a rule of thumb markup
factor of 4.5 from factory retail costs to after market retail
costs. If averaged over the total U.S. new vehicle
registration, this becomes an annual cost of $44 million to the
manufacturers. Additional savings could likely be realized as
at lower volatility manufacturers would once again have greater
flexibility in the design and manufacturing of the fuel systems.
-------
5-78
The driveability cost estimates cited above are not based
on hard evidence, but rather on the limited information
provided by GM and Chrysler. Although the analysis was
qualitatively believable, there exists no verifiable basis for
relying on any precise estimates.
In addition to the direct cost savings presented, there
are also many "intangible costs" associated with driveability
problems which although difficult to quantify are
never-the-less real. These costs include direct costs to the
consumer for the reduction in fuel economy due to poor
driveability, as well as indirect costs to society associated
with the increase in emissions from poorly operating vehicles.
In addition to these, Chrysler also mentioned some "intangible"
costs associated with driveability problems due to high fuel
volati1ity:
1) The costs of "substitute transportation" in the form
of a towing fee and car rental and/or taxi costs when the
vehicle stalls due to vapor lock or fuel foaming;
2) Payments to repair garages for attempted fixes when
the problem is improperly diagnosed;
3) Higher prices on new cars due to manufacturer
warranty costs on existing vehicles;
4) Costs associated with motorist insurance and auto
club fees in case of vehicle breakdown;
5) Costs associated with the willingness of many people
to pay from it to 30 extra per gallon for fuel which they know
will give better performance;
6) Costs to motorists through taxes to support public
agencies which must cope with vehicles which breakdown; and
7) Litigation costs involving product quality.
These "intangible" costs, although legitimate costs, are
not all mutually exclusive of each other. Often "substitute
transportation" is paid for by motorist's insurance and auto
club memberships, and many of the payments to repair garages
for improper fixes are covered under warranty costs. In
addition, by paying to 31 extra per gallon of fuel, those
consumers avoid recurrence of driveability problems and the
further costs associated with them. Thus, this cost may not
really be intangible, but rather an alternative way to
establish the total societal cost associated with driveability.
There was no actual data shown by Chrysler in support of
the 1<£ to 3$ price range. However, it seems reasonable that
people who are experiencing driveability problems would be
-------
5-79
willing to pay extra per gallon for fuel on which their vehicle
will perform better. AAA in their testimony at the hearing on
the 1985 volatility study stated that people will willingly pay
It to extra per gallon. Chrysler appears to accept this
testimony by using it in their comments. It also seems
reasonable in light of the broad range of fuel prices at the
retailers today which vary by as much as 10 cents per gallon,
that to 30 extra per gallon may be an accurate if not low
estimate. If people whose vehicles are experiencing
dnveability problems are willing to spend extra per gallon of
fuel, then combining this with the information shown in Figure
5-13 on the satisfaction of vehicle operation with fuels of
varying volatility, a cost to the consumer can be determined.
CRC stated that for an increase in the ambient temperature,
there is a corresponding decrease in the temperature at which
the vapor to liguid ratio eguals 20. Therefore, the
satisfaction of driveability of vehicles at a given ambient
temperature can be determined by raising or lowering the
vertical axis of Figure 5-13. Using Figure 5-13 and available
ambient temperature data, the number of vehicles, and the age
of those vehicles in a given area should enable the
determination of the number of vehicles in that area which have
unsatisfactory performance due to fuel volatility. Knowing the
fuel consumption of these vehicles and the amount extra per
gallon they are willing to pay now provides the total cost
associated with driveability.
For the calculation, the temperature data for the summer
months of April through October for the ten largest
(non-California) ozone non-attainment cities in the U.S. (see
Table 5-22) were chosen since it was readily available and
fairly representative of the nation as a whole. The vehicle
population for these cities was assumed to be proportional to
population (see Table 5-22). The age of the vehicles was
proportioned according to the MOBILE3 VMT mix for vehicles of
various model years as seen in Table 5-23. As Figure 5-13 does
not show a curve for each model year (MY), vehicles of 1971
through 1974 MY were assumed to correspond to the line labeled
1971 model cars, 1975-6 MY vehicles to the line labeled 1975,
and all vehicles of 1982 and later MY to correspond to the line
labeled 1982 model cars. Once the percentage of vehicles
dissatisfied for these 10 cities was determined, the results
were applied to the national fuel consumption data. Using the
range of 1$ to per gallon that the people with driveability
problems are willing to pay yields the cost savings associated
with volatility control. These cost savings as seen in Table
5-24 are as high as $78.0 million dollars per year depending on
the level of volatility control, the calendar year, and the
amount extra per gallon that people are willing to pay.
Although certain assumptions such as lumping all of the
1982 and later vehicles into one group may tend to
over-estimate these costs associated with driveability, the use
-------
5-80
FIGURE 5-13
distribution of vapor lock limiting tv/l,20
AT 100°F AMBIENT IN FOUR CRC PROGRAMS[ ]
Percent of Cars Satisfied
-------
5-81
Table 5-22
City Specific Information
Average Daily High Temperature
i ty
Population
Vehicle Population*
Apr
My
JE
JY
Aug
Sept
Oct
w York
17,190,221
9,836,528
60.7
71.4
80.5
85.2
83.4
76.8
66.8
i. ] cago
7,794,851
4,460,342
59.4
69.7
79.1
83.1
82.3
75.4
65.5
11 ladelphia
5,547,902
3,174,601
63.5
74.1
83.0
86.8
84.8
78.4
67.9
._lroit
4,488,072
2,568,149
57.6
68.5
79.1
83.1
81.6
74.2
63.4
• r.ton
4,303,311
2,462,426
56.3
67.1
76.6
81.4
79.3
72.2
63.2
. -.hington, D.C.
3,250,822
1,860,174
67.1
76.6
84.6
88.2
86.6
80.2
69.8
i -inn
3,220,844
1,843,020
82.7
85.3
88.0
89.1
89.9
88.3
84.6
. 'ust on
2,735,766
1,565,450
79.4
85.9
91.3
93.8
94.3
90.1
83.5
t tsburgh
2,218,870
1,269.674
60.9
70.8
79.5
82.5
80.9
74.9
63.9
• .nnecticut
2,206,318
1,262,492
58.9
70.3
79.5
84.1
81.9
74.5
64.3
Met ro
Vehicle population estimated to be 57.222 percent of population based on national average.
-------
5-82
Table 5-23
VMT Mix of Vehicle Fleet*
Age of Vehicle % VMT (as of January)
0 .036
1 .137
2 . 122
3 . 109
4 . 097
5 . 085
6 . 075
7 . 064
8 . 056
9 . 048
10 .041
11 .041
12 .034
13 .027
14 .021
15 .016
16 .011
Based on MOBILE3 information.
-------
5-83
Table 5-24
Drn/ability Cost In Millions*
Fuel RVP
1986
1988
1990**
1992**
1994**
1996**
1998**
2000**
2010
11.5
0
0
0
0
0
0
0
0
0
11.0**
6 . 2
4 7
3.7
3.0
2 . 5
2.2
1. 8
1.7
1.3
10 5
1 L. 5
8 ?
6 . 6
5.5
4 . 6
4.0
3 5
3 . 2
2 . 1
10.0**
15.3
11.0
9.1
7 . 7
6.6
5.8
5.0
4.4
2.6
9.5**
18.7
13. 2
11.0
9.4
8.2
7 . 1
6 . 2
5.5
3.0
9.0
21.5
15.0
12.4
10.6
9 . 3
8.1
7 .1
6 . 3
3 . 2
8.5**
24 . 0
16 . 4
13. 5
11. 5
10.0
8.7
7.6
6.7
3 . 3
8 .0**
26.0
17 . 7
14 . 5
12 . 2
10 . 5
9.1
7 . 9
6.9
3 . 3
Cost assuming people willing to pay le/gal. extra for better performance fuel.
Based on a curve fit.
-------
5-84
of average monthly high temperatures rather than daily high
temperatures may severely under-estimate the associated costs.
Also, this analysis was done assuming that people are only
paying more for fuel at those times when conditions require
it. In reality, these individuals will not know in advance
when they will need the better fuel. As a result, they will be
paying the extra cost per gallon for a much longer period than
we were able to take into account. In addition, these costs
savings for volatility control are calculated assuming current
fuel volatility is at current ASTM standards. As has been
discussed earlier, the national average is already above the
ASTM limits, and if it follows the current trends, it could be
at much higher levels by the year 2010. Thus, the cost savings
in future years to the consumer for volatility control due to
driveability would probably also be much higher.
The warranty and vehicle design change costs based on the
manufacturer information are not verifiable and required many
assumptions. The costs based on the \
-------
5-85
could not be verified. An independent method of determining
the value to society of improved fuel quality is to use the
costs to the consumers due to their willingness to pay an extra
10 to 3(t per gallon extra for better performing fuel. In
comparison to other cost estimates seems relatively
reasonable. In addition, it is easily broken down by fuel RVP,
and can be predicted into the future. The majority of costs
based on manufacturer information are costs m the past for
design changes. The majority of these may no longer apply to
future costs as their effect has already been made. Therefore,
although there are many other costs associated with
driveability problems due to high volatility, those based on
higher fuel costs are the only ones which will receive any
further consideration in this study.
VII. Enforcement Costs
The Agency is considering
compliance with gasoline RVP
refiners and importers, 2) in
3) a combination of both.
three overall schemes to monitor
controls: 1) self-reporting by
-field sampling and testing, and
Under the first option, refiners and importers would be
required to test the volatility of fuel and report it to EPA.
EPA would prescribe sampling and testing methodologies
Regulated parties would be required to submit periodic reports
to EPA and to maintain any records necessary to substantiate
their reports. EPA would perform a limited number of audits of
regulated parties' records and testing capability to monitor
compliance, in addition to reviewing the periodic reports.
Under the second options, the regulatory standard would
apply to fuel in the field. Inspectors would take samples at
one or more types of fuel distribution facilities according to
a prescribed sampling methodology. If an acceptable RVP field
screening test were available, it would be performed by the
inspector to determine which samples are likely to be in
violation of the applicable standard, and only these samples
would be shipped to a testing laboratory. If such a screening
test were not available, and samples would be tested at the
laboratory according to a prescribed methodology. Regulated
parties might be required to maintain records to facilitate the
tracing of violating samples through the distribution network.
The third option would include both self-reporting and
in-field sampling and testing.
A full report investigating the cost of the options above
is currently being developed ai.d will be placed in the docket
when complete. Preliminary estimates of the costs of different
enforcement scenarios based on the above three options are
presented in Table 5-25. Option one involves self-reporting by
refiners and importers, options two through six involve
-------
5-86
Table 5-25
Preliminary RVP Enforcement Cost Estimates
Cost
Scenario (1000 Dollars)
I. Self-Reporting 295-477
II In-Field Sampling and Testing 844-2,093
(10,000 Retail Inspections)
III. In-Field Sampling and Testing 1,465-2,182
(8,000 Retail and 2,000 Terminal
Inspections)
IV. In-Field Sampling and Testing 1,526-2,289
(7,000 Retail, 2,000 Terminal,
and 1,000 Refinery Inspections)
V. In-Field Sampling and Testing—Inspection 1,347-1,811
Scheme same as in IV but
with Field Screening Test Kit
VI. In-Field Sampling and Testing 1,081-1,811
(8,500 Retail Inspections with
States doing 1,500 Retail
Inspections themselves)
VII. Self-Reporting and In-Field Sampling and 1,356-1,956
Testing with 50% of the Inspection
Scheme in IV
-------
5-87
in-field sampling and testing scenarios, and option seven
involves a combination of self-reporting and in-field sampling
and testing. These estimates are based on a year-round basis.
With a 5-month standard some estimates might change. For
example, additional testing equipment and manpower would be
needed to do 10,000 tests in 5 months; also, a contractor might
charge more to handle such an intensive inspection scheme. EPA
inspector time, travel cost, sampling materials, shipping costs
and enforcement resources would remain the same.
The enforcement costs of Table 5-25 are small relative to
the other cost of the RVP control program. Even the most
expensive option is only approximately two million dollars
(including manpower costs) compared to a refinery cost of 490
million dollars for the 9.0 psi control options (total refinery
cost to be derived in Chapter 6). For this reason the
preliminary cost of enforcement were not included in the cost
effectiveness calculations of Chapter 6. The costs will,
however, be incorporated when they become finalized. Even if
the enforcement cost were included they would not effect the
incremental cost effectiveness since the enforcement cost are
independent of the RVP level chosen an hence equal at each RVP
control step.
-------
5-88
References (Chapter 5)
1. "Estimated Refining Cost Impact of Reduced Gasoline
Vapor Pressure," Final Report, Bonner and Moore Management
Science under SwRI Work Assignment #28 of EPA Contract No.
68-03-3162, July 1985.
2. "Supplemental Report to Estimated Refining Cost
Impact of Reduced Gasoline Vapor Pressure," Bonner and Moore,
February 25, 1986.
3. "Cost and Feasibility of Gasoline Volatility
Reductions," Sobotka and Company, Inc., for EPA Office of
Policy Analysis, April 21, 1986.
4. "Refining Cost for Reducing Gasoline Vapor
Pressure," Turner, Mason and Company, for Motor Vehicle
Manufacturers Association, January 17, 1986.
5. "Gas Processors Association: Comments on Gasoline
Volatility and Hydrocarbon Emissions from Motor Vehicles,"
February 27, 1986. (Public Docket No. A-85-21, Entry II-D-40.)
6. "Impact Assessment of Reducing Gasoline Volatility,"
Bonner and Moore Management Science, for California Air
Resources Board under Contract No. A2-051-32, November 30, 1983.
7. "The Butane Industry: An Overview and Analysis of
the Effects of Gasoline Volatility Control on Prices and
Demand," Draft Report, Jack Faucett Associates, Work Assignment
#5, EPA Contract No. 68-03-3244, May 30, 1985.
8. Harry Weaver, Director, Environmental Department,
Motor Vehicle Manufacturers Association, letter to Charles
Gray, Director, ECTD, OMS, EPA, December 4, 1984.
9. Ronald L. Jones, American Petroleum Institute,
letter to Charles L. Gray, Jr., Director ECTD, OMS, EPA,
December 17, 1984.
10. W.S. Freas, Manager, Emission and Fuel Economy
Operations, General Motors, letter to R.E. Maxwell, Director,
Cert. Division, OMS, EPA, August 15, 1984.
11. J.C. Ingomells, Chevron Research Company, "Fuel
Economy and Cold-Start Driveability with Same Recent Model
Cars", SAE Paper No. 740522.
12. T.M Fisher, Director, Automotive Emission Control,
General Motors Corporation, letter to EPA, Central Docket
Section (LE-131), Docket A-83-44, January 22, 1985. (See public
docket # A-83-44).
-------
5-89
13. D.R. Buist, Director, Automotive Emissions and Fuel
Economy Office, Environmental and Safety Engineering Staff,
Ford, letter to Richard D. Wilson, Director, OMS, EPA, January
22, 1985. (Available in public docket # A-83-44).
14. Fred W. Bowditch, Motor Vehicle Manufacturers
Association, letter to Richard D. Wilson, OMS, EPA, January 22,
1985. (Available in public docket # A-83-44).
15. D.R. Buist, Director, Automotive Emissions and Fuel
Economy Office, Environmental and Safety Engineering Staff,
Ford, letter to Richard D. Wilson, Director, OMS, EPA, March
27, 1986.
16. R.S. Sickler and S.A. Pezda, Ford Motor Company,
"The Effect of Increasing Fuel Volatility on Exhaust Emissions"
SAE Paper No. 860533.
17. American Petroleum Institute, "Comments on Gasoline
Volatility and Vehicle Hydrocarbon Emissions," letter to EPA,
March 27, 1986.
18. Charles F. Taylor, "The Internal Combustion Engine
in Theory and in Practice," MIT Press, Vol l, 1980.
19. Fred Hendron, "Propane Power for Light-Duty
Vehicles: An Overview" SAE Paper No. 830383.
20. From Department of Energy, Monthly Energy Review,
March 1985, as published in the National Petroleum News 1985
Factbook, p. 101.
21. O. Pinkus and D. F. Wilcock, "Strategy for Energy
conservation Through Tribology," American Society of Mechanical
Engineers, New York, 1978.
22. Cheryl L. Dickson and Paul W. Woodward, "Motor
Gasolines, Summer 1985," National Institute for Petroleum and
Energy Research (NIPER) for API, June 1986.
23. "Evaluation of a Temperature Driveability Test
Procedure - 1971 CRC Yuma Program," CRC Report No. 455, June
1973.
24. "Customer Perception of Hot-Weather Driveability in
1977-1981 Passenger Vehicles," CRC Report NO. 543, July 1985.
25. Philip A Yaccarino, "Hot Weather Driveability and
Vapor-Lock Performance with Alcohol-Gasoline Blends," SAE
852117, 1985.
-------
5-90
26. V.M. Tertois and B.D. Caddock, "Carburetor Foaming
and its Influence on the Hot Weather Performance of Motor
Vehicles," SAE 821202, 1982.
27. State of California Air Resources Board, "Comments
on EPA's November 1985 Study of Gasoline Volatility and
Hydrocarbon Emissions from Motor Vehicles," letter to EPA, Feb.
13, 1986.
28. American Petroleum Institute, "Comments on Gasoline
Volatility and Vehicle Hydrocarbon Emissions," letter to EPA,
March 1986.
29. W.C. Williams, D.K. Lawrence, J.E. Robinson, D.B.
Heck, "CRC Investigates Cool-Weather Driveability and Customer
Satisfaction," SAE 801350, 1980.
30. John P. Doner, Coating and Chemical Laboratory, "A
Predictive Study for Defining Limiting Temperatures and their
Application in Petroleum Product Specification," Distributed by
the National Technical Information Service, U.S. Department of
Commerce, November, 1972.
31. Toshio Maeda, Nissan Research and Development, Inc.,
"Comments Concerning Fuel Volatility and Hydrocarbon Emissions
from Motor Vehicles," letter to EPA, March 7, 1986.
32. T.M. Fisher, Director Compliance and Planning,
General Motors Corporation, letter to Charles L. Gray,
Director, Emission Control Technology Division, OMS, EPA,
January 14, 1986.
33. J. W. Furlong, Planning Specialist, Certification
and Regulatory Programs, Chrysler Motors, Letter to Paul
Machiele, Engineer, Emission Control Technology Division, OMS,
EPA, July 14, 1986.
34. "Cost Estimates for Emission Control Related
Components/Systems and Cost Methodology Description,"
EPA-460/3-78-002, NTIS tt PB 279195, March 1978.
-------
CHAPTER 6
Analysis of Alternatives: Gasoline
This chapter draws on the findings presented earlier in
the report and provides a direct comparison of the various HC
control strategies being examined (except for alcohol blend
strategies, which are addressed in Chapter 7) The bulk of the
chapter (Section I) is devoted to the evaluation of RVP
control, which is described further below. Following this, in
Section II, the control of %i60 is evaluated. Two additional
analyses are presented in appendices. Appendix 6A evaluates
changes in certification fuel at constant in-use RVP (i.e.,
should certification fuel be matched to Class A, B or C fuel)
Appendix 6B evaluates the cost effectiveness of inspection and
maintenance programs for evaporative emissions
The main comparison of RVP control options is based on the
estimated costs of motor vehicle-related controls and in-use
fuel controls (presented, respectively, in Chapters 4 and 5)
and on projected emissions benefits (discussed in Chapter 3)
associated with each of the long- and short-term control
scenarios. Using this information, cost effectiveness ($ per
ton) figures were developed as a basis for evaluation The
emission reductions associated with each control scenario,
along with their relative cost effectiveness, will be the focus
of this chapter.
Both long-term control strategies, where in-use RVP and
certification fuel RVP are made equal, and short-term control
strategies, where in-use fuel is temporarily controlled to a
lower RVP level than certification fuel, will be examined in
terms of cost, emission reductions, and resulting cost
effectiveness. The long-term analysis will focus on the year
2010 as a "steady-state" point at which essentially the entire
motor vehicle fleet will have turned over (i e., revised
certification fuel and test procedure will have affected the
design of almost all of the vehicles in the field). The
short-term discussion will examine the years 1988, 1990, 1992,
1995, 1997, and 2000, and will focus on benefits achievable
with additional m-use RVP control over and above those
benefits resulting from the long-term strategies, again on a
calendar year basis (i.e., certification RVP equal to in-use
RVP) .
Part A of Section I outlines the methodologies and
assumptions used to calculate cost effectiveness estimates for
both the long- and short-term control strategies. (The
discussions on the development of costs and emissions
reductions are merely reviews, as the details on both are
presented in Chapters 3 through 5.) Following the methodology
descriptions, Part B presents the results of the analyses,
first for the long-term and then for the short-term strategies
-------
6-2
In Section I, an analysis of alternatives based on best
estimates and current conditions will be presented first. This
"base" case includes control of vehicle refueling losses
(onboard refueling controls are in place) and no inspection and
maintenance (I/M) programs for evaporative emissions. As
discussed earlier in Chapter 2, ozone-related HC control
appears to be most valuable during the summer months, and a
5-month (May-September) control period appeared to be most
appropriate. In the analyses performed below both the control
period for m-use volatility control and the period of
consideration of emission benefits will be five months. This
is addressed m greater detail later in Part B The second
analysis presented in Part B adjusts the cost effectiveness of
RVP control so that it is comparable to that of other ozone
strategies currently being investigated, particularly those
which are year-round and focus on these emission reductions in
ozone non-attainment areas.
The third analysis presented examines the effect of
discounting both costs and emission reductions over a 33 year
period prior to calculating a cost per ton. This methodology
further increases comparability between RVP control values and
those of other ozone programs.
As discussed in Chapter 4, the ratio of percentage change
in fuel economy to percentage change in fuel energy content
(due to butane removal), designated "R", is between 0.82 and
0.95. All cost effectiveness analyses have been conducted
using both ends of this range.
In addition to the base case analysis, Part B will also
present the results of various sensitivity analyses. The first
sensitivity analysis assumes onboard refueling controls are not
in place (i.e., the effect of not implementing onboard vehicle
refueling controls or of implementing them "after" RVP
controls, either chronologically or for accounting purposes).
However, the vehicle-oriented control of excess evaporative HC
emissions due to high in-use RVP, which would under some
scenarios be controlled with the addition of an onboard
canister, is assumed to be controlled via RVP control
(certification or in-use) and not via onboard refueling
control. This was done because the current evaporative
emission excess is not related to refueling and can be
controlled without the use of an onboard refueling canister.
If the onboard system is not assumed to be present, some
control of refueling emissions would be realized with in-use
RVP control and reduce the cost per ton of an in-use RVP
control program.
The second and third sensitivity analyses will examine the
effect of crude oil price on the cost effectiveness. Crude oil
prices of $15 per barrel and $25 per barrel will be
investigated. The sensitivity of refining cost to crude oil
price was described in Chapter 5.
-------
6-3
I. RVP Control
A. Methodology
1. Long-Term Analysis
The methodology used to estimate the long-term
(steady-state) costs, emission reductions, and resulting cost
effectiveness of controlling evaporative HC emissions through
equating m-use and certification gasoline RVPs along with
revisions to the certification test procedure is detailed in
this section. The methodology for calculating the actual cost
effectiveness (total costs over total emission reductions) will
be presented first for all sensitivity cases, followed by that
for adjusting the cost effectiveness of the base case to be
comparable to those of other strategies The HC emission
reductions considered in the former calculation all occur in
the summer and are, thus, not directly comparable to those of
year-round control strategies. The latter calculation adjusts
for this difference
The year 2010 was used to represent the long-term, as the
vehicle fleet would consist entirely of post-1990 (i.e.,
controlled) vehicles by this time.
a. Actual Cost Effectiveness
The 2010 emission reductions associated with the base case
(including exhaust HC benefits, with onboard refueling loss
control and with no evaporative I/M program), and the without
onboard case are estimated in Chapter 3 and were simply taken
from total inventory projections of Tables 3-21 and 3-22. As
discussed in Chapter 3 the emission reductions used m the cost
effectiveness analysis are based on design-value day
temperatures. Net and incremental emission reductions were
estimated for each long-term control scenario (11.5-psi RVP
down to 8.0-psi RVP in 0.5-psi increments).* The sources of
NMHC emission reductions include motor vehicle evaporative and
exhaust emissions, and gasoline storage and distribution
sources (i.e., bulk storage. Stage I, and refueling).
The uncontrolled baseline assumes an m-use. Class C
equivalent RVP of roughly 11.7 psi — it varies from city
to city — and a certification fuel RVP of 9.0 psi.
-------
6-4
The net and incremental emission reductions estimated for
the long-term control scenarios are presented in Table 6-1 for
the base case and the without onboard sensitivity analysis.
The emission reductions are representative of 5-months of
control and are therefore five-twelfth's of the values
presented in Tables 3-21 and 3-22. Emission reductions for the
crude cost sensitivity cases are the same as the base case.
Emission reductions are detailed here in order to provide a
perspective on the relative contribution of various sources —
motor vehicle evaporative losses, motor vehicle exhaust
emissions, refueling, bulk storage, and Stage I. As shown,
motor vehicle evaporative emissions make up the largest
controlled category, accounting for 84 to 88 percent of total
NMHC reductions in 2010 (assuming the base case).
The only difference between the emission reductions of the
without onboard case and the base case is the refueling
emission reduction. In the without onboard case, total
emission reductions are somewhat higher because in-use RVP
control reduces refueling losses if they are not already being
controlled by onboard systems. Since in-use RVP control
affects even controlled refueling losses (by the same
proportions as uncontrolled emissions) and onboard controls
would be subject to some degree of tampering, some control of
refueling emissions occurs even if onboard controls are
implemented. This is demonstrated by the refueling values in
the base case of Table 6-1. Based on previous EPA studies,
onboard controls are at least 97 percent effective; accounting
for the projected tampering incidence (the same as then-current
evaporative control tampering rate), refueling emissions are
expected to be controlled by 93 percent in 2010 (i.e.,
essentially all models in the fleet will be equipped with
onboard controls).[1 ]
The net cost of commercial gasoline and motor
vehicle-related controls are broken down in Table 6-2 and are a
function of several individual components. These include. 1)
the refinery costs of reducing gasoline RVP, 2) the value of
the increased energy content of commercial gasoline, 3) the
value of recovered or prevented evaporative HC losses, 4) the
cost of motor vehicle redesign, 5) the fuel economy penalty
associated with increased vehicle weight due to the enlarged
canister, and 6) the driveability credits. The combination of
these individual components is described below. The detailed
derivation of the individual vehicle and fuel-related control
costs are provided in Chapters 4 and 5, respectively.
The refinery and NGL industry costs of reducing RVP,
relative to crude oil price, are taken from Table 5-3 of
Chapter 5. For this 2010 scenario the "with investment" costs
are most applicable. Nationwide 5-month costs are calculated
-------
d-5
T Case" (Net >:
Z s ) - -
^05
5 >4
5j3
8 7 0 '
• 3 3
u52
1-7 1
Z -s .it -.J
•o 7
l7
j 7
17 >7
) 7
u7
n 7
Ref .\e ± nig
0
1
1
2 3
3
4
4
Bi. 1 % Storage
*>
L.
7
11
15 21
26
31
3 b
St^qe I
2
S
13
I"- 20
2 1
22
23
Tot.i-
57 =
Ol'J
;55
t^-G 7 20
750
7 7-o
aoi
"Base Case" (Incremental)
e; ?j. -ic
2 0
2 5
24
22
24
10
10
L O t HC
3
J
0
0
0
0
Se f ^e i i ng
1
0
1
1
0
1
0
Bui-' Storage
5
4
4
3
5
5
5
Stage I
¦;
5
6
1
1
1
1
Total
41
38
35
30
30
26
25
"Without Onboard" Case (Net):
Evap. HC
505
534
563
587
609
633
652
671
Exhaust HC
67
67
67
67
67
67
67
67
Refue1ing
0
10
19
2 S
36
45
55
64
Bulk Storage
2
7
11
15
21
26
31
36
Stage I
2
8
13
15
20
21
22
23
Total
576
52 6
673
716
753
792
827
862
"Without Onboard" Case (Incremental):
Evap. HC
-
29
29
24
22
24
19
19
Exhaust HC
-
0
0
0
0
0
0
0
Refueling
-
10
9
5
8
9
10
9
Bulx Storage
-
5
4
4
6
5
5
5
Stage I
;
6
5
6
1
1
1
1
Total
-
50
47
43
37
39
35
34
Differences lr total emissions oetveen Table *5-1 and Table 6-3
are a result of round-off in calculating the components of Table
6-1. Vehicle related emission reductions
-------
6-6
Table 6-2
Net Costs Under Long-Term, 5-month
Control Scenarios in 2010 (10 $/y r)
In-Use - Cert. RVP (psi)
Case/Cateqory
11.5
11.0
10.5
10.0
9.5
9.0
8.5
8.0
"Base Case":
- Re fine ry Cost
0
59
135
223
329
447
583
720
- Vehicle Cost
61
4 8
16
25
14
4
-7
-15
- Fuel Econ. Ciedit*
0
-22
-52
-84
-118
-153
-191
-230
- Evap.Recov.Credit
-96
-106
-115
-124
-132
-140
-147
-153
- Dnveability Credit
0
-1
-1
-1
-1
-1
-1
-1
- Weight Penalty
11
8
6
4
2
0
-2
-3
- Total Cost
-24
-14
9
43
94
157
235
316
"Without Onboard" Case:
Refinery Cost
0
59
135
223
329
447
583
720
Vehicle Cost
61
4 fi
36
25
14
4
-7
-15
Fuel Econ. Credit*
0
-22
-52
-84
-118
-153
-191
-230
Evap.Recov.Credit
-96
-108
-120
-132
-143
-153
-163
-173
Dnveability Credit
0
-1
-1
-1
-1
-1
-1
-1
Weight Penalty
U
8
6
4
2
0
-2
-3
Total Cost
-24
-16
4
35
83
144
219
298
"$25 $/bbl Crude Oil" Case:
Refinery Cost
0
71
157
264
396
54 3
706
870
Vehicle Cost
61
48
36
25
14
4
-7
-15
Fuel Econ.Credit*
0
-24
-57
-91
-128
-167
-209
-252
Evap.Recov.Credit
-106
-116
-127
-137
-145
-154
-161
-169
Dnveability Credit
0
-1
-1
-1
-1
-1
-1
-1
Weight Penalty
12
9
7
5
2
0
~2
-4
Total Cost
-33
-13
15
65
138
225
326
429
"15 $/bbl Crude Oil" Case:
Refinery Cost
0
39
102
178
263
358
473
595
Vehicle Cost
61
48
36
25
14
4
-7
-15
Fuel Econ.Credit*
0
-20
-46
-74
-104
-136
-170
-205
Evap.Recov.Credit
-86
-94
-103
-110
-117
-125
-131
-137
Dnveability Credit
0
-1
-1
-1
-1
-1
-1
-1
Weight Penalty
10
7
6
4
2
0
-1
-3
Total Cost
-15
-21
-6
22
57
100
163
234
For R = 0.82; when R = 0. 9 5, fuel economy credit is increased by
16 percent.
-------
6-7
simply by multiplying these refinery costs by nationwide
gasoline consumption (excluding off-highway gasoline
consumption) from EPA's MOBILE3 Fuel Consumption Model (FCM)
for the year 2010.[2] The annual consumption projected by the
MOBILE3 FCM is 74.93 billion gallons (1.784 million barrels)
and must be multiplied by five-twelfth's to represent the
5-month control period. The base case refinery costs assume
$20 per barrel crude oil. This value was also used in the
without onboard and winter-time credit cases. These refinery
costs are summarized in Table 6-2 The refinery costs for two
additional sensitivity cases assuming $15 per barrel and $25
per barrel crude oil are also shown.
Reducing m-use gasoline RVP also increases the energy
density of the gasoline, which, in turn, results in increased
fuel economy in motor vehicles. The effects are summarized in
Table 5-19 of Chapter 5. The credit is estimated oy
multiplying the increases in vehicular fuel economy of Table
5-19 by the nationwide fuel consumption described above and an
estimated value of gasoline of $0.82 per gallon (for the S20
per barrel crude oil case). Gasoline values of $0.73 and $0 90
per gallon were used m the $15 per barrel and $25 per barrel
cases, respectively. A factor of five-twelfths is also
included since the credit only occurs during the control
period. The resulting credit was multiplied by a R factor of
0.82 to relate the fraction of the increased energy that is
fully utilized by the engine. The credits are shown in Table
6-2. For cases where the R-value is 0.95 the credits are
simply increased by 16 percent.
The methodology for valuing recovered/prevented
evaporative HCs from the gasoline storage and distribution
system, vehicle refueling, and motor vehicle operation was
derived in Section V of Chapter 5. There, the value of HC
(butane) control was determined to be $290.32 per ton (for the
$20 barrel per crude oil case). The evaporative and refueling
emission reductions can simply be multiplied by this value to
derive the recovery credits. The emission reductions used to
obtain this credit are based on July-average temperature as
discussed in Chapter 3. These emission reductions are shown on
a 12-month basis in Tables 3-23 and 3-24 and are generally 28
percent less than the emission reductions of Table 6-1 which
are based on design-value temperatures. The emission
reductions used to develop an economic credit exclude the
portion which is attributed to exhaust emissions (67,000 tons
in 2010). The 5-month evaporative recovery credits are shown
in Table 6-2.
The costs of vehicle redesign solely for evaporative
emission control, on a dollar-per-vehicle basis, are taken fiotr,
Tables 4-5 thru 4-7 in Chapter 4. These are multiplied by-
Energy and Environmental Analysis (EEA) vehicle sales
-------
6-8
projections for the year 2010 to determine annual costs.[3]
The resulting annual vehicle design costs are summarized in
Table 6-2 and do not vary with sensitivity case
As alluded to in Chapter 4, redesigning motor vehicles for
higher volatility would increase the weight of the motor
vehicle slightly, as a larger canister would be required. This
weight increase would cause the motor vehicle to consume more
fuel over the course of its lifetime than if the vehicle did
not have added weight due to redesign. This weight penalty and
associated discounted lifetime costs per vehicle are derived in
Section II of Chapter 4 and are summarized in Table 4-9 These
per-vehicle costs are multiplied by the sales projections
described earlier to derive annual costs which are summarized
m Table 6-2. The total annual weight penalties are summarized
an Table 6-2 for each case.
The dnveability credit, derived in Chapter 5, is also
summarized m Table 6-2. This credit is taken from Table 5-24
and multiplied by five-twelfth's since the credit only occurs
when ln-use volatility control is occurring.
The net costs for each control scenario are then
calculated by simply adding costs and subtracting credits
Note that the cost calculations take into account the fact that
vehicle-related costs occur year-round, while costs and credits
associated wath commercial fuel volatility control occur over
the 5-month period only. (Vehicle-related credits occur either
year-round or just in the summer, depending on the case.) The
net, long-term, steady-state cost effectiveness is determined
by simply dividing the net cost by the 5-month emission
reduction. The ancremental long-term cost effectiveness is
determined first by calculating the incremental cost, then an
incremental emission reduction, and finally dividing the
incremental cost by the incremental emission reduction.
b. Cost Effectiveness Comparable to Other Strategies
The volatility control cost effectiveness calculated by
the above methodology is unique since the control program is
seasonally focused. Thus, it is desirable to convert the
previous analysis to one that is on the same basis as other
ozone reductions programs (e.g., refueling, Stage II,
Inspection/Maintenance programs, etc.). The methodology of
adjusting the previous analysis for this difference is detailed
below.
Nearly all other ozone reduction programs produce emission
reductions that occur over a 12 month period, though the
ozone-related benefits accrue primarily in the summer.
(However, it should be noted that many of the supporting
analyses use emission factors that are only representative of
-------
6-9
summer conditions.) Some programs control emissions only in
ozone non-attainment areas. Per ton of emission control, these
are the most valuable since they focus control in the areas
needing it most. Thus, a cost per ton which assumes year round
effectiveness and control only in non-attainment areas is the
most useful when comparing strategies.
The base case emission reduction values in Table 6-1 can
be converted to a year-round basis by first multiplying by 2.4
(i.e., twelve-fifth's), then by multiplying by 0.395. The
first factor simply expands 5-month emission reductions to
their 12-month equivalents. The second value represents the
fuel consumption in the 61 current, ozone non-attainment areas
as a percentage of nationwide fuel consumption. This
percentage was calculated using the NEDs data base, which
includes county-by-county fuel consumption data for the entire
U.S.[4]
The adverse health and welfare effects associated with
ozone concentrations in the non-attainment areas are of
principal concern, and are the primary focus of this analysis.
There are, however, benefits of lowering ozone concentrations
in other areas These benefits occur because some adverse
effects appear to have either no threshold level or no clearly
defined level below which damage does not occur. Research has
identified three areas which are adversely effected at
relatively low ozone concentrations: agriculture,
non-agriculture vegetation, and materials. The types of damage
that occur due to ozone include reduced plant growth, yield,
and quality, as well as damage to elastomers, paints, textile
fibers, and dyes. These effects are discussed in greater
detail in the Onboard/Stage II Regulatory Impact Analysis
(RIA).[1]
Assigning an exact monetary value to ozone reductions in
attainment areas is difficult. An estimate can be identified,
however, based on generally accepted figures. For year-round
VOC reductions in attainment areas, $250 per ton has been cited
by General Motors in a previous evaluation of hydrocarbon
control benefits.[5] This value has been utilized in the
Onboard RIA and will be here, as well.
To obtain the attainment area emission reductions, the
annual emission reductions which are based on July-average
temperature (Table 3-23) were simply multiplied by 0.605. This
value represents the fuel consumption in attainment areas
(i.e., one minus 0.395). Effective annual emission reductions
are used because that was assumed m the development of the
$250 per ton estimate. This assumption was made because nearly
all VOC control strategies produce year-round emission control
and no conversion of the $250 per ton figure would be needed
prior to its use. Since the actual benefits occur primarily in
-------
6-10.
the summer, as in non-attainment areas, the same conversion
from 5-month to year-round is needed.
The costs for this annualized cost effectiveness analysis
are the same as those cited above in Section 1., except for the
attainment area credit, since actual costs are used in both
cases, which are unaffected by methods to account for ozone
benefits. The net long-term, steady state cost-effectiveness
is determined by dividing net cost by annual non-attainment
area emission reductions. Incremental cost-effectiveness
values are determined in the same manner as before using the
adjusted incremental costs and emission reductions.
c. 33-Year Discounted Analysis
Both the actual and adjusted cost effectiveness
methodologies described above for the long-term are
steady-state in nature. That is, they focus on costs and
benefits in a given calendar year once all vehicles in the
fleet have controls. Start-up costs, such as the cost of
implementing vehicle controls while a portion of the fleet is
still uncontrolled and not producing emission reductions, are
ignored. At the same time, start-up costs only occur once, so
long-term programs, such as RVP control, should not be unduly
weighed by this effect.
One methodology which attempts to properly weight start-up
costs is the 33-year discounted cost effectiveness. It has
been used in the analysis of Stage II and onboard vehicle
refueling controls, along with the steady-state analysis.[1]
Greater weight in this analysis will be placed on the
steady-state cost effectiveness. However, the 33-year
discounted figures will provide insight into the effect of
start-up costs on the analysis, as well as improve
comparability with those programs for which a 33-year
discounted cost effectiveness has been estimated.
To determine the discounted cost effectiveness the costs
and emission reductions for each year over a 33 year period
need to be calculated. The net present value of costs and
emissions reductions will be determined using a discounting
rate of 10 percent. These factors were chosen to be consistent
with the analysis performed in the Onboard Refueling RIA.[1]
The emission reductions and costs for each of the 33 years are
determined analogously to the 2010 analysis above. The annual
emission reductions were already presented in Table 3-21.
Emission reductions for those years not modeled are linearly
interpolated from the modeled years. Emission reductions that
occur in the years beyond 2010 are set equal to those achieved
in 2010 since the vehicle fleet would consist entirely of
post-1990 (i.e., controlled) vehicles by this time and little
change will be occurring. Any mileage growth that occurs
-------
6-11
beyond 2010 would affect both emission reductions and costs to
the same degree and should not significantly affect the
cost-effectiveness calculation.
The net costs for each year are derived in the same manner
as in the long-term methodology outlined above. The specific
fuel consumption values for each year were taken from the
MOBILE3 Fuel Consumption Model and multiplied by the refinery
costs in Table 5-3 (which are based on $20 crude oil price) to
determine the refinery cost. The costs of vehicle redesign, on
a dollar-per-vehicle basis, are taken from Table 4-9 in Chapter
5 Short-term vehicle costs are used for years 1990-1994.
These are multiplied by EEA vehicle sales projections for each
individual year.[3] The evaporative recovery credits are
obtained by multiplying the July-average emission reductions
from Table 3-23 by the butane value of $290 32 per ton
developed in Section V of Chapter 4. Net costs are calculated
for each year in which emission modeling was performed. Net
cost for the years not modeled are linearly interpolated from
the modeled years. Net costs beyond 2010 are set equal to the
2010 net cost, consistent with the handling of emission
reductions
The net present value of both emission reductions and net
costs are then determined for the 33 year period, then
multiplied by the annual annuity factor (33 years at 10
percent). To increase comparability to other strategies, the
net present values of both costs and emission reductions are
also adjusted, as described in Section 6.2 above, for
seasonality and non-attainment area focus prior to the
calculation of the cost per ton ratio. Thus, as with the
steady-state analysis, both actual and adjusted cost
effectiveness figures are presented.
All cost effectiveness estimates (steady-state actual and
adjusted and 33-year discounted actual and adjusted) will be
presented in Section II. Their relative significance will be
analyzed and interpreted as the estimates are presented.
2. Short-Term Analysis
As highlighted elsewhere in this study, ozone control is
the primary focus of the various HC reduction strategies being
evaluated. Any reductions achievable in the shorter term could
be important in view of the Clean Air Act requirement that all
urban areas be in attainment of the ozone NAAQS by 1987.
Although the long-term strategy is to ensure that certification
fuel RVP is representative of in-use levels, a short-term
strategy could be to control in-use fuel RVP to a level lower
-------
6-12
than the long-term certification specification in order to
achieve the additional benefits associated only with in-use RVP
control. These benefits include: 1) an immediate effect on
the entire motor vehicle fleet, including older vehicles not
affected by the revised certification fuel, and 2) further
control of in-use RVP-related emissions such as gasoline
storage, distribution, refueling, and vehicular emissions due
to tampering, malmaintenance and defects.
Therefore, a short-term analysis has been performed to
focus on the incremental costs and emission reductions of
additional in-use RVP control over and above the long-term RVP
control strategies. This analysis focuses on the years 1988,
1990, 1992, 1995, 1997 and 2000. As before, the actual
cost-effectiveness analysis is described first, followed by the
adjusted analysis.
Year-round HC emission inventories with additional
short-term in-use RVP control were estimated in Chapter 3 and
summarized in Table 3-14; emission reductions were calculated
from this table and used directly here for the base case and
the without onboard case. As with the long-term analysis,
emission reductions for the crude cost sensitivity cases and
the winter-time credit case are the same as the base case. As
these emission reductions are due solely to in-use RVP control,
the emission reductions here are simply five-twelfths of the
year-round reductions.
The costs of additional in-use RVP control consist of
three parts: 1) refinery costs, 2) credit due to increased
fuel density, and 3) credit due to recovered or prevented
evaporative emissions. As before, the refinery cost of each
0.5 psi of RVP control was taken from Table 5-3. Here,
however, the "no-investment" costs were used for 1988 and 1990,
as there would not likely be time for refiners to invest in
capital equipment for the most economic RVP control refinery
processes. As discussed in Chapter 5 only 0.5 and 1.0 psi
decreases in RVP were investigated for this time period since
this appears to be the feasible limit of RVP control in a
no-investment situation. For the year 1992 it is assumed that
sufficient leadtime will be available to enable capital
investment to be made in order to meet lower RVP standards.
The refinery control costs however would not be equal to the
long-term "with-investment" costs. In the long-term analysis
annual capital charges were calculated based on amortization of
capital over an economic life of 13 years. In the short-term
RVP control scenario where RVP restraints may be imposed for a
limited period of time (5-10 years), the useful life of the
investments would be less than 13 years and the required annual
capital charges would be greater than the long-term refinery
costs. This issue was investigated in a contractor report by
Sobotka and Company for the EPA.[6] The report determined that
-------
6-13
when capital is amortized over ten years instead of 15 years
the cost of RVP reduction increased by an average of 20
percent. This increase is applied to the long-term refinery
costs for additional short-term RVP control in the years 1992
to 2000. These costs per barrel were again multiplied by
5-month, on-highway, fuel consumption projections from EPA's
MOBILE3 Fuel Consumption Model.[2]
The density-related fuel economy credit for each 0.5 psi
of additional RVP control is again taken from Table 5-19 and
multiplied by 5-month fuel consumption. The evaporative
emission recovery/prevention credit is again obtained by
multiplying the 5-month, July-average, emission reductions
calculated from Table 3-23 (minus exhaust emissions since their
energy is not recoverable by the engine) for additional in-use
RVP control by the butane value of $290.32 per ton
The cost effectiveness is simply the net cost divided by
the emission reduction. These estimates of short-term costs
per ton will be summarized along with long-term estimates in
Section II below.
The adjustments to the actual short-term cost
effectiveness to make it comparable with other strategies
follows exactly the same methodology as that used in the
long-term. Emission reductions were annualized and reduced to
represent effective annual reductions achieved in
non-attainment areas. The annualized emission reductions
occurring in attainment areas were credited at $250/ton and
subtracted from the actual net costs to arrive at the adjusted
net costs.
B. Cost Effectiveness Comparison of RVP Control
Scenarios
1. "Base" Case: Steady-State Analysis
The base case represents the combination of the most
likely future regulatory situation and EPA's best technical
estimates. As outlined earlier, this includes: 1) onboard
control of refueling losses, 2) no evaporative I/M program and
3) an assumption that m-use RVP will continue to rise until
ASTM limits are reached (i.e , Class C summertime RVP will
equal 11.5 psi, on average, by 1988), except that it will
remain constant where it already exceeds ASTM limits.
Actual summation emission reductions, costs, and cost
effectiveness estimates are shown for the base case m Tables
6-3 and 6-4 for R-values of 0.82 and 0.95, respectively. The
top portions present the long-term (2010) analysis, where
certification fuel and in-use fuel RVPs are made equal. The
11.5-psi scenario represents purely vehicle-oriented control as
it only requires a change in certification fuel RVP (i.e.,
-------
6-14
Table 6-3
"Base" Case: Actual Steady-State Cost Effectiveness, R = 0.82
In-Use Gasoline RVP Control Equal to Revised
Cert. Fuel RVP in the Long Term (2010)
In-Use = Emission Cost
Revised Reductions Net Cost Effectiveness
Cert. Fuel
(101 Tons/Yr)
( 106
$/Yr)
($/Ton)
RVP (psi)
Net
Incr.
Net
Incr.
Net
Incr .
11.5
576
576
-25
-25
-43
-43
11.0
616
40
-13
11
-22
279
10 . 5
655
39
8
22
13
562
10 . 0
689
34
43
35
63
1026
9.5
718
29
95
51
132
1762
9 . 0
748
30
156
62
209
2084
8.5
774
26
235
79
304
2999
8.0
799
25
316
81
395
3177
Additional In-Use
Gasoline
RVP Control
in
the Short
Term (1988
-2000)
Incremental
Control Step
Emission Reductions
(10JTons)
RVP (psi)
1988
1990
1992
1995
1997
2000
11.5-11.0
200
185
147
105
83
60
11.0-10.5
169
155
126
90
73
54
10.5-10.0
-
-
101
73
60
46
10.0- 9.5
-
-
80
58
48
38
9.5- 9.0
-
-
68
52
43
35
9.0- 8.5
-
-
40
34
30
28
8.5- 8.0
-
-
36
30
28
25
Incremental
Control Step Incremental Cost Effectiveness ($/Ton)*
RVP (psi)
1988
1990
1992
1995
1997
2000
11.5-11.0
99
121
129
251
356
542
11.0-10.5
186
229
301
474
619
871
10.5-10 . 0
-
-
527
764
957
1266
10 0- 9.5
-
-
1012
1384
1669
2144
9.5- 9.0
-
-
1401
1807
2149
2583
9.0- 8.5
-
-
3033
3500
3845
4110
8.5- 8.0
-
-
3393
3906
4107
4480
1988 and 1990 costs are based on no investment; 1992 and
-------
6-15
Table 6-4
"Base" Case: Actual Steady-State Cost-Effectiveness, R = 0.95
In-Use Gasoline RVP Control Equal to Revised
Cert. Fuel RVP in the Long Term (2010)
In-Use =
Emission
Cost
Revised
Reductions
Net Cost
Effectiveness
Cert. Fuel
(103 Tons/Yr)
(10G
$/Yr)
($/Ton)
RVP (psi)
Net
Incr.
Net
Incr.
Net Incr.
11.5
576
576
-22
-22
-39
-39
11. 0
616
40
-16
7
-25
166
10 . 5
655
39
1
17
2
430
10 . 0
689
34
31
30
45
873
9 . 5
718
29
77
46
107
1575
9 . 0
748
30
133
56
178
1894
8.5
774
26
206
73
266
2777
8 . 0
779
25
280
75
351
2935
Additional In-Use
Gasoline
RVP Control
in
the Short Term (1988
-2000)
Incremental
Control Step
Emission Reductions (10
3 Tons)
RVP (ps i)
1988
1990
1992
1995
1997
2000
11.5-11. 0
200
185
147
105
83
60
11.0-10 . 5
169
155
126
90
73
54
10.5-10.0
-
-
101
73
60
46
10.0- 9.5
-
-
80
58
48
38
9.5- 9.0
-
-
68
52
43
35
9.0- 8.5
-
-
40
34
30
28
8.5- 8.0
—
—
36
30
28
25
Incremental
Control Step
Incremental
Cost Effectiveness
($/Ton)*
RVP (psi)
1988
1990
1992
1995
1997
2000
11.5-11. 0
64
88
91
202
298
466
11.0-10 . 5
144
189
253
414
546
778
10.5-10 . 0
-
-
459
675
853
1139
10.0- 9.5
-
-
929
1278
1545
1991
9.5- 9.0
-
-
1305
1686
2009
2418
9.0- 8.5
-
-
2884
3328
3657
3908
8.5- 8.0
-
-
3215
3701
3893
4245
1^88 <-" ""i 1990 i e or. r.o investment, 1992
-------
6-16
m-use RVP is expected to average 11.5 psi); the 8.0-, 8.5- and
9.0-psi cases, on the other hand, requires primarily in-use
fuel control, with improvements to the evaporative emission
test procedure but no change in the certification fuel. All of
the other RVP scenarios combine the fuel-related and
vehicle-oriented approaches. Tables 6-3 and 6-4 show that, in
2010, the vehicle-oriented approach (ll.5-psi scenario) is
significantly more cost-effective than the scenarios involving
fuel control, and actually results in an overall cost savings
(i.e., negative cost per ton). This occurs because fuel
economy credits from recovered evaporative emissions outweigh
the costs of vehicle redesign. However, additional incremental
emission control is achievable via more in-use fuel-orlented
programs at increasing cost per ton.
The center and bottom portions of Tables 6-3 and 6-4
summarize the short-term analysis, where the strategy is to
control in-use fuel RVP to a level lower than the long-term
certification specification; after a specified time, in-use
fuel control would be relaxed to the long-term certification
RVP level chosen via the long-term analysis. The short-term
analysis focuses on years between 1988 and 2000 (inclusive).
The first column in this portion of the table shows the
increment between the long-term certification/in-use
specifications and the short-term m-use RVP being evaluated;
0.5-psi increments are shown (e.g., long-term RVP = 11.5,
short-term in-use RVP lower at 11.0). The incremental values
(based on 0.5-psi intervals) at a given in-use RVP (e.g., 10.0
psi) are the same regardless of long-term certification fuel
RVP (e.g., 10.5, 11.0, or 11.5 psi). It should be noted that
the emission reductions associated with equating certification
and in-use RVPs (i.e., the long-term strategy) are not included
in the center portion of the table, but rather only the
additional reductions that can be obtained with short-term
additional in-use control. The cost per ton estimates for each
of these control increments are shown in the bottom portion of
the table.
As indicated in Tables 6-3 and 6-4, short-term cost
effectiveness ($/ton) rises between 1988 and 2000. This occurs
primarily because the emission reductions achievable with
additional in-use RVP control decrease with time as the vehicle
fleet gradually turns over (i.e., vehicles with larger
canisters, designed for the revised long-term certification
RVP, beginning with the 1990 model year, start to make up more
and more of the m-use fleet). Therefore, incremental costs
per ton rises between 1988 and 2000 and eventually exceed the
long-term figures shown in the top portion of Tables 6-3 and
6-4, because the short-term refinery costs are higher than
those in the long-term due to a shorter, assumed amortization
period.
-------
6-17
Conceptually, the short-term cost effectiveness figures
are used to determine if cost effective emission reductions are
achievable at lower RVPs in the short term than in the long
term. For example, Table 6-3 shows the long-term cost
effectiveness of the 9.5 RVP scenario to be $1762 per ton, and
that of the 9.0 RVP scenario to be $2084 per ton. In 1992,
emission reductions are available with 9.0 RVP control at $1401
per ton. Thus, one might decide that 9.0 RVP was too costly in
the long term but reasonable in the short term. However, Table
6-3 also shows that the short-term cost per ton rises very
sharply, so that by 1997 it is essentially the same as the
long-term cost effectiveness Thus, short-term in-use RVP
control is generally only available at a lower cost per ton
than long-term control for 4-5 years. As the short-term
refinery costs assumes amortization of capital investment over
10 years, reducing this to 4-5 years would increase the
short-term costs per ton further and probably shorten even
further the period during which additional short-term in-use
RVP control is more cost effective than long-term control.
This is true for essentially all of the control scenarios
evaluated in this chapter. Thus, little will be said further
about short-term RVP control.
The effect of R-value is shown for the base case by a
comparison of Tables 6-3 and 6-4. As indicated m the tables,
the higher R-value results in slightly lower costs (due to
higher fuel economy credits) and, thus, arithmetically lower
cost effectiveness. For example, in the base case analysis of
the 9.0-psi long-term scenario, incremental cost/ton decreases
by 9 percent, or by $190 per ton, with the higher R-value.
As described in detail in the methodology section above a
number of changes were made to the base case, actual cost
effectiveness analysis, so that the results would be comparable
to year-round ozone reduction strategies in non-attainment
areas. The results of this fifth sensitivity analysis are
presented in Tables 6-5 and 6-6 for R-values of 0.82 and 0.95,
respectively.
The effect of the adjustment to the emission reductions
lowered the incremental emission reductions by 1,000 to 2,000
tons per year. This decrease in emission reductions is offset,
however, by the $250 per ton credit which is given to the
attainment area reductions, which lowered the incremental costs
by an average of 10 million dollars. The result is a lowering
of the incremental cost effectiveness by approximately $230 per
ton for the 9.0 psi RVP control step. It should also be noted
that the actual net cost effectiveness was negative for the
11.5 psi and the 11.0 psi control steps, while the adjusted net
cost effectiveness is negative at 11.5 psi down to the 8.5 or
9.0 psi step, depending on R-value.
-------
6-18
Table 6-5
Base Case: Adjusted Steady-State Cost Effectiveness R = 0.82
In-Use Gasoline RVP Control Equal to Revised
Cert. Fuel RVP in the Long Term (2010)
In-Use =
Emission
Cost
Revised
Reductions
Net <
Cost
Effectiveness
Cert. Fuel
(103 Tons/Yr)
(10
s $/Yr)
<$/Ton)
RVP (psi)
Net
Incr .
Net
Incr.
Net
Incr.
11. 5
546
546
-169
-169
-310
-310
11.0
584
38
-170
-1
-291
-21
10 . 5
621
37
-160
10
-258
267
10 . 0
653
32
-136
24
-208
750
9 . 5
681
28
-94
42
-138
1509
9 . 0
709
28
-42
52
-59
1854
8.5
734
25
27
70
37
2798
8.0
758
24
100
72
132
3000
Additional In-Use Gasoline
RVP Control
in
the Short Term (1988-2000)
Incremental
Control Step
Emission
Reductions (10
Tons)
RVP (psi)
1988
1990
1992
1995
1997
2000
11.5-11.0
190
175
139
99
78
57
11.0-10.5
160
147
119
85
69
51
10.5-10.0
-
-
96
70
56
43
10.0- 9.5
-
-
75
55
46
36
9.5- 9.0
-
-
64
49
41
34
9.0- 8.5
-
-
38
32
28
26
8.5- 8.0
—
—
34
28
26
24
Incremental
Control Step
Incremental
Cost Effectiveness ($/Ton)
*
RVP (ps i)
1988
1990
1992
1995
1997
2000
11.5-11. 0
-187
-161
-154
-22
86
278
11.0-10 . 5
-108
-62
17
193
348
607
10.5-10 . 0
-
-
247
494
693
1008
10.0- 9.5
-
-
758
1142
1447
1934
9.5- 9.0
-
-
1157
1582
1929
2387
9.0- 8.5
-
-
2876
3361
3710
3998
8.5- 8.0
-
-
3223
3758
3984
4362
* 1988 and 1990 costs are based on no investment; 1992 and later
assume investment has taken place; refinery costs are midway
-Or-- •<=<=," --P -I— - -> ; ' --Appi -iQS
-------
6-19
Table 6-6
Base Case: Adjusted Steady-State Cost Effectiveness R = 0.95
In-Use Gasoline RVP Control Equal to Revised
Cert. Fuel RVP in the Long Term (2010)
In-Use -
Emission
Cost
Revised
Reductions
Net Cost
Effectiveness
Cert. Fuel
(10 J Tons/Yr)
(106
$/Yr)
($/Ton)
RVP (psi)
Net Incr
Net
Incr .
Net
Incr .
11.5
546 546
-167
-167
-306
-306
11.0
584 38
-172
-5
-295
-140
10 . 5
621 37
-167
5
-270
129
10 . 0
653 32
-148
19
-227
590
9 . 5
681 28
-112
36
-165
1311
9 . 0
709 28
-66
46
-93
1653
8 . 5
734 25
-2
64
-3
2565
8.0
758 24
64
66
85
2744
Additional In-Use
Gasoline
RVP Contr
ol
in the Short
Term (1988-2000)
Incremental
Control Step
Emission Reductions (10J
Tons)
RVP (psi)
1988 1990
1992
1995
1997
2000
11.5-11.0
190 175
139
99
78
57
11.0-10.5
160 147
119
85
69
51
10.5-10.0
-
96
70
56
43
10.0- 9.5
-
75
55
46
36
9.5- 9.0
-
64
49
41
34
9.0- 8.5
-
38
32
28
26
8.5- 8.0
34
28
26
24
Incremental
Control Step
Incremental Cost Effectiveness ($/Ton)
*
RVP (psi)
1988 1990
1992
1995
1997
2000
11.5-11.0
-223 -196
-194
-74
24
198
11.0-10.5
-153 -104
-33
130
270
508
10.5-10 . 0
-
176
401
584
875
10.0- 9.5
-
672
1031
1316
1772
9.5- 9.0
-
1055
1454
1781
2213
9.0- 8.5
-
2718
3180
3512
3786
8.5- 8.0
-
3035
3543
3758
4115
* 1988 and 1990 costs are based on no investment; 1992 and later
assume investment has taken place, refinery costs are midway
ho'wac- - _i - - -i- - - • ~ rrui^r-f ccenarios
-------
6-20
2. Base Case: 33-Year Discounted Analysis
The 33-year discounted cost effectiveness figures are
presented in Tables 6-7 and 6-8. The format of the two tables
is slightly different than that of the previous tables since,
with 33-year discounting, there is no long- or short-term. The
first table (6-7) presents the analysis of actual costs and
emission reductions for R equal to 0.82 (top half) and .95
(bottom half). The second table (6-8) present the analysis of
adjusted costs and emission reductions.
Comparison of the results of the discounted analysis with
the steady-state analysis reveals a number of differences. Net
emission reductions are much lower for the discounted analysis,
especially at 11.5 psi control (vehicle control) and the first
few increments of control. As RVP control approaches 8.0 psi,
the net emission reductions approach those of the steady-state
analysis. This is explained by the fact that in the
steady-state, long-term analysis, it is assumed that
essentially the entire fleet consists of vehicles with
evaporative emission control systems. The discounted analysis,
on the other hand, weighs all of the years prior to the
steady-state analysis where, in the first few year, very few
vehicles have control systems. Therefore, the more
vehicle-oriented control scenarios in the discounted analysis
show lower net emissions reductions than the steady-state
analysis. This results in an increase in the incremental
emission reductions due to in-use RVP control. More emissions
are controlled in the early years of a volatility control
program by in-use fuel control, and this is reflected in an
increase in the incremental emission reductions of the
discounted analysis.
The same factors that influence the emission reductions
also influence the net costs in the discounted analysis.
Initial vehicle costs in all the years of the control program
are now realized immediately while the evaporative emission
recovery credit in the vehicle-oriented scenarios grows slowly
over time. Thus, the evaporative emission recovery credit,
which outweighs the vehicle cost in the steady-state analysis,
is not large enough in the discounted analysis to offset the
cost of the new vehicle hardware. As a result the first two
steps of control which were available to the economy at a net
savings m the base case are now occurring at a small net
cost. As shown m Tables 6-7 and 6-8, although net costs
increase, incremental costs do not change nearly as
significantly as the incremental emission reductions. The
combination is an increase in the net cost effectiveness of all
control scenarios. Since the increase however is greater at
the vehicle-control end of the cost effectiveness analysis
(11.5-11.0 psi RVP) than on the fuel control side (9.0 psi) the
result is a decrease in the incremental costs per ton of the
-------
6-21
Table 6-7
33-Year Discounted, Actual Cost Effectiveness
In-Use =
Revised
Cert. Fuel
RVP (psi)
R = .82
Emission
Reductions
(105 Tons/Yr)
Net Cost
(10s $/Yr)
Cost
Effectiveness
($/Ton)
Net
Incr
Net
Incr
Net
Incr
11.5
411
411
11
11
26
26
11 . 0
492
82
17
6
35
80
10 . 5
564
72
35
18
62
250
10 . 0
624
59
67
32
108
541
9 5
672
48
117
50
175
1038
9 . 0
715
44
178
60
248
1385
8. 5
746
31
257
80
345
2575
8.0
775
28
339
82
438
2670
R = . 95
In-Use =
Emission
Cost
Revised
Reduct ions
Net
Cost*
Effectiveness
Cert. Fuel
(103
Tons/Yr)
( 106
$/Yr)
($/Ton)
RVP (psi)
Net
Incr .
Net
Incr .
Net
Incr .
11. 5
411
411
12
12
30
30
11. 0
492
82
14
2
28
19
10. 5
564
72
26
12
46
171
10.0
624
59
52
26
84
443
9.5
672
48
96
44
143
913
9.0
715
44
151
54
211
1246
8.5
746
31
225
74
301
2388
8.0
775
28
300
75
387
2654
-------
6-22
Table 6-8
33-Year Discounted, Adjusted Cost Effectiveness
R = .82
In-Use = Emission Cost
Revised Reductions Net Cost Effectiveness
Cert. Fuel (103 Tons/Yr) (106 $/Yr) ($/Ton)
RVP (psi)
Net
Incr.
Net
Incr .
Net
Incr .
11 . 5
389
389
- 92
-92
-237
-237
11. 0
467
77
-108
-16
-232
-209
10 . 5
535
68
-111
-3
-208
-45
9 . 5
591
56
-97
14
-164
254
9.0
637
46
-62
35
-97
773
9 . 0
678
41
-15
47
-22
1129
8.5
708
29
55
70
77
2376
8.0
734
27
127
72
172
2671
R = .95
In-Use =
Emission
Cost
Revised
Reductions
Net Cost
Effectiveness
Zert. Fuel
(10 1
Tons/Yr)
(10G
$/Yr)
($/Ton)
RVP (psi)
Net
Incr.
Net
Incr .
Net
Incr .
11. 5
389
389
-91
-91
-233
-233
11.0
467
77
-112
-21
-239
-273
10. 5
535
68
-120
-9
-225
-127
10 . 0
591
56
-112
8
-189
150
9.5
637
46
-83
29
-130
642
9.0
678
41
-42
41
-62
982
8.5
708
29
22
64
31
2178
8.0
734
27
88
66
120
2444
-------
6-23
8.0-11.0 RVP scenarios. For 9.0 psi RVP control, the actual
incremental cost effectiveness decreases by 34 percent from
$1894-2084 per ton to $1246-1385 per ton, depending on R
value. The same holds true for the adjusted cost-effectiveness
figures.
3. "Without Onboard Control" Case
The first sensitivity case examines the effect of assuming
that onboard refueling loss control is not implemented prior to
RVP control. Only the actual, steady-state cost effectiveness
figures will be compared.
In the base case an onboard implementation date of 1990
was assumed. For all years except 1988 (pre-onboard control),
the emission reductions are somewhat higher than under the base
case because in-use RVP control would capture a certain
percentage of the reductions in refueling losses previously
attributed to onboard control. This increase in emission
reductions also lowers the cost of the RVP control program to a
certain extent due to the slight increase m evaporative
recovery credits.
The emission reductions, costs, and resulting cost
effectiveness estimates for this without onboard control
sensitivity case are presented in Tables 6-9 and 6-10 for R
values of 0.82 and 0.95, respectively. The overall effect is
an improvement of the cost effectiveness for the fuel-oriented
RVP control strategies. For example, the incremental cost
effectiveness of the 9 RVP scenario in 2010 without onboard is
27 percent lower ($1380-1519 per ton) versus the base case
($1894-2084 per ton).
4. "Crude Oil Cost" Cases
The second and third sensitivity cases examine the effect
of crude oil cost on the actual, steady-state cost
effectiveness of RVP control. Two sensitivity cases were run,
one with a crude value of $15 per barrel and one with a value
of $25 per barrel. The emission reductions, costs, and
resulting cost effectiveness estimates for these cases are
presented in Table 6-11 through 6-14 with R-value and crude oil
price changing alternately.
Table 5-4 of Chapter 5 illustrates the effect of crude oil
cost on refinery RVP control cost. Refinery costs are highly
dependent on crude oil price since additional crude oil
purchases make up a large part of the RVP control cost. This
sensitivity is also evident in the net cost of the RVP control
program but to a lesser degree. Due to the fixed costs of
producing and distributing gasoline, the value of the recovered
emissions (retail price minus taxes) is not as strongly
-------
6-24
Table 6-9
"Without Onboard" Case: Actual Steady-State
Cost Effectiveness, R = 0.82
In-Use Gasoline RVP Control Equal to Revised
Cert. Fuel RVP in the Long Term (2010)
In-Use =
Emission
Cost
Revised
Reductions
Net Cost
Effectiveness
Cert. Fuel
(103 Tons/Yr)
(106
$/Yr)
($/Ton)
RVP (psi)
Net
Incr.
Net
Incr .
Net
Incr .
11. 5
576
576
-24
-24
-42
-42
11. 0
625
49
-16
8
-26
172
10 . 5
673
48
3
19
5
399
10 . 0
716
43
35
32
49
744
9.5
754
38
84
49
112
1309
9.0
793
39
143
59
181
1519
8.5
828
35
220
76
265
2151
8.0
863
35
298
78
345
2258
Additional In-Use
Gasoline
RVP Contr
ol
in
the Short '
rerm (1988
-2000)
Incremental
Control Step
Emission Reductions (10
Tons)
RVP (ps i)
1988
1990
1992
1995
1997
2000
11.5-11.0
200
185
150
110
89
68
11.0-10.5
169
156
128
96
79
62
10.5-10.0
-
-
103
78
66
53
10.0- 9.5
-
-
83
64
54
45
9.5- 9.0
-
-
70
57
50
43
9.0- 8.5
-
-
43
38
36
34
8.5- 8.0
-
—
38
35
33
33
Incremental
Control Step
Incremental Cost Effectiveness ($/Ton)
*
RVP (psi)
1988
1990
1992
1995
1997
2000
11.5-11.0
99
120
121
225
311
448
11.0-10.5
186
228
289
430
547
730
10.5-10.0
-
-
507
698
835
1049
10.0- 9.5
-
-
966
1242
1451
1725
9.5- 9.0
-
-
1341
1622
1811
2078
9.0- 8.5
-
-
2843
3037
3135
3244
8.5- 8.0
-
-
3152
3310
3382
3379
1988 and 1990 costs are based on no investment; 1992 and
later assume investment has taken place; refinery costs are
midway between those of open and fixed NGL purchase
-------
6-25
Table 6-10
"Without Onboard" Case: Actual Steady-State
Cost Effectiveness, R = 0.95
In-Use Gasoline RVP Control Equal to Revised
Cert. Fuel RVP in the Long Term (2010)
In-Use =
Emission
Cost
Revised
Reductions
Net i
Cost
Effectiveness
Cert. Fuel
(103 Tons/Yr)
(10
6 $/Yr)
($/Ton)
RVP (psi)
Net
Incr
Net
Incr .
Net Incr.
11 . 5
576
576
-22
-22
-38
-38
11.0
625
49
-17
5
-27
103
10 5
673
48
-2
15
-3
308
10 . 0
716
43
25
28
35
635
9.5
754
38
69
44
92
1174
9.0
793
39
123
53
155
1380
8 . 5
828
35
193
70
233
1988
8.0
863
35
265
72
307
2082
Additional In-
Use
Gasoline
RVP Contr
Ol
ln
the Short
Term (1988-2000)
Incremental
Control Step
Emission Reductions (105
Tons)
RVP (psi)
1988
1990
1992
1995
1997
2000
11.5-11.0
200
185
150
110
89
68
11.0-10.5
169
156
128
96
79
62
10.5-10.0
-
-
103
78
66
53
10.0- 9.5
-
-
83
64
54
45
9.5- 9.0
-
-
70
57
50
43
9.0- 8 5
-
-
43
38
36
34
8.5- 8.0
-
-
38
35
33
33
Incremental
Control Step Incremental Cost Effectiveness ($/Ton)*
RVP (psi)
1988
1990
1992
1995
1997
2000
11. 5-11.0
64
104
101
197
276
402
11.0-10.5
144
203
259
390
499
669
10.5-10.0
-
-
459
635
762
960
10.0- 9.5
-
-
904
] 164
1361
1620
9.5- 9.0
-
-
1262
1528
1707
1959
9.0- 8.5
-
-
2694
2887
2981
3083
8.5- 8.0
-
-
2986
3136
3204
3200
1988 and 1990 costs are based on no investment; 1992 and
later assume investment has taken place; refinery costs are
nidv/ay betveen tKose of open and fixed NGL purchase
-------
6-26
Table 6-11
"15 $/bbl Crude Oil" Case: Actual Steady-State
Cost Effectiveness Analysis, R = .82
In-Use Gasoline RVP Control Equal to Revised
Cert. Fuel RVP in the Long Term (2010)
In-Use =
Revised
Cert. Fuel
Emission
Reductions
(103 Tons/Yr)
Net Cost
(1Q6 $/Yr)
Cost
Effectiveness
($/Ton)
RVP (psi)
Net
Incr .
Net
Incr.
Net
Incr.
11 . 5
576
576
-15
-15
-26
-26
11 . 0
616
40
-20
-5
-33
-126
10 . 5
655
39
-7
14
-10
352
10 . 0
689
34
21
28
31
813
9 . 5
718
29
56
35
78
1188
9 . 0
748
30
99
44
133
1473
8.5
774
26
162
63
210
2400
8.0
799
25
234
71
292
2801
Incremental
Control Step
RVP (psi)
Additional In-Use Gasoline RVP Control
in the Short Term ( 1988-2000)
Emission Reduction (lO'Tons)
1990 1992
1988
1995
1997
2000
11.5-11.0
200
185
147
105
83
60
11.0-10.5
169
155
126
90
73
54
10.5-10.0
-
-
101
73
60
46
10.0- 9.5
-
-
80
58
48
38
9.5- 9.0
-
-
68
52
43
35
9.0- 8.5
-
-
40
34
30
28
8.5- 8.0
-
-
36
30
28
25
Incremental
Control Stet
Incremental Cost Effectiveness ($/Ton)*
RVP (psi)
1988
1990
1992
1995
1997
2000
11.5-10.0
-12
3
-1
68
125
226
11.0-10 . 5
134
170
223
363
480
682
10.5-10.0
-
-
441
643
808
1072
10.0- 9.5
-
-
745
1029
1246
1608
9.5- 9.0
-
-
1060
1374
1637
1972
9.0- 8.5
-
-
2519
2908
3195
3415
8.5- 8.0
-
-
3025
3483
3663
3995
1988 and 1990 costs are based on no investment; 1992 and
later assume investment has taken place; refinery costs are
midway between those of open and fixed NGL purchase
-------
6-27
Table 6-12
"15 $/bbl Crude Oil" Case: Actual Steady-State
Cost
Effectiveness Analysis, R =
0 . 95
In-Use Gasoline RVP
Control Equal to
Revised
Cert.
Fuel RVP in
the Lonq
Term (2010)
In-Use =
Emission
Cost
Revised
Reductions
Net
Cost
Effectiveness
Zert. Fuel
( 10 3
Tons/Yr)
(10
6 $/Yr)
($/Ton)
RVP (ps i)
Net
Incr .
Net
Incr .
Net
Incr
11. 5
576
576
-13
-13
-23
-23
11.0
616
40
-21
-8
-34
-202
10. 5
655
39
-12
10
-18
250
10 . 0
689
34
12
24
17
688
9 . 5
718
29
42
30
59
1031
9 . 0
748
30
81
39
108
1309
8 . 5
774
26
139
58
179
2203
8 . 0
799
25
204
66
256
2585
Additional In-Use Gasoline RVP Control
in the Short Term ( 1988-2000)
Incremental
Control Step Emission Reduction (10* Tons)
RVP (ps i)
1988
1990
1992
1995
1997
2000
11.5-11.0
200
185
147
105
83
60
11.0-10.5
169
155
126
90
73
54
10.5-10.0
-
-
101
73
60
46
10.0-9.5
-
-
80
58
48
38
9.5-9.0
-
-
68
52
43
35
9.0-8.5
-
-
40
34
30
28
8.5-8.0
-
-
36
30
28
25
Incremental
Control Step Incremental Cost Effectiveness ($/Ton)*
RVP (psi)
1988
1990
1992
1995
1997
2000
11.5-11.0
-25
-12
-20
41
91
179
11.0-10.5
112
148
196
325
433
619
10.5-10.0
-
-
397
583
735
978
10.0-9.5
-
-
688
953
1156
1493
9.5-9.0
-
-
987
1282
1528
1842
9.0-8.5
-
-
2386
2755
3027
3236
8.5-8.0
-
-
2867
3301
3472
3786
1988 and 1990 costs are based on no investment; 1992 and
later assume investment has taken place; refinery costs are
midway between open and fixed NGL purchase scenarios.
-------
6-28
Table 6-13
"25 $/bbl Crude Oil" Case: Actual Steady-State
Cost Effectiveness Analysis. R - 0.82
In-Use Gasoline RVP Control Equal to Revised
Cert. Fuel RVP in the Long Term (2010)
In-Use = Emission Cost
Revised Reductions Net Cost Effectiveness
Cert. Fuel (10 1 Tons/Yr) (lO* $/Yr) ($/Ton)
RVP (psi)
Net
I nc r .
Net
I nc r.
Net
I nc i
11.5
576
576
-33
-33
-57
-57
11.0
'jib
40
-10
23
-17
564
10. 5
655
39
27
38
42
97 0
10.0
689
29
76
49
111
1427
9.5
718
30
138
61
192
2101
9.0
748
30
217
80
290
2089
iTi
00
774
26
311
93
401
3562
8.0
799
25
406
95
508
37 52
Additional In-Use Gasoline RVP Control
in the Short Term (1988-2000)
Incremental
Control Step
Emission Reduction (10 Tons)
RVP (psi)
1988
1990
1992 1995
1997
2000
11.5-10.5
200
185
147
105
83
60
11.0-10.5
169
155
126
90
73
54
10.5-10.0
-
-
101
73
60
46
10.0- 9.5
-
-
80
58
48
38
9.5- 9.0
-
-
68
52
43
35
9.0- 8.0
-
-
40
34
30
28
8.5- 8.0
_
_
36
30
28
25
Incremental
Control Step
Incremental Cost Effectiveness ($/Ton)*
RVP (psi)
1988
1990
1992
1995
1997
2000
11.5-11.0
177
204
219
379
520
766
11.0-10.5
306
358
457
690
885
1225
10.5-10.0
-
-
693
989
1231
1619
10.0- 9.5
-
-
1170
1596
1923
2469
9.5- 9.0
-
-
17 39
2 234
2654
3184
9.0- 8.5
-
-
3515
4055
4455
4761
8.5- 8.0
-
-
3931
4524
4758
5189
] 0 o p J ( ^ ° ; h c l • o l* - i; J o ^ tf) \ 1 C"1 0 2 P n ^
-------
6-29
Table b-14
25 $/bbl Crude Oil" Case: Actual Steady-State
Cost
Effectiveness Analysis
, R =
0.95
In-Use Gasoline RVP
Control Equal to
Revised
Cert.
Fuel RVP in
the Long Term (2010)
In-Use =
Emission
Cos t
Revised
Reductions
Net Cost
Effectiveness
Zert. Fuel
(10
Tons/Yr)
(10
S/Yr )
($/Ton)
RVP (psi)
Net
I nc r .
Net
Incr .
Net
I nc i
11.5
576
576
-30
-30
-53
-53
11.0
616
40
-13
18
-20
443
10. 5
655
39
20
32
30
8 34
10.0
689
34
63
43
91
12 b 2
9.5
718
29
118
55
164
1896
9.0
748
30
191
73
256
2480
8.5
774
26
279
87
360
3319
8.0
799
25
367
89
459
3486
Additional In-Use Gasoline RVP Control
m the Short Term (1988-2000)
Incremental
Control Step
Emission Reduction (10 Tons)
RVP (psi)
1988
1990
1992
1995
1997
2000
11.5-11.0
11.0-11.5
10.5-10.0
10.0-9.5
9.5-9.0
9.0-8.5
8.5-8.0
200
169
185
155
147
105
83
60
126
90
73
54
101
73
60
46
80
58
48
38
68
52
43
35
40
34
30
28
36
30
28
25
Incremental
Control Step Incremental Cost Effectiveness ($/Ton)*
RVP (psi)
1988
1990
1992
1995
1997
2000
11.5-11.0
141
169
198
327
458
685
11.0-10.0
261
316
426
626
807
1125
10.5-10.0
-
-
640
894
1119
1482
10.0-9.5
-
-
1101
1482
1789
2302
9.5-9.0
-
-
1654
2103
2502
3005
9.0-8.5
-
-
3369
3867
4248
4540
8.5-8.0
-
-
3755
4300
4522
4932
k
lOQR 1OC0 CO^1" " ' a ^ fs » »•} D'-t-/nn i|- ; -a o 1
-------
6-30
dependent on crude oil price as the refinery control cost. For
example, when crude oil drops 19 percent from $20 per barrel to
$15 per barrel, the value of recovered emission credits
decreases by only 11 percent from $0.82 per gallon to $0.73 per
gallon. Thus, the net cost of control will change by a greater
percentage (either up or down) than the drop in crude oil
price. For example, at 9.0 psi RVP the incremental cost of
control for the $15 per barrel case is approximately 29 percent
lower than the $20 per barrel base case. This results in a
similar percentage reduction in incremental cost
effectiveness. With crude oil at $25 per barrel incremental
cost increase by 38 percent at the 9.0 psi RVP control step
with a similar percent increase in cost effectiveness.
Overall, a $15 per barrel crude price lowers the cost per ton
of RVP control so that one additional 0.5 psi step of RVP
control is achievable of the same cost effectiveness
Likewise, a $25 per barrel crude oil price results in one less
0.5 psi of RVP control being available at the same cost per ton.
C. Summary
Based on the technical estimates and assumptions used in
the base case analysis, purely vehicle-oriented control (i.e.,
certification fuel RVP revised to 11.5 psi and improved test
procedure) appears to be the most cost-effective approach in
the long term. As shown earlier in Table 6-3, a total of
576,000 tons of summertime HC emissions can be eliminated via
this strategy in the year 2010 at a net savings of roughly $43
per ton to the public. However, because a change in
certification fuel and test procedure can only affect vehicle
design, strategies involving in-use RVP control can reduce
emissions further. This is possible because in-use RVP levels
affect those portions of evaporative emissions attributable to
malmaintenance/defects and evaporative system tampering, as
well as stationary sources such as gasoline storage,
distribution, and vehicle refueling, whereas certification fuel
RVP has no impact on these emissions. In 2010, in-use fuel RVP
control could eliminate up to an additional 223,000 tons of
summertime HC emissions at an incremental cost per ton ranging
between $166 and $3177 per ton. When adjusted to be comparable
with year-round strategies focused entirely on non-attainment
area emissions, the incremental cost per ton of in-use RVP
control becomes $140 to $3000 per ton. The 33-year discounted
cost per ton for in-use RVP control were $100-850 per ton lower
than the actual and adjusted steady-state (year 2010) figures.
Additional short-term in-use RVP control is only as cost
effective as long-term RVP control for a period of 4-5 years
(i.e., 1992-1997). As this provides little time for
amortization of capital costs, it does not appear to offer a
significant advantage over long-term control.
-------
6-31
The absence of an onboard control program has very little
effect on the 11.5 psi RVP control scenario. However, with
respect to all other scenarios, it increases incremental
emission benefits and slightly decreases incremental costs, so
the actual incremental steady-state (2010) cost per ton is
decreased by as much as 34 percent, or up to $1000 per ton.
The assumption of crude oil cost also influences the cost
effectiveness of the 8.0-11.0 RVP scenarios. A 25 percent
decrease in the crude oil cost to $15 per barrel decreases cost
effectiveness by roughly the same proportions, while a 25
percent increase raises the cost effectiveness by roughly 35
percent.
11. %iso Control
Control of the mid-range volatility has been discussed as
a means to achieve further evaporative emission reductions, in
addition to in-use fuel RVP control. The purpose of this
section is to determine the cost effectiveness of controlling
the % i $ o of in-use fuels (gasohol, methanol blends, and
straight gasolines) The evaporative emission reductions, on a
per vehicle basis, were determined in Chapter 2 and are
extrapolated to the fleet in this section. The costs of
controlling the %i6o of m-use fuels were developed in
Chapter 5.
The %igo point of in-use fuels varies widely for both
straight gasolines and alcohol blends even when comparing
within the same ASTM classes. Alcohol blends generally have
greater %i60 values than straight gasolines with increases of
up to 10 percent for volatility matched methanol blends and 15
percent for splash-blended gasohol. This is because of the
effects of the higher volatile azeotropes formed when alcohol
is mixed with straight gasoline. Based on recent summer MVMA
fuel survey results, the %iSo point of all alcohol blends was
assumed to be 10 percent greater than the %i6o point of ASTM
Class C straight gasolines which was assumed to be 35 percent.
The total emission reductions attributable to a reduction
in the %iso point of in-use fuels is a weighting of the
tampered and non-tampered emission reductions based on the
percentages of tampered and non-tampered vehicles in the
fleet. This must be adjusted for the percentage of vehicles in
the total fleet which are carbureted (based on miles traveled
rather than number of vehicles) and also the percent of the
fuel supply of which the %i6o is being controlled. A final
adjustment is also necessary to extrapolate the emissions
reductions from LDVs to the entire vehicle fleet including LDTs
and HDVs based on MOBILE3 results. The evaporative HC emission
inventories based on a certification fuel RVP and in-use fuel
RVP of 9.0 psi were used to calculate the LDV to fleet
adjustment factor. The use of other higher RVP cases results
-------
6-32
in a small change in the adjustment factor but does not affect
the final calculations to any extent.
The equation and inputs from the %]6o analysis presented
in Chapter 2 used to calculate the fleetwide emission
reductions are presented in Table 6-15. Tables 6-16 and 6-17
contain the predicted fleetwide emissions reductions for both
1990 and 2010 on a gram per mile and annual tonnage basis for
the additive and multiplicative approaches (as explained in
Section II, Chapter 2), respectively.
The cost effectiveness results in 1990 and 2010 of
reducing the %iso point of the alcohol blends to current
m-use gasoline levels and then reducing all m-use fuels
further are shown in Tables 6-18 and 6-19. The cost
effectiveness results are based on annual costs (based on the
cost of %,so control developed in Chapter 5) and annual
emission reductions, and are calculated by dividing the cost of
volatility control by the emission reductions within each
scenario. In 1990 the cost effectiveness range for reducing
the %i6o value of all in-use fuels from 35 percent to 25
percent is $6,670 to $18,610 per ton. The cost effectiveness
increases by 2010 to a range of $21,970 to $61,290 per ton
The increase in cost effectiveness is due primarily to the
decrease in the percentage of carbureted vehicles in the
vehicle fleet whose hot-soak emissions are affected by control
of the %ieo point of m-use fuels.
These cost effectiveness results do not compare favorably
with other volatility control cost effectiveness results
(calculated on the same annual basis) determined in this
chapter. For example, in the case of reducing in-use fuel RVP
to 10.5 psi, the short-term incremental cost effectiveness is
$229 per ton m 1990 (see Table 6-3). In the long-term, the
incremental cost effectiveness for making in-use fuel RVP equal
certification fuel RVP at 9.0 psi, is $2,084 per ton in 2010
(see Table 6-3). RVP control is more cost effective than
%iso control in both the short-term and long-term.
Pursuing the control of the %i6o of in-use fuels does
not appear reasonable at this time from a cost-effectiveness
standpoint for either alcohol blends or straight gasoline. It
should be noted, these results are based on a limited number of
emission test results which were tested at a laboratory whose
hot-soak emissions do not match closely EPA's measured hot-soak
emissions tested under the same conditions. In addition, the
results do not rule out the possibility of placing a cap on
%iso of certification fuel to current m-use levels to
prevent future emission increases due to the effect of %i6o
on hot-soak emissions. As explained in Chapter 2, control of
certification fuel %i6o appears to be cost effective and is
already included in the cost effectiveness results of this
chapter.
-------
6-33
"^ibl o G-i5
Emission Reduction Obtainable TVirouqh Control
of %iftp of Tn-llse Fuels
Emission^- = (Aq/test )Nnn_ ) + (Attest
Reluct inn Tamnerel Fraction •ira,T r 11 Fraction
."triosMav. .Carbureted fraction of fL.
"< ("3—; n—^ x (Annual VMT) x ( , )
fMiles/dav bised on miles tnvylfil
. , F.nV to FU.«<
x (Fuel market penetration fraction) x ,., )
' Adiustment Factor
(Z\q/test)Mon_t^mpf.,re^ = dervndent on relucti on m °ifi0 below)
(Aq/test)Tampered = dependent on reduction m ?str-,0 (see IvLow)
Non-tamr>ered fraction = 0.97
Tampered fraction^ = 0.03
#tnps/day = 3.02
#mii.es/dav = 31. l
FueL market penetration fraction: GasoLine - 0.RO
GasolioL = 0.07
Methanol = 0.01
%l60 Reduction
45-35
35-30
35-25
Ad 1111 ve Me thod
(/\q/test ^on-^amnered
Straiaht Line Curve
0.20
0. L0
0.?0
0.34
O.lL
0. l9
Multiplicative Method
(/\a ! t est) Non-Tampe red
Straiqht. Line Curve
0 .GO
0. ?0
0.G0
0.05
0.35
0.57
(Aq/test) Tampered
4.00
">.50
4.00
Year
Fraction of fleet affected^
Annual Vehicle Miles Travels (W)'*
LDV to Fleet Adjustment Factor^
L000
2010
0.455 0.1L2
l5GI.75 X 10° LRn5.0G X 10q
1.7L L.56
1 Emission Reductions in crrams. To obtain q/miLe, divide bv annua L WTT
2 Based on mileage of 50,000 miles and evap. survev. (See Fiqure 2-R of
this report.)
3 From "Percent of VMT Accumulated bv Fuel Imected Vehicles," EPA memo
from Hark Wolcott to Rick Rvkowski, -Tanuarv 8, L90G.
4 Based on M0BILE3 fuel consumotion modeL predictions.
5 LDV to FLeet = M0BILE3 FLeetwide Evaporative HC Emissions Inventory
Adiustment Factor M0BILE3 LOV Evaporative TIC Emission Inventory
-------
6-34
Table 6-16
Emission Reductions Obtainable
Through Control of %i6q of In-Use Fuels
(Additive Method)
Fuel
Gasohol
Methanol
All Fuels
Final
45
35
45
35
35
30
25
Fleetwide Emission Reductions in 1990
Straight Line
g/mi tons/yr
0 0018
0.0010
0 0130
0.0260
*
3 , 130
*
1 , 860
22,350
44,660
Curve
q/mi tons/yr
0 . 0025
0.0014
0.0137
0 . 0245
4 ,360
*
2 , 490
23,610
42,140
Fleetwide Emission Reductions in 2010
Fuel
Gasohol
Methanol
All Fuels
Final
45
35
45
35
35
30
25
Straight Line
q/mi tons/yr
0.0004
0.0002
*
850
*
480
* *
0.0029 6,060
0.0058 12,110
Curve
g/mi tons/yr
0 0006
0 .0003
0.0031
0 0055
1, 180
*
680
6 ,400
11,430
Base from which emission reductions are determined
-------
6-35
Table 6-17
Fuel
Gasoho1
Methanol
Emission Reductions Obtainable
Through Control ot oibo <->£ In-Use Fuel:
(Multiplicative Method)
F ma 1
45
15
45
35
FLeetwide Emission Reductions in 1990
Straight Line
g/mi tons/yr
0 0039
6 , 660
0 0022 3,800
Curve
g/mi tons/yr
* *
0.0057 9,740
0.0032
*
5, 570
All Fuels
35
30
25
0.0277
0.0553
47,560
95,090
0.0313
0 0531
53,860
91,300
Fuel
Final oii.o
Fleetwide Emission Reductions in 2010
Straight Line
g/mi tons/yr
Curve
g/mi tons/yr
Gasohol
45
35
0 0009 1,800
0 0013
*
2 , 640
Methanol
45
35
0.0005 1,030
0.0007
*
1,510
All Fuels
35
30
25
0.0062 12,900
0 0124 25,790
0.0070 14,610
0.0119 24,760
Base from which emission reductions are determined.
-------
b-36
Table b-18
1^0 Cost: Effectiveness of Reducing ? i n of In-Use Fuels
Final
POOi
~ost
L 3 S ' j i )
Emissions Reductions
St: aiqht Line Oii ve _
_ ' ^ o i, % v: ) (_' o u./y )
Cost
E C_L e rt_ive nc_s s '
St t a 1 qht.
Lino C_u 1 vp
( $ i (• i' ) ( S / t on )
Gasohol 45
35
34 . 4
33.7
*
23.7
130-
¦ ')T
*
4 , 31) 0 -
'i, 740
¦k
3, r"j0-
/ , SoO
*
2.4 30-
'i,44')
Methanol
45
35
35.0
34 . 0
*
13.5
~
7 '• 0 -
-:0 i
*
2,490-
5, 570
~
3,50 0 -
7 , ',¦)()
*
2,4 30-
'i,'!1 )
All
Fuels
35
30
33.3
26.2
30',
387
*
22, 3 50¦
4 7 , 5'>0
*
2 3 , '3 L 0
53,860
, 4 90 -
17,350
5, 730
16,420
25
23.2
¦">34.0-
7 84 . 1
•i 4,1,', 0
f<5, 0 90
42 , 140-
91,300
<5 , ') 7 0
17,560
0, 94 0
18,blO
* Base from which costs, emission reductions, and cost effectiveness
results are determined.
1 Based on M0BILE3 fuel consumption monel prediction of gasoline sales of
79.901 x 10'' gallons in 1990.
2 The range of emission reductions listed is for the range from the
additive method to the multiplicative method.
3 The range of cost effectiveness only lists the minimum and maximum ends
of the range.
-------
6-37
Table 6-19
2010 Cost Effectiveness of Reducing %ieo of In-Use Fuels
Emissions Reductions
Cost
Effectiveness
Final
Fuel %iso
Pool
^16 0
Cost'
(106$/yr)
Straiqht Line
Curve
Straight
Line Curve
(tons/yr)
(tons/yr)
($/ton) ($/ton)
Gasohol 45
34.4
*
*
*
*
35
33.7
21.2
850-
1,180-
11,760-
8,'
1.800
2, 640
25,070
17,
Methanol 45
35.0
*
*
*
*
35
34.6
12.1
480-
680-
11,760-
8,<
1,030
1, 510
25,070
17.
All
Fuels 35
33.3
*
*
*
*
*
30
28.2
275.7-
6.060-
6,400-
21,370-
18,
346.4
12,900
14,610
57,160
54,
25
23.2
566.5-
12,110-
11,430-
21,970-
22,
700.6
25,790
24,760
57,850
61,
* Base from which costs, emission reductions, and cost effectiveness results
are determined.
1 Based on M0BILE3 fuel consumption model prediction of gasoline sales of
71.395 x 109 gallons in 2010.
2 The range of emission reductions listed is for the range from the additive
method to the multiplicative method.
3 The range of cost effectiveness only lists the minimum and maximum ends of
the range.
-------
6-38
References (Chapter 6)
1. "Draft Regulatory Impact Analysis: Proposed
Refueling Emission Regulation for 1990 and Later Model Year
Gasoline - Fueled Motor Vehicles, Volume I, Analysis of
Gasoline Marketing Regulatory Strategies." EPA, Office of Air
Quality Planning and Standards and Office of Mobile Sources.
(Available in public docket A-87-11.)
2. "MOBILE3 Fuel Consumption Model," Mark A. Wolcott,
EPA, and Dennis F. Kahlbaum, CSC, February 1985,
EPA-AA-TEB-EF-8 5-2.
3. "The Motor Fuel Consumption Model: Twelfth
Periodical Report," Energy and Environmental Analysis, Inc.,
For U.S. Department of Energy, November 1985.
4. "Fuel Consumption Fraction in Ozone Problem Areas
and Impact on Onboard Cost-Effectiveness Analysis," EPA
memorandum from Terry P. Newell to Charles L. Gray, Jr., July
7, 1986.
5. "Public Hearing on the EPA Study of Gasoline
Volatility and Hydrocarbon Emissions from Motor Vehicles," Ann
Arbor, MI., February 4-5. 1986. (Public Docket No. A-85-21,
Entry II-F-1.)
6. "Cost and Feasibility of Gasoline Volatility
Reductions," Sobotka and Company, Inc., for EPA Office of
Policy Analysis, April 21, 1986.
-------
Appendix 6A
Selection of a Certification Fuel RVP
The selection of the severity of conditions for the FTP
for evaporative emissions is of critical importance, as this
will determine the capacity of the evaporative emission control
system for all new gasoline-fueled vehicles. With respect to
exhaust emissions it has been the goal to set the conditions of
the FTP to be representative of what vehicles experience in the
real world. However, since control systems for evaporative
emissions are not proportional control devices (i.e. once
canister breakthrough conditions are reached none of the
subsequent evaporative emissions are trapped until such time as
the canister is purged), controlling emissions from vehicles
experiencing conditions of "average" severity is not
sufficient. Therefore, it is appropriate to design the FTP to
represent more "worst case" conditions.
When comparing the temperature and fuel tank level
conditions of the current FTP to m-use conditions, it appears
that the FTP resembles ASTM Class C areas fairly well. Using
the methodology outlined in Chapter 2, the Uncontrolled Diurnal
Indices (UDI) for the 61 non-California urban ozone
non-attainment areas were calculated using design value day
temperatures and ASTM fuel RVP with average vehicle conditions
(i.e., 61 percent full fuel tank). The UDI for the FTP was
calculated using the temperatures and fuel tank level of the
diurnal test procedure and fuel RVP levels of 9.0 (Class A),
10.0 (Class B) and 11.5 psi (Class C). The results of these
calculations are shown in Table 6-A-l.
The analysis shows that, for 9.0 and 10.0 RVP fuel, the
UDI of the FTP is less than that calculated for the great
majority of the cities (59 and 53 out of 61, respectively).
Using Class C fuel (RVP=11.5), the FTP UDI is much more
representative of the UDIs for the 61 non-attainment cities,
(28 out of 61 UDIs are greater than that of the FTP). Hence,
setting certification fuel volatility specifications to match
ASTM Class C fuel volatility specifications appears to be the
most reasonable approach. However, since it is desirable to
cost effectively control as many emissions as possible, ideally
an analysis of the cost effectiveness of increasing
certification test stringency would identify the point at which
the increased emission reductions achieved are no longer worth
the cost.
The ability to perform such an analysis is limited by our
current model of evaporative emissions, notably the treatment
of driving cycles. For instance, many vehicles may experience
only partial diurnals each day, being interrupted by driving
(and therefore purging of the canister) during the day. Other
vehicles may not be driven for periods of two days or more,
-------
6-A-2
during which time multiple diurnals are experienced and the
evaporative canister may become saturated with hydrocarbons.
Some vehicles are driven consistently on long trips resulting
in complete cannister purge, while others are consistently
driven for very short periods, where the ensuing hot-soaks may
always exceed the cannister purge. In addition to driving
cycles, at this time it is difficult to address the impact the
presence of an onboard refueling vapor recover canister would
have on the results of this analysis.
These limitations are much more relevant in an analysis of
certification test stringency than in assessing the point at
which to match in-use and certification fuel. Increasing
certification fuel RVP and, therefore, design stringency only
affects the emissions from vehicles with operational control
systems, albeit the great majority of vehicles. Since canister
purge plays an integral role in the operation of any
evaporative control system, and purge is strongly dependent on
driving cycle. The in-use efficiency of the control system is
very dependent on vehicle driving cycle. This deficiency is
not too serious for estimating average emissions since the
certification procedure simulates these conditions directly.
However, the deficiency increases as one departs from the
average and moves toward those conditions where evaporative
emissions are high, such as those mentioned above. Since the
increased capture of these highly uncontrolled emission
situations is precisely the focus of a cost effectiveness
analysis of certification test stringency, the potential for
error due to the limitations of the current evaporative
emission procedure are particularly high. Therefore, it must
be stressed that the analysis performed below must be
considered very preliminary.
The certification test can be made more stringent in a
number of ways. It can be done by decreasing the fuel tank
level of test vehicles, increasing the span of the diurnal heat
build, or by adjusting the certification fuel RVP. However,
since both diurnal and hotsoak emissions are best correlated
with fuel RVP, RVP was the parameter which was varied in this
analysis to simulate different levels of stringency for the FTP.
The first step of the cost effectiveness analysis was to
determine the cost of raising the certification fuel RVP from
the present level of 9.0 psi RVP up to as high as 12.5 psi RVP
(in increments of 0.5 psi RVP). The vehicle control costs for
the different certification fuel RVP levels could be taken
directly from Table 4-3.
-------
6-A-3
In the second step, average non-tampered emissions were
calculated on a per vehicle basis for each certification fuel
scenario for the nine largest, non-California U.S. cities*, and
were then weighted according to population and combined to
obtain an overall average emission value. All emission
calculations were performed using ozone non-attainment design
value temperatures and a 61 percent full fuel tank (as derived
in Chapter 2, Section III-A-ld), while assuming that in-use
fuel RVP matched ASTM summer specifications (e.g.. Class C =
11.5 RVP). From these emission calculations, emission
reductions for each incremental increase in fuel RVP were
determined.
A fuel recovery credit was also calculated for each
certification fuel RVP level, due to the expected increased
efficiency of vehicle control systems with increasing
certification fuel RVP. These fuel recovery credits were
determined using average July temperatures for each city and
assuming a 53 percent full fuel tank (see Chapter 2, Section
III-A-le). Two scenarios were used, one in which it was
assumed that the fuel recovery credit was experienced over the
entire year, and one in which the fuel recovery was assumed to
take place only five months of the year (i.e., only during the
summer months when evaporative emissions are well understood
and presumably higher).
Cost effectiveness calculations were then performed
assuming a ten year vehicle life. Emission reductions and fuel
recovery credits were discounted over the ten year period using
both a 0 percent (which is analogous to the year 2010, steady
state analysis of Chapter 6) and 10 percent (standard for EPA
mobile source analysis) discounting rate. The results of this
analysis, including the fraction of nontampered emissions which
would be eliminated under each certification fuel scenario are
shown in Table 6-A-2. It should be pointed out that the cost
effectiveness numbers shown here differ somewhat from those
presented in the main body of Chapter 6. This is due to the
fact that the emissions model used in this appendix was
relatively simplistic compared to the emissions model used in
the other sections.
Focusing on the results of the analysis assuming a zero
percent discount rate and 5-month recovery credit, which are
most consistent with those of Chapter 6, it appears that
raising certification fuel from 9.0 psi RVP to 10.0 psi RVP
Miami was not used in the analysis as the ozone design
value day for this city was atypical in that it was a
winter day.
-------
6-A-4
would not only reduce emissions, but would also introduce a net
savings to consumers, since savings due to the fuel recovery
credit would outweigh the cost of improving the evaporative
emission control system on new vehicles. Raising certification
fuel volatility further, up to 11.5 psi RVP (independent of the
fuel recovery credit timeframe), would produce a net consumer
cost (i.e. the incremental manufacturing cost would outweigh
the incremental fuel recovery credit), but would still reduce
emissions at a relatively low incremental cost effectiveness
(less than $200 per ton). Within the accuracy of the emissions
model this confirms that setting certification fuel to match
m-use fuel in Class C areas is reasonable. Since it appears
that emissions reductions would be achieved at moderate
expense, one could conclude therefore that the test procedure
was not being made too stringent. In addition, once
certification fuel RVP is increased above 11.5 psi RVP, it
appears that emission reductions would become increasingly
costly, but would still not be unreasonable, on the order of
$500-1000 per ton. However, due to the aforementioned
limitations in the analysis, particularly the treatment of
driving cycles, no confident conclusion can be drawn at this
time.
It should also be pointed out that the results of this
analysis are influenced by certain assumptions contained in the
equations used to model emissions. Reasonably accurate
equations exist which correlate diurnal emissions with fuel
RVP, fuel temperature increase, and fuel tank vapor space
volume. Hot soak emissions, however, can currently only be
correlated with in-use fuel RVP. According to the model, once
certification fuel RVP is raised past the in-use level (11.5
psi here), hot soak emission reduction through increasing
certification RVP is no longer possible. The sharp increase in
the cost of emission reductions once certification RVP exceeds
11.5 psi (see Table 6-A-2), therefore,is the result of this
aspect of the model.
One other useful result shown in Table 6-A-2 is the
fraction of current non-tampered evaporative emissions which
would be eliminated by raising the certification test procedure
RVP to various levels. Since many cities currently are not in
compliance with ozone regulations, it may be fair to say that
the current test procedure conditions fall considerably short
of the desired level of control. Looking at the fraction of
evaporative emissions eliminated in conjunction with the cost
effectiveness of each certification test stringency scenario
may help in determining how stringent the FTP for evaporative
emissions ought to be.
-------
A. 5 Month Fuel Credit
Initial
New
Additional
Emission
Oect
Cert
Manufacturer
Reduction
KVP
KVP
Cost <$)
(q/yr)
9.0
9.5
0.59
1266.9
9.5
10.0
0.62
924.8
10.0
10.5
0.62
674.2
10.5
11.0
0.66
487.2
11.0
11.5
0.70
346.5
11.5
12.0
0.84
125.4
12.0
12.5
0.86
77.6
B. 12 Month Fuel
Credit
Initial
New
Additional
Emission
Cect
Cect
Manufacturer
Reduction
KVP
RVP
Oost ($)
(q/yr)
9.0
9.5
0.59
1266.9
9.5
10.0
0.62
924.8
10.0
10.5
0.62
674.2
10.5
11.0
0.66
487.2
11.0
11.5
0.70
346.5
11.5
12.0
0.84
125.4
12.0
12.5
0.86
77.6
Table 6-A-2
Posts Effectiveness of increasing certification Fuel kvp
(pec Vehicle Basis!
Per- Discounted
centage
Emission
Discounted
incremental
of Present
Fuel
Fuel
Reductions
Fuel
Oast
Emissions
Credit
Credit
(grams)
Credit ($)
Effectiveness (t/ton)
eliminated
(g/yrI
($/yr)
0% Disc.
10% Disc.
0% Disc. 10% Disc.
0% Disc.
10% Disc.
20.3
354.95
.0937
12669.0
8173.8
.937 .6045
-24.8
-1.6
35.1
258.61
.0683
9248.0
5966.7
.683 .4404
- 6.2
27.3
45.9
187.26
.0494
6742.0
4349.7
.494 .3189
17.0
62.8
53.6
133.61
.0353
4872.0
3143.4
.353 .2275
57.2
124.8
59.2
93.13
.0246
3465.0
2235.6
.246 .1586
118.9
219.7
61.2
14.91
.0039
1254.0
806.8
.039 .0254
579.5
913.7
62.4
5.77
.0015
776.0
500.7
.015 .0098
987.8
1540.4
Per-
Discounted
centage
Emission
Discounted
Incremental
of present
Fuel
Fuel
Reductions
Fuel
Oost
Emissions
Credit
Credit
(grams)
Credit ($)
Effectiveness ($/ton)
Eliminated
(g/yr)
<$/yr)
0% Disc.
10% Disc.
0% Disc. 10% Disc.
0% Disc.
10% Disc.
20.3
851.88
.2249
12669.0
8173.8
2.249 1.4510
-118.8
-95.6
35.1
620.67
.1638
9248.0
5966.7
1.638 1.0568
- 99.9
-66.4
45.9
449.41
.1186
6742.0
4349.7
1.186 .7652
- 76.2
-30.3
53.6
320.66
.0846
4872.0
3143.4
.846 .5461
- 34.6
32.9
59.2
223.52
.0590
3465.0
2235.6
.590 .3807
28.8
129.6
61.2
35.79
.0094
1254.0
808.8
.094 .0610
539.7
873.8
62.4
13.84
.0037
776.0
500.7
.037 .0236
962.1
1515.4
-------
6-A-6
Table 6-A-l
61 Non-California Urban Ozone Non-Attainment
Areas and Associated Design Value Day UDIs
Area
ASTM Class
UDI
EPA Region l
Boston Metro Area
Greater Connecticut Metro Area
New Bedford, MA
Portland, ME
Portsmouth-Dover-Rochester, NH-ME
Providence, RI
Springfield, MA
Worcester, MA
EPA Region 2
Atlantic City, NJ
New York Metro Area
Vineland-Millville-Bridgeton, NJ
EPA Region 3
A1lentown-Bethlehem, PA
Baltimore, MD
Erie, PA
Harrisburg-Lebanon-Carlisle, PA
Lancaster, PA
Philadelphia Metro Area
Pittsburgh, PA
Reading, PA
Richmond-Petersburg, VA
Scranton-Wilkes Barre, PA
Washington, DC-MD-VA
York, PA
EPA Region 4
Atlanta, GA
Birmingham, AL
Charlotte-Gastonia-Rock Hill, NC-SC
Chattanooga, TN-GA
Huntington-Ashland, WV-KY-OH
Louisville, KY-IN
Memphis, TN-AR-MS
Miami-Hialeah, FL ***
Nashville, TN
Tampa-St. Petersburg-Clearwater, FL
C
C
C
C
D
C
D
C
C
C
C
C
C
C
C
C* */D
C
C
C
D/C**
D
C
C
C/B**
C
B
B
C
C
B* */C
C
C
C
3 .017
1 . 649
1 .603
0 .838
2 .285
2 . 521
3 .309
1 . 730
2.915
1 .976
2.310
2
2
1
1
1
1
1
1
1
2
1
2
462
813
505
722
297
665
594
858
773
661
432
045
619
470
840
257
274
263
240
198
,772
, 603
-------
6-A-7
Area
Table 6-A-l (cont)
ASTM Class
UDI
EPA Region 5
Akron, OH
Canton, OH
Chicago Metro Area
Cincinnati Metro Area
Cleveland, OH
Dayton-Springfield, OH
Detroit, MI
Grand Rapids, MI
Indianapolis, IN
Milwaukee Metro Area
Muskegon, MI
EPA Region 6
Baton Rouge, LA
Beaumont-Port Arthur, TX
Brazoria, TX
Dallas-Fort Worth, TX
El Paso, TX
Galveston-Texas City, TX
Houston, TX
Lake Charles, LA
Longview-Marshal1, TX
New Orleans, LA
San Antonio, TX
Tulsa, OK
EPA Region 7
Kansas City, MO-KS
St. Louis, MO-IL
EPA Region 8
Denver-Boulder, CO
Salt Lake City-Ogden, UT
* * *
C
C
C
c
c
c
c
c
c
c
c
B**/C
B
B
B
A
B
B
C/B**
B
C
B
B
B
B
B**/A
B* */A
432
500
270
361
087
034
517
806
684
510
929
EPA Region 9
Phoenix, AZ
1. 102
1. 619
1 . 606
1 . 535
1 . 196
0 . 750
1 . 181
1.063
1. 798
2.356
1.673
2. 524
1 . 490
1. 533
2.447
2 . 102
2. 168
Certification Temperature A(9.0) 1.000
& Tank Level Conditions B(10.0) 1.309
C(11.5) 1.958
* Assuming a 61% full gas tank.
** ASTM Class assumed in analysis.
*** Based on highest ozone day during May to September period
(measured at same monitoring station at which the ozone
design value occurred).
-------
Appendix 6-B
Evaluation of an Inspection/Maintenance
Program for Evaporative Emission Control Systems
This appendix details the calculations used to derive the
benefits and cost effectiveness of an in-use vehicle inspection
and maintenance (I/M) program for evaporative emission control
systems. The program is assumed to begin in 1988. This was the
earliest feasible implementation date for evaporative I/M.
Application was restricted to 1978 and later model year LDVs and
LDTs. Earlier vehicles were not certified using the
comprehensive SHED test and so their evaporative emission
control systems are not very effective and are not amenable to
cost-effective repair. Heavy-duty vehicles were not included
since many current I/M programs for exhaust emissions do not
inspect HDVs.
This appendix is arranged into six sections describing how
the emissions benefits, costs and cost effectiveness of
evaporative I/M were estimated. The sections are:
Sources of Excess Evaporative Emissions
Evaporative I/M Program Design
- Effectiveness of Evaporative I/M
- Evaporative I/M Program Scenarios
- Evaporative I/M Costs
Evaporative I/M Cost Effectiveness
The first section describes how the excess emissions of
evaporative systems were estimated. The second section
describes how an evaporative inspection would be performed. The
next section discusses the expected effectiveness of evaporative
I/M programs. The fourth section describes the two I/M programs
analyzed in this appendix. The fifth section presents estimates
of the costs associated with evaporative I/M programs. The last
section calculates the cost effectiveness for 1988 and for 2010,
representative of the long-run.
Sources of Excess Evaporative Emissions
The emission reductions obtainable through an evaporative
I/M program were based on results from EPA's in-use emission
factors (EF) test program, which is described in Chapter 2.
Tables 6-B-l and 6-B-2 present the types of malmaintenance and
defect (M&D) problems discovered in the EF test program and the
rate of occurrence of each problem for fuel-injected (FI) and
carbureted vehicles, respectively. Tables 6-B-l and 6-B-2 also
present the average diurnal and hot-soak emission effect
associated with each problem, as measured in the EF test program
vehicles.
-------
6-3-2
Table 6-B-l
In-Use EF Test Program M&D Types, Rates of Occurrence and
Average Emissions for Fuel-Iniected Vehicles (11.7 psi Fuel)
Vehicles
Vehicles
Vehicle
Evaporative
With
With
Average
Emissions
Control
Diurna1
Hot Soak
(grams/test)
Problem
Problems
Problems
Diurnal
Hot Soak
Gas Cap Leaks
4
4
20. 64
3 .35
Purge System
Plugged/Damaged
3
3
18 . 50
10 . 48
Purge Hose Disabled
— *
1
-
38. 15
Saturated Canister
2
2
16.33
2.11
Plugged/Dirty Filter
7
7
14 . 49
5 . 66
Canister Damaged
-
1
-
13 . 06
Air Cleaner Throttle
Body Gasket Missing
1
1
22 . 96
3 . 22
Average Emissions
Vehicles with
Any Problem
17
19
17 .36
7 . 53
Average Emissions
Vehicles with
No Problems
146
144
7. 56
2. 07
Average Emissions
Of All Vehicles
163
163
8 . 51
2. 70
Fleet M&D Effect
0 . 95
0 . 63
If the average emissions of vehicles with a problem was
less than the average of vehicle without problems; then
they were added to the without problem sample for that mode.
-------
Table 6-B-2
In-Use EF Test Program M&D Types, Rates of Occurrence and
Average Emissions for Carbureted Vehicles (11.7 psi Fuel)
Vehicles
Vehicles
Vehicle
Evaporative
With
With
Average
Emissions
Control
Diurnal
Hot Soak
(grams/test)
Problem
Problems
Problems
Diurnal
Hot Soak
Gas Cap Leaks
11
11
13 .87
6.61
Canister Hose Damaged
1
1
17 .18
3.91
Purge System
Plugged/Damaged
19
19
15.22
7 . 38
Saturated Canister
8
_ *
10 . 55
-
Plugged/Dirty Filters
4
4
8.66
3.23
Bowl Hoses Damaged
4
-
11. 73
-
Air Cleaner Throttle
Body Gasket Missing
1
-
17 . 67
-
Combinations
6
6
14 .48
10 .39
Average Emissions
Vehicles with
Any Problem
45
41
13 . 51
7 .11
Average Emissions
of Vehicles with
No Problems
110
114
8.63
3.32
Average Emissions
Of All Vehicles
155
155
10.33
4 . 26
Fleet M&D Effect
1. 70
0.96
If the average emissions of vehicles with a problem was
less than the average of vehicle without problems; then
they were added to the without problem sample for that mode.
-------
6-B-4
Although the percentages used for M&D problems in this
analysis are based on the observed rates in the sample, there is
no reason to believe that many of the problems observed in the
fuel-injected sample or the carbureted sample will occur only in
those groups. For example, the rate of leaking gas caps is
larger for carbureted than fuel-injected vehicles. There is
probably no good reason why the rates would be so different,
other than sampling variation. The emission levels themselves
are subject to this same sampling variability. As a result some
vehicles with identified M&D problems have emission levels which
are less than the average emissions of vehicles without
identified problems. Both of these factors make the estimates
of the potential emission reductions from the evaporative I/M
uncertain at best. Only more information on the extent of M&D
problems and the actual measured reductions from repairs of
those problems can help reduce this uncertainty. This analysis
will, however, provide a basis for discussion of the relative
value of the concept of evaporative I/M and whether more
investigation is warranted.
The percentage of the total M&D effect due to each defect
is contained in Tables 6-B-3 and 6-B-4 for fuel-injected and
carbureted vehicles, respectively. The percentages were
calculated using the following equation and then normalized
(All negative percentage contributions were assumed to be zero
before the total percentage was normalized.)
(Avg Evap Emissions Emissions With\ Rate of Percent
With a M&D Problem - No Problems ) x Occurrence = of Fleet
Overall of M&D M&D
Fleet M&D Effect Problem Effect
Evaporative I/M Program Design
There are two basic I/M program designs. Centralized
programs inspect all vehicles at a few large inspection stations
with multiple "lanes" so that more than one vehicle can be
inspected at once. These stations are usually run by the local
I/M agency or by a contractor. Repairs are not done at the
inspection station. The inspection in centralized programs must
be done quickly in order to avoid long waiting lines.
Decentralized I/M programs license independent garages or repair
facilities to conduct inspections. Usually, the vehicles are
both inspected and repaired at the time of the inspection.
Evaporative emissions caused by malmaintenance and defects
(M&D) can usually be detected by a careful functional and visual
inspection of the system. To detect a leaky gas cap, the fuel
tank is sealed off and pressurized through its connection to the
charcoal canister. A drop in pressure noted with a pressure
gauge indicates a possible leaking gas cap. The exhaust
-------
6-B-5
Table 6-B-3
Percentages of Total M&D Effect Due to
Specific M&D Problems on Fuel-Iniected Vehicles
Percentages of Total M & D
Problem Diurnal Hot Soak
Gas Cap Leaks* 31.35 4.95
Purge System Plugged/ 19.69 24.30
Damaged
Purge Hose Disabled* - 34.75
Saturated Canister 10.54 0.10
Plugged/Dirty Filter 29.20 24.22
Canister Damaged* - 10.54
Air Cleaner Throttle 9.22 l.ll
Body Gasket Missing
Total 100.00 100.00
Total Percentage 51.04 74.54
Addressable with
Evaporative I/M
* Problem is addressable with MS>D Evaporative I/M
-------
6-B-6
Table 6-B-4
Percentages of Total M&D Effect Due to
Specific M&D Problems on Carbureted Vehicles
Percentages of Total M & D
Problem Diurnal Hot Soak
Gas Cap Leaks*
21.87
23 . 14
Canister Hose Damaged*
3.24
0.42
Purge System Plugged/
47 . 52
49 . 43
Damaged
Saturated Canister
5.83
-
Plugged/Dirty Filters
0 . 05
0 . 0
Bowl Hoses Damaged*
4 . 72
-
Air Cleaner Throttle
3 . 44
-
Body Gasket Missing
Combinations
13 .33
27 . 01
Total
100.00
100.00
Total Percentage
77.35
72. 99
Addressable with
Evaporative I/M
Problem is addressable with M&D evaporative I/M.
-------
6-B-7
emission analyzer can also be used to "sniff" for leaks while
the system is pressurized. To detect a broken/missing gasket,
propane is sprayed around the intake manifold in an engine at
idle. An increase m engine RPM when the propane is applied
would indicate a broken/missing gasket. Broken or cracked
canisters and damaged hoses are detected visually.
Leaking gas caps, missing or broken carburetor gaskets,
broken canisters and damaged hoses are all assumed to be
detectable. Tables 6-B-3 and 6-B-4 contain the maximum
percentage of the M&D effect addressable through an evaporative
I/M program broken down into the diurnal and hot-soak components.
Any gas caps, canisters and connecting hoses which have
been removed or disconnected are considered to be tampered.
Tampering can usually be detected by a quick visual inspection
of the system. Table 6-B-5 contains rates of tampering observed
in recent EPA tampering surveys.[7]
Effectiveness of Evaporative I/M
In operating I/M programs, the actual effectiveness of
tampering inspections in eliminating the emission impacts of
tampering is less than 100 percent. This is due to factors such
as sloppy inspections, fraud and retampenng between
inspections. It is assumed that inspections for evaporative
system tampering can be up to 70 percent effective when
conducted in centralized I/M programs.[6] Actual experience
with decentralized underhood inspections, however, suggest that
an evaporative I/M inspection as part of a decentralized program
may be only about half as effective. This observed difference
in effectiveness is thought to be caused by differences in
inspector training, motivation and oversight in centralized
versus decentralized programs. Benefits from inspection of
vehicles for tampering are estimated by reducing the rate of
tampering and then recalculating the emission impact of the
remaining tampering. The emission impact of evaporative system
tampering has already been discussed in Chapter 2.
Table 6-B-6 contains the maximum portion of M&D effects
addressable through an evaporative I/M program which is 100%
effective. Table 6-B-6 also shows the portion of M&D effects
addressable through evaporative I/M programs which are 70%
effective. For this analysis, functional M&D inspections in I/M
programs will be assumed to be only as effective in identifying
and correcting evaporative system problems as visual tampering
inspections.
Evaporative I/M Program Scenarios
Two scenarios for annual inspections of the evaporative
system in I/M programs were investigated. The first assumes
that thorough inspection for M&D problems combined with an
evaporative system tampering check can be accomplished m
-------
6-B-8
Table 6-B-5
Types of Tampering Problems
And Typical Rates of Occurrence
Problem Rate of Occurrence (%)
Gas Cap Removed 1.2%
Canister Vacuum Disconnected 1.7
Cap Removed & Canister Vacuum Disconnected 0.1
Canister Removed 0.3
Non-vacuum Canister Disconnection 0.2
Total Disablements 3.5%
Rates calculated from the combined data from the EPA
Tampering surveys performed in 1982, 1983 and 1984 (9,142
vehicles).
-------
6-B-9
Table 6-B-6
Portion of M&D Effects
Addressable
Through M&D Evaporative I/M
g 100% Effectiveness @ 70% Effectiveness
Vehicle Type Diurnal Hot Soak Diurnal Hot Soak
Fuel Injected 77% 73% 54% 51%
Carbureted 51% 75% 36% 52%
-------
6-B-10
centralized I/M programs (Case #1). This represents the most
optimistic case for evaporative I/M. The second scenario
assumes that there is only the visual inspection for evaporative
system tampering (no M&D checks) and that this inspection occurs
in a decentralized setting (Case #2). This case represents a
program design more likely to be implemented. The two scenarios
show a range of possible reductions possible from an evaporative
I/M program.
Although the first optimistic scenario (Case #1) is
included in the analysis it is not likely to be implemented in
any of the existing I/M areas. Existing centralized inspection
programs were designed with an expected vehicle inspection
throughput and inspection times as short as two minutes per
vehicle. Any change in the inspection procedure which would
result in more time for each inspection would cause long lines
at the inspection stations and angry vehicle owners. It would
be necessary to build more inspection lanes at each station or
build new stations to absorb longer inspection times. The
required evaporative system pressure checks and propane leak
checks also generate concerns about liability for accidents,
accidental damage to vehicles and possible negative public
reaction. It is not likely that any State agency responsible
for an I/M program would proceed with an evaporative I/M
approach with these program elements without considerable
pressure from EPA or environmental groups. As long as there are
other, more easily implemented options available, centralized
I/M program planners will likely avoid adding an evaporative I/M
program of this type.
Many I/M programs are considering changing the frequency of
their inspections to biennial (every other year). Biennial
inspections would reduce the number of inspections required each
year and allow inspection times at centralized inspection
stations to be lengthened without building new inspection
facilities. However, the inspections necessary in Case #1
described above will more than double the inspection time in
most centralized programs. This will more than offset the
reduction in the inspection load from switching from an annual
to a biennial inspection frequency. Also, changing inspection
frequency will not address the liability concerns and will
reduce the benefits of both the evaporative and exhaust emission
portions of the program.
An underhood tampering inspection in a decentralized
inspection program (Case #2), however, can be done and is
currently being done in some areas. Liability is not as much a
concern for this scenario since the inspection is strictly
visual. Also, requiring repairs to tampering is not considered
as controversial since some vehicle owner responsibility is
assumed.
-------
6-B-ll
M0BILE3 emission factors were determined based on the
predicted percentage reductions in M&D emissions and tampering
occurrences. The MOBILE3 M&D emission components (i.e.,
carbureted hot-soak and diurnal, fuel injected hot-soak and
diurnal) were reduced by the percentages listed in Table 2-B-6
for the centralized (Case #1) I/M program. The M0BILE3 M&D
components were not changed for the decentralized (Case #2) I/M
program. The M0BILE3 LDV and LDT tampering rates (on 1978 and
later vehicles) were reduced 70 percent or 35 percent depending
upon which I/M program was being evaluated. The MOBILE3 results
were then used to determine the emissions inventory according to
the methodology described in Chapter 3.
An evaporative I/M program would only be practically
instituted in areas with existing I/M programs. These are best
represented by ozone non-attainment areas. VOC emission
reductions are shown in Table 6-B-7 for 61 non-California ozone
non-attainment areas obtained with an evaporative I/M program
assuming the current in-use RVP or ASTM limit, whichever RVP is
higher for each city. For the year 2010, a case is also
presented which assumes volatility control to 9.0 psi in ASTM
Class C areas. This assumes proportional reductions (from 22
percent) in ASTM non-Class C areas. For MOBILE3 modeling, 15 of
the 61 areas, including the 10 largest areas and five other
geographically diverse areas, were chosen to represent the 61
non-attainment areas. The use of only these 15 areas should not
significantly affect the results of the analysis.
Evaporative I/M Costs
The costs of an evaporative I/M program arise from the two
steps of an I/M program, the inspection and the repair of the
malmaintained, defective and/or tampered parts.
The cost per inspection assumes additional inspector time
per vehicle (in addition to the time required for an exhaust
inspection) at a labor rate of $20/hour. The increase in time
for a M&D inspection is primarily due to the procedure necessary
to check for leaky gas caps. When combined with an underhood
inspection for other M&D problems and for tampering, the
inspection is assumed to add six inspector minutes to the
existing inspection time or $2 per inspection. The underhood
tampering inspection alone is assumed to add three inspector
minutes or $1 per inspection.
Table 6-B-8 contains the estimated costs of the parts and
the amount of time necessary to carry out the repairs. The part
costs and labor time estimates are based on typical costs of
parts and labor time found in Mitchells Mechanical Parts/Labor
Estimating Guides. Labor costs are based on a basic shop fee of
$35/hour. The assumed repair costs associated with each problem
for fuel-injected and carbureted vehicles on both a repaired
-------
6-B-12
Table 6-B-7
Emission Reductions Obtainable with
Evaporative I/M Programs
In 61 Non-California Non-Attainment Areas*
(1000 tons/year)
Calendar Year
Assuming Current Case #1 Case #2
In-Use ASTM RVP Fuel* 1988 2010 1988 2010
Baseline (no program) Case 6998.7 7698.9 6998.7 7698.9
With Evaporative I/M 6935.7 7603.9 6986.7 7678 2
Program Benefit 63.0 95.0 12.0 20.7
Assuming Volatility
Control to 9.0 psi
in ASTM Class C**
Baseline (no program) Case
With Evaporative I/M
Program Benefit
Calendar Year
Case #1
2010
7028.4
6976.1
52.3
Case #2
2010
7028.4
7017.3
11.1
* *
The model assumes whichever RVP is higher for each city.
All scenarios assume a 9.0 psi RVP certification fuel.
Proportional reductions (from 11.5 psi) are assumed for
ASTM non-Class C areas.
Case #1 refers to an annual inspection of the evaporative
control system which includes a leak check and a thorough
underhood examination for tampering, cracked hoses and
damaged canisters in a centralized program.
Case #2 refers to an annual visual inspection of the
evaporative control system for missing and disabled
components (tampering) in a decentralized program.
-------
6-B-13
Table 6-B-8
Evaporative I/M Estimated
Parts and Labor Repair Costs
Evaporative
Part Replaced
Parts
Cost
(?)
Labor
Time
(hours)
Total
Cost
Gas Cap
Intake Gasket
Vacuum Hoses
Carburetor Gasket
Evaporative Canister
10
10
2
40
50
0.0
3.0
0.3
1. 4
0 . 5
10 . 00
115.00
12 . 50
89 . 00
67 . 50
Parts cost and labor time was estimated based on "Mitchell
Mechanical Parts/Labor Estimating Guide" for Domestic Cars
and Imported Cars and Trucks 1984. Published by Mitchell
Manuals Inc., San Diego, California, 1984. A basic shop
fee of $35.00/hour labor cost was used for all assumed
repairs.
-------
6-B-14
vehicle and average m-use vehicle basis are listed in Tables
6-B-9 and 6-B-10. Inspection costs and the combined inspection
and repair costs are shown as well.
The total first year repair cost will be greater than the
total second (and subsequent) year repair cost. This occurs
because during the first year of the evaporative I/M program
(1988), existing problems in all of the vehicles from model
years back to 1978 must be repaired. In the second and
subsequent years, only the cars which have had malmaintenance
and defect problems and/or have been tampered with within the
last year will need to be repaired.
If we assume that none of the M&D problems existed when the
vehicles were new, then based on the average age of the vehicles
an average rate for the occurrence of new problems could be
estimated. The average age of vehicles in the fleet is about
five years, therefore 20 percent of the observed problems would
occur every year if the rate of occurrence is assumed to be
linear. This is complicated by the fact that some of the
problems are likely to be the result of manufacturing defects
which should not be counted in the rate of new occurrences. It
is not possible to determine for certain if a M&D problem is a
defect or a result of in-use deterioration, therefore for this
analysis a reoccurrence rate of 15 percent will be assumed.
Thus, the second year (and subsequent) repair costs are 15
percent of the first year repair costs. The incremental
inspection cost remains unchanged since all vehicles must still
be inspected.
Tampering is assumed to also reoccur. Since deliberate
tampering is a behavior, it's reoccurrence cannot be assumed to
be random and therefore linear. The assumed penalty for
tampering, in this case simply the repair of the tampered item,
is not likely to be a strong deterrent factor. The reoccurrence
rate for tampering is assumed to be 25 percent for this
analysis. This includes both new occurrences on previously
untampered vehicles and actual reoccurrence on previously
tampered vehicles.
The combined repair and incremental inspection costs are
calculated in Table 6-B-ii and Table 6-B-12 separately for each
scenario. The resultant evaporative I/M costs for the
fuel-injected and carbureted vehicles are the same for the
tampering only inspection (Case #2).
An economic credit is realized from the emission repairs
required in the evaporative I/M program. The excess emissions
which would have been lost without repairs to the evaporative
control system will now be captured by the charcoal canister and
combusted in the engine. The economic credit is determined as
explained previously in Chapter 5, by assuming the composition
of the emissions is all butane. Chapter 5 contains the values
used to convert the butane to a gasoline equivalent and then to
economic credits (i.e., to develop the butane value of $290.32
-------
6-B-15
Table 6-B-9
Evaporative I/M Repair Cost
Per Vehicle Inspected
For Fuel-Iniected Vehicles
Repair
Repair Cost
MEtD Problem
Rate(%)
Cost($)
Per Vehicle(:
Leaking Gas Cap
3.6
10 . 00
$0.36
Leaking Gasket
3.6
115.00
4 .14
Broken Canister
1. 8
67 . 50
1 . 22
M&D
Repair Cost
per Inspection
$5. 72
Tamperinq Problem
Gas Cap Removed
1. 2
10.00
$0 . 12
Canister Vacuum
Disconnected
1. 7
12. 50
0 .21
Cap Removed S Canister
Vacuum Disconnected
0.1
22. 50
0 . 02
Canister Removed
0.3
67.50
0 . 20
Canister Mechanically
Disconnected
0.2
12 . 50
0 . 03
Tampering Repair Cost per Inspection $0.58
-------
6-B-16
Table 6-B-10
Evaporative I/M Repair Cost
Per Vehicle Inspected
For Carbureted Vehicles
Repair Repair Cost
M&D Problem Rate(%) Cost($) per Vehicle($)
Leaking Gas Cap
5. 5
10 . 00
$0 . 55
Damaged Vacuum Line
5 . 5
12 . 50
0 . 69
Damaged Vent Line
0 . 9
12 . 50
0.11
Leaking Carburetor
0.9
89 . 00
o
00
o
M&D Repair Costs per Inspection $2.15
Tampering Problems
Gas Cap Removed
1.2
10 . 00
$0 . 12
Canister Vacuum
Disconnected
1 . 7
12 . 50
0 .21
Cap Removed & Canister
Vacuum Disconnected
0 . 1
22 . 50
0 . 02
Canister Removed
0.3
67 . 50
0 . 20
Canister Mechanically
Disconnected
0 . 2
12 . 50
0 . 03
Tampering Repair Cost per Inspection $0.58
-------
6-B-17
Table 6-B-ll
Calculation of Total Evaporative I/M Cost
Per Vehicle Inspected
For Fuel-Injected Vehicles
Case ttl (Full Inspection)
First Year M&D Repair Cost (at 70% Effectiveness) $4.00
First Year Tampering Repair Cost
at 70% Effectiveness 0.41
Incremental Inspection Cost 2.00
First Year Cost per Vehicle $6.41
Subsequent Year M&D Repair Cost
with 15% Reoccurrence and at 70% Effectiveness $0.60
Subsequent Year Tampering Repair Cost
with 25% Reoccurrence and at 70% Effectiveness 0.10
Incremental Inspection Cost 2.00
Subsequent Year Cost per Vehicle $2.70
Case tt2 (Tampering Inspection)
First Year Tampering Repair Cost
at 35% Effectiveness $0.20
Incremental Inspection Cost 1.00
First Year Cost per Vehicle $1.20
Subsequent Year Tampering Repair Cost
with 25% Reoccurrence and at 35% Effectiveness $0.05
Incremental Inspection Cost 1.00
Subsequent Year Cost per Vehicle
$1.05
-------
6-B-18
Table 6-B-12
Calculation of Total Evaporative I/M Cost
Per Vehicle Inspected
For Carbureted Vehicles
Case ttl (Full Inspection)
First Year M&D Repair Cost (at 70% Effectiveness) $1.51
First Year Tampering Repair Cost
at 70% Effectiveness 0.41
Incremental Inspection Cost 2 . 00
First Year Cost per Vehicle $3.92
Subsequent year M&D Repair Cost
with 15% Reoccurrence and at 70% Effectiveness $0.23
Subsequent Year Tampering Repair Cost
with 25% Reoccurrence and at 70% Effectiveness 0.10
Incremental Inspection Cost 2.00
Subsequent Year Cost per Vehicle $2.33
Case #2 (Tampering Inspection)
First Year Tampering Repair Cost
at 35% Effectiveness $0.20
Incremental Inspection Cost l.00
First Year Repair & Inspection Cost per Vehicle $1.20
Subsequent year Tampering Repair Cost
with 25% Reoccurrence and at 35% Effectiveness $0.05
Incremental Inspection Cost l.00
Subsequent Year Cost per Vehicle
$1. 05
-------
6-B-19
per ton, or $0.32 per kilogram). Tables 6-B-13, 6-B-14 and
6-B-15 contain the resultant fuel recovery credits for the
evaporative I/M program for 1988 and 2010, respectively.
Evaporative I/M Cost Effectiveness
The total number of vehicles affected by an evaporative I/M
program were based on 61 non-California ozone non-attainment
areas. It was determined that 39.5 percent of nationwide VMT
occurs in these areas and, therefore, it was assumed that 39.5
percent of the nationwide fleet is found in the ozone
non-attainment areas. The total size of the nationwide fleet
was taken from estimates in the MOBILE3 fuel consumption
model.[8] Carbureted and fuel-injection projections back to
1978 (for the 1988 analysis), and back to 1990 (for the 2010
analysis) were taken from MOBILE3. The I/M cost without the
fuel recovery credit is a weighted-average of the total
inspection and repair cost for fuel-injected and carbureted
vehicles. The fuel recovery credit, as described in the
previous section, is subtracted from the inspection and repair
cost to obtain the I/M cost with the fuel recovery credit. The
I/M cost with the fuel recovery credit is the final cost of the
evaporative I/M program and is divided by the emission
reductions listed in Table 6-B-7 to determine the cost
effectiveness (C/E) of the evaporative I/M program. The
resulting C/E numbers are shown in Table 6-B-13, 6-B-14 and
5-A-15.
Tables 6-B-13 and 6-B-14 present the derivation of the C/E
of an evaporative I/M program in 1988 (first year) and 2010,
respectively, assuming current in-use or ASTM RVP fuel. The
first scenario analyzed assumes that a full inspection of the
evaporative control system is performed each year which will
identify both maintenance and defects and identify evaporative
system disablements and removals (tampering) in a program with
centralized inspection stations. The second scenario assumes
that evaporative I/M consists of only the visual inspection for
evaporative system tampering performed in a decentralized
(garage based) program. As was discussed earlier, these two
scenarios show the range of program benefits. The second
(tampering only) scenario, however, is the most likely to be
implemented if evaporative I/M were to be added to the existing
exhaust emission I/M programs. This is due to the impact a full
M&D and tampering inspection would have on the total inspection
time per vehicle in a centralized inspection program.
Both Tables 6-B-13 and 6-B-14 utilize costs and emission
reductions based on the vehicle population of 61 non-California
ozone non-attainment areas. This has little effect on the final
cost-effectiveness for any particular evaporative I/M program
since most areas' vehicles should be similar in vehicle mix and
emissions to this fleet. Therefore, the cost effectiveness
numbers could be applied to any city where an evaporative I/M
-------
6-B-20
Table 6-B-13
Cost Effectiveness of Evaporative I/M
1988 (First Year, No RVP Control)
Total number of vehicles affected by Evaporative I/M program*
FI Carb.
LDV 15.2 X 10b 22.8 X 10b
LDT 2.3 X 10s 7 1 x 1Q6
Total 17.5X106 29.9 X 10s
Cost of Inspection and Repair per Vehicle (first year)
Casetfl FI = $6.41 Carb. = $3.92
Case#2 FI = $1.20 Carb. - $1.20
Costs and Benefits
Case #1 Case #2
Cost of Inspection
and Repair $2.29 x 10a $0.57 x 10"
Fuel Recovery Credit** $0.18 x 108 $0.04 x 108
Total Costs & Credits: $2.11 X 108 $0.53 x 10s
Emission Reduction
due to Evap I/M*** 63,000 tons 12,000 tons
Cost Effectiveness: $3350/ton $4420/ton
* Based on the vehicle VMT of 61 non-California ozone
non-attainment areas, MOBILE3 fuel consumption model total
number of 1978 and newer model year vehicles, MOBILE3
vehicle registration distributions and MOBILE3 projections
of the number of carbureted and fuel injected vehicles.
** Fuel recovery credit was determined at $290.32/ton assuming
the recovered emissions were butane. The recovered
emissions were determined using MOBILE3 with average July
temperatures.
*** Emission reductions were determined using MOBILE3 with
ozone design value day temperatures.
-------
6-B-21
Table 6-B-14
Cost Effectiveness of Evaporative I/M
2010 (Subsequent Years, No RVP Control)
Total number of vehicles affected by Evaporative I/M program*
FI Carb.
LDV 53.6 x 10B 6.8 x 106
LDT 11.9 x 106 1.5 x 10s
Total 65.5 x 10G 8.3 x 10s
Cost of Inspection and Repair per Vehicle (after first year)
Casettl FI = $2.70 Carb. = $2.33
Case#2 FI = $1.05 Carb. = $1.05
Costs and Benefits
Case #1
Case #2
Cost of Inspection
and Repair
Fuel Recovery Credit**
Total Costs & Credits
Emission Reduction
due to Evap I/M***
Cost Effectiveness
$1.96 x 10'
$0.26 x 10'
$1.70 X 10'
95,000 tons
$1790/ton
$0.77 x 108
$0.06 x 10s
$0.71 x 10s
20,700 tons
$3430/ton
* Based on the vehicle VMT of 61 non-California ozone
non-attainment areas, MOBILE3 fuel consumption model total
number of 1978 and newer model year vehicles, MOBILE3
vehicle registration distributions and MOBILE3 projections
of the number of carbureted and fuel injected vehicles.
** Fuel recovery credit was determined at $290.32/ton assuming
the recovered emissions were butane. The recovered
emissions were determined using MOBILE3 with average July
temperatures.
*** Emission reductions were determined using MOBILE3 with
ozone design value day temperatures.
-------
6-B-22
Table 6-B-l5
Cost Effectiveness of Evaporative I/M
2010 (Subsequent Years, With RVP Control)
Total number of vehicles affected by Evaporative I/M program*
FI Carb.
LDV 53.6 x 10" 6.8 x 10b
LDT 11.9 x 10s 1.5 x 106
Total 65.5 x 106 8.3 x 10G
Cost of Inspection and Repair per Vehicle (after first year)
Case#l FI = $2.70 Carb. = $2.33
Case#2 FI = $1.05 Carb. = $1.05
Costs and Benefits
Case #1 Case #2
Cost of Inspection
and Repair $1.96 x 108 $0.77 x 10s
Fuel Recovery Credit** $0.13 x 108 $0.03 x I0a
Total Costs & Credits $1.83 X 108 $0.74 x 10®
Emission Reduction
due to Evap I/M*** 52,300 tons 11,100 tons
Cost Effectiveness $3500/ton $6670/ton
* Based on the vehicle VMT of 61 non-California ozone
non-attainment areas, MOBILE3 fuel consumption model total
number of 1978 and newer model year vehicles, MOBILE3
vehicle registration distributions and MOBILE3 projections
of the number of carbureted and fuel injected vehicles.
** Fuel recovery credit was determined at $290.32/ton assuming
the recovered emissions were butane. The recovered
emissions were determined using MOBILE3 with average July
temperatures.
*** Emission reductions were determined using MOBILE3 with
ozone design value day temperatures.
-------
6-B-23
program might be implemented. First year benefits from the full
centralized evaporative I/M program (Case #1) are 63,000 tons
per year VOC (0.9 percent of the total VOC inventory) with a
cost effectiveness of $3,350 per ton. The decentralized
tampering only program (Case #2) reduced VOC by 12,000 tons per
year (0.2 percent of the total VOC inventory) at $4,420/ton. By
the year 2010 Case #1 will reduce VOC by 95,000 tons per year
(1.2 percent of the total VOC inventory) at $1,790 per ton. In
2010, Case #2 will reduce VOC by 20,700 tons per year (0.3
percent of the total VOC inventory) at $3,430/ton.
Table 6-B-15 repeats the analysis for the year 2010, but
assumes volatility control to 9.0 psi in ASTM Class C areas and
proportional reductions in ASTM non-class C areas. This affects
both the benefits of evaporative I/M and the fuel recovery
credit, but does not affect the inspection and repair costs.
With these changes in 2010, Case #1 will reduce VOC by 52,300
tons per year (0.7 percent of the total VOC inventory) at $3,500
per ton and Case #2 will reduce VOC by 11,100 tons per year (0.2
percent of the total VOC inventory) at $6,670 per ton.
-------
CHAPTER 7
Analysis of Alternatives: Alcohol Blends
During the past several years commercial fuel has been
introduced which consists of gasoline blended with ethanol,
methanol, and other alcohols * Blending an alcohol into
gasoline increases the volatility of the final product,
making the potential increase in evaporative and exhaust
emissions a special concern. However, certain approaches
to volatility control could have serious economic impacts
on the fuel-alcohol industries, particularly on ethanol
blenders. This chapter evaluates alcohol blends separately
from straight gasoline and considers the desirability of
separate regulatory approaches for alcohol blends.
I. Industry Economics and Blend RVP
The fuel ethanol and methanol industries have
developed in ways that give them unique economic
characteristics. It is important to view the impact blends
have on HC emissions in the light of these characteristics.
A. Gasohol
Gasohol began entering certain fuel markets in the
late 1970's with the enactment of a tax credit applicable
to the federal gasoline tax. The credit has grown to 6
cents/gallon of gasohol (equivalent to 60 cents/gallon of
ethanol) and was the result of dual Congressional desires
to reduce U.S. dependence on imported oil and to make use
of excess agricultural production (specifically, corn). A
number of states have also enacted tax credits ranging from
one-half to 14 cents/gallon of gasohol.
New alcohol blends require an EPA waiver in order to
be introduced commercially (as provided in Section 211(f)
of the C ean Air Act). In 1978, when a waiver was
requested for gasohol, the Agency did not act to disapprove
the waiver application for gasoho] within the statutorily
required 180 days and the waiver for gasohol was
automatically granted under Section 211(f).
For convenience, alcohol blends containing only
ethanol are referred to as gasohol. In practice,
these blends rarely contain more or less than 9-10
percent ethanol. Any blend containing methanol, even
though ethanol may be present as a cosolvent, will be
referred to as a methanol blend.
-------
7-2
An important result of the automatic approval of the
gasohol waiver was that in accordance with the waiver
request no constraints were imposed on the volatility of
gasohol. This factor has in turn meant a great deal of
flexibility for the industry in the gasolines they can use
for blending with ethanol. Since final product volatility
is not regulated, ethanol blenders have been able to use
gasolines that were originally refined as final-use fuels.
These fuels are more easily and cheaply obtained than
specially-formulated low-RVP base fuels, and the ability to
use the standard finished gasoline has continued to play a
significant role in the development of the gasohol market.
The term "splash blending" is used here to describe this
practice of blending alcohols with "final-use" gasoline.
The practice of splash blending increases the RVP of
gasohol by up to about 1.0 psi as compared to straight
gasoline as discussed in Section II.B. of Chapter 2.
Another physical phenomenon which affects the actual
volatility effect of gasohol (or any alcohol blend) in the
field is called commingling. It occurs when vehicle fuel
tanks are intermittently filled with alcohol blends and
then with straight gasoline. The RVP of the resulting
mixture can be significantly higher than that of the blend
or the straight gasoline. At current market levels the
effect of commingling is equivalent to an overall increase
in in-use blend RVP of about 0.2 psi RVP for those vehicles
using some amount of gasohol, as discussed in Section II.B.
of Chapter 2.
Section II of Chapter 3 states that nationally,
gasohol accounts for about 7 percent of the gasoline
market, varying among states (because of different state
tax exemptions) from zero to more than 30 percent. About
80 percent of gasohol is sold in Midwest states (that is,
close to major corn producing areas). [1]
B. Methanol Blends
Several blends of methanol and gasoline have received
EPA waivers permitting their sale at one time or another.
The two waivered methanol blends which are expected to be
most common are an ARCO blend called "Oxinol" and a DuPont
blend. Each blend also includes a higher-alcohol
"cosolvent" (e.g , ethanol) to help prevent phase
separation in the presence of water. Oxinol is typically
4.75 percent methanol with 4.75 percent tertiary butyl
alcohol (TBA). The Dupont blend is 5 percent methanol with
2.5 percent ethanol or another higher alcohol.
-------
7-3
Unlike gasohol, these methanol blends are subject to
volatility limits as a provision of their EPA waivers. The
requirement is that the RVP of the final blend not exceed
the current ASTM limits for gasoline.* The volatility
limit ensures that evaporative and exhaust emissions from
methanol blend-fueled vehicles will not greatly exceed
emissions from gasoline-fueled vehicles.
Since adding methanol and a cosolvent to gasoline to
produce these blends increases RVP by about 2 psi (see
Section II. B. of Chapter 2) and since the RVPs of few
gasolines are 2 psi below the ASTM limits, these methanol
blends require specially formulated lower-RVP base
gasolines. Because of this, methanol cannot be "splash
blended" (as defined above).
At a 10 percent local market penetration by methanol
blends, the commingling effect would boost the effective
RVP for those vehicles using the blends by about. 0.7 psi
RVP, as discussed in Section II.B. of Chapter 2.
To date, only the ARCO "Oxinol" blend has actually
been marketed, achieving a 1.3 percent market share
nationwide during 1985 (mostly in the Northeast). Today
neither Oxinol nor the DuPont blend is available
commercially, in large part because the low price of crude
oil has removed much of their economic advantage.
II. Emission Effects of Blend Usage
Several factors interact to affect the evaporative and
exhaust VOC emissions of alcohol blends and their effect on
urban ozone formation (see Chapter 2, Section II.B. of this
report). The analysis in that chapter compares the effect
on ozone formation of VOC emissions from vehicle fleets
including some alcohol blend use with a gasoline-only fleet
at the same RVP. Certain factors would tend to reduce the
fleetwide ozone potential of both alcohol blends relative
to gasoline, including lower VOC exhaust emission and less
photochemically reactive evaporative emissions. However,
assuming 10 percent market shares for the blends, other
factors such as commingling offset these positive effects.
The resulting fleetwide increase or decrease in ozone
formation potential between gasoline and either gasohol or
methanol blends if RVPs are matched is less than 1 percent.
The DuPont blend originally had more stringent
requirements which have recently been relaxed (51 FR
39800, October 31, 1986).
-------
7-4
The RVP of current gasohol, however, is about 1.0 psi
higher than gasoline RVP, as stated above. Because of
this, blend-fueled vehicles have significantly higher
evaporative emissions than gasoline-fueled vehicles,
although their exhaust emissions are slightly lower. The
evaporative increase, however, more than offsets the
exhaust reduction for both ethanol and methanol blends.
The increase in ozone potential of evaporative plus exhaust
emissions from a gasohol-fueled vehicle is about 15 to 35
percent (depending on RVP level) compared to
gasoline-fueled vehicle emissions. On a fleetwide basis
(assuming a 10 percent market penetration by gasohol), the
increase in total ozone potential over a gasolme-only
fleet would be about 2 to 3 percent. (At a similar market
penetration, methanol blends would have about the same
fleetwide effect on ozone potential as gasohol.)
Ill. Effect of RVP Control on Alcohol Blends
A. Effect on Emissions
For each psi that alcohol blend RVP were reduced,
essentially the same VOC emission reductions would be
achievable as for gasoline. However, gasohol is currently
about 1.0 psi RVP higher than gasoline (and methanol
blends) as stated above. Thus if gasohol were controlled
to the same RVP as gasoline, about 2 to 3 percent
additional ozone control would be possible.
B. Effect on Refinery and Distribution Costs
As with emissions, refinery-level costs resulting from
alcohol blend RVP control would be similar to gasoline RVP
control costs, since reducing RVP would require about the
same effort for both final-use gasoline and for gasoline
base stocks for alcohol blends.
Costs might be slightly higher (roughly 0.5 cents/
gallon) in cases where lower RVP base gasolines were
required (as is the case today for methanol blends) because
the cost effects of RVP control can become more pronounced
at lower RVP levels (see Section III of Chapter 5).
Offsetting these higher costs, however, would be the
savings to refiners of making use of ethanol's octane value
as discussed in Chapter 5, Section III. Currently, much of
the octane-enhancing quality of ethanol is "wasted" in a
sense. This happens because with splash blending, existing
final-use gasolines that already have a sufficient octane
rating are blended with ethanol, further increasing octane
unnecessarlly.
-------
7-5
However, if refiners were to produce lower-RVP fuels
as blend stock for gasohol, the octane value of those fuels
would not need to be as high as for final-use fuels because
ethanol would be added. The blending could happen at the
refinery or later if the lower RVP, lower octane base fuel
were kept segregated during distribution. This ability to
take advantage of ethanol's octane value at the refinery
would be an economic benefit to refiners because the costs
of boosting octane through additives or additional refining
would be reduced. The value of this octane savings
(discussed in Chapter 5, Section III) would more than make
up for the cost of producing the lower-RVP base gasoline.
Despite this ability by refiners to take advantage of
ethanol's octane value, a requirement to meet the same RVP
as gasoline could introduce new distribution-related costs
to gasohol producers If this cost issue is ignored for
the moment, the discussion above indicates that both the
emission reductions and refinery-level control costs would
be about the same as those associated with gasoline RVP
control — and thus the cost per ton of RVP control for
alcohol blends would be almost exactly the same as for
gasoline. This means that absent the distribution effects,
cost effectiveness (i.e., cost per ton) is not a deciding
factor m determining whether to treat blends differently
from gasoline; absent such effects, EPA would control all
fuels to the same RVP level.
In reality, however, a requirement that gasohol meet
gasoline RVP limits would create serious new problems. As
stated above, gasoline is subject to non-binding ASTM
volatility limits, methanol blend RVPs are controlled to
typical gasoline levels, but gasohol is free from any
administrative volatility constraints. The gasohol
industry has relied heavily on their ability to market
their product at a higher RVP than the other fuels, and
hence to splash blend ethanol with gasoline.
If volatility controls required gasohol to meet the
same limits as gasoline, the gasohol industry might be
unable to obtain sufficient amounts of special reduced-RVP
gasoline stock; costs could be prohibitive in any case,
largely because separate distribution and storage
facilities would be needed in order to keep the special
stock segregated from higher RVP fuels. EPA contractors
have conducted three studies of this economic impact
[2,3,4], but the dependence of the result on local factors
such as number of storage tanks and their current
availability makes the task of developing reliable
estimates very difficult. Nevertheless, it is clear that a
major impact on gasohol sales and the industry as a whole
is possible.
-------
7-6
Another factor that could potentially affect gasohol
distribution should be mentioned: The state of Colorado is
considering a carbon monoxide emission control strategy
that could mandate the use of oxygenated fuels during the
winter months. If enacted, this program could expand
gasohol's current market. In addition, since distribution
of low-octane low-RVP gasoline to Colorado gasohol blenders
would probably increase, a transition to a summertime
gasohol RVP control program could conceivably be less
costly with these wintertime oxygenated fuel requirements
in place.
IV. Gasohol Control Options
Without the potential for major costs to the gasohol
industry, EPA would simply require that gasohol meet the
same RVP standards as gasoline (i.e., the same requirement
as for methanol blends). However, because of the public
policy concerns about the importance of a gasohol. industry
in the U.S., additional options have been developed and
evaluated. In light of the conclusions about gasohol, two
options that could apply to methanol blends are also
considered.
Each of the following gasohol control options were
considered:
Option 1. Continue the existing gasohol exemption
from RVP control.
Option 2. Instead of a total exemption, establish a
gasohol RVP standard 1.0 psi higher than that for gasoline
(i.e., a 1.0 psi RVP allowance).
Option 3, Require gasohol RVP to equal the RVP of
straight gasolines.
a. Apply this requirement nationwide.
b. Apply requirement only to gasohol sold in ozone
non-attainment areas; provide a 1.0 psi RVP
allowance in attainment areas.
c. Delay requirement until 1993; provide a 1.0 psi
RVP allowance in the interim.
The following paragraphs discuss each gasohol option.
Option 1. Administratively, this option represents
the current situation—gasohol blenders could continue to
-------
7-7
use commercial gasolines for base stock. For this reason,
it is unlikely that the ethanol industry would experience
any adverse economic impact.
However, if EPA required gasoline refiners to
significantly reduce the volatility of their fuels refiners
would have surpluses of certain very high RVP gasoline
components and a desire to market them. If gasohol
continued to have no RVP restrictions, it is very likely
that more of these very high RVP components would be used
as gasohol base stock, a current but not yet widespread
practice. Although gasohol today averages 1.0 psi RVP
above gasoline RVP or less, this practice could boost
gasohol RVP much higher, with an associated increase in VOC
emissions. Another result would be to make ethanol more
attractive and thus more valuable as a gasoline extender,
possibly increasing gasohol usage. The loss of emission
control associated with a gasohol RVP exemption under an
in-use RVP control program would probably be even greater
than what would occur under the similar Option 2 below.
Gasohol industry representatives are not strongly
advocating this option.[5]
Option 2. A l.o psi RVP allowance for gasohol, like
the exemption, would continue to allow splash-blending
(ethanol adds up to 1.0 psi to gasoline RVP) and hence
would not economically threaten the gasohol industry.
Unlike the exemption, the fixed allowance would eliminate
the ability to market very high RVP gasoline as gasohol
base stock.
The emissions impact of allowing a 1.0 psi RVP
disparity between gasoline and gasohol, as with the total
RVP exemption, is substantial. We estimate that at least 2
to 3 percent of the potential fleetwide ozone control
achievable in the year 2010 by reducing Class C gasoline
RVP to 9.0 psi could be lost because of gasohol" s higher
RVP (see Section II above). In areas with high gasohol
market shares, the emissions effect would be greater — a
very significant 2 percent of all Chicago VOC emissions,
for example. Appendix 7-A contains a detailed derivation
of the effect of the gasohol RVP allowance on the Chicago
VOC inventory.
Option 3. Under this option EPA would require that
gasohol's RVP be no higher than the RVP of in-use
gasolines. Environmentally, this approach is most logical
since emissions are a function of RVP and all gasolines and
blends would meet the same RVP standard. Thus, gasohol use
under this option would add very little to evaporative and
vehicle exhaust emissions and their effect on ozone
production as compared to a fleet using only straight
gasoline.
-------
7-8
However, as discussed earlier, requiring gasohol to
meet the same RVP limits as gasoline would create serious
economic burdens for the gasohol industry, On the other
hand, refiners could take advantage of ethanol's octane
value, as discussed above, which might make gasohol cheaper.
a. Implementing Option 3 nationwide simultaneously
with new gasoline volatility controls would mean that the
impact of distribution-related costs would be felt by the
gasohol industry immediately If measures are available to
the industry which could cushion the effect, there would be
very little time to implement them.
b. Another approach might be to implement Option 3
only in ozone non-attainment areas. The adverse effects of
requiring m-use gasohol RVP to equal gasoline RVP might
occur, but now only in non-attainment areas. Outside
non-attainment areas, the program under this option would
have the characteristics described above for Option 2 (1.0
psi RVP allowance for gasohol), in effect providing a
continued but more limited splash-blending market for
gasohol producers.
This kind of approach would indeed target emission
control primarily in the areas that need it most,
eliminating some of the environmental concerns associated
with Option 2 above. There would remain the concern that
HC emissions would still be higher in attainment areas,
where emission reductions do have some value (see Section I
of Chapter 6.) In any case, it does not appear that such
targeting would sufficiently relieve the economic burden on
the gasohol industry. According to the Ad Hoc Ethanol
Committee[l] the reason is that the widespread nature of
the ozone problem results in over half of the gasohol
market being within current non-attainment areas. It is
unlikely that gasohol market penetration could be increased
in attainment areas enough to make up the difference.
Thus, if prohibiting splash blending gasohol in
non-attainment areas would effectively eliminate that
market, as the industry claims, then the nationwide gasohol
market would be severely reduced, resulting in increased
production costs, higher prices or lower profits margins,
and still greater reduced sales.
Finally, as with any policy which is geographically
targeted, issues of competition are likely to arise among
businesses near the boundaries. On one side of a boundary,
service stations might be able to obtain gasohol to sell,
and on the other side, not.
-------
7-9
c. A third approach would delay Option 3 until
1993, the year in which the Federal tax credit far gasohol
is set to end. The effect of this alternative would be
that the additional emissions as well as the ability to
splash blend associated with Option 2 (1.0 psi RVP
allowance) would occur from the date of implementation of
gasoline RVP control through December 31, 1992 (essentially
until May 1993, since RVP control is not being considered
during non-summer months). After that, gasohol RVP would
need to match gasoline RVP, with the same ramifications we
described for Option 3a above
This option would allow ethanol producers about four
years to change their marketing strategy (i.e., develop
sources of low-RVP, low-octane base gasoline stock). Since
splash-blended gasohol is not expected to be profitable
without the Federal tax credit (unless perhaps crude oil
prices exceed their previous highs), the application of the
RVP limit after the tax credit expires will have little
economic impact. This option would also allow time for
Congress to comprehensively review the issues of RVP, the
gasohol tax credit, and the overall national economic
policy toward gasohol.
V. Methanol-Blend Control Options
Methanol blends have not historically received the
same degree of public attention and special incentives that
gasohol has. One example is the tighter volatility
requirements that are part of all EPA methanol blend
waivers (e.g., ARCO "Oxmol" and the DuPont blend, as
discussed earlier). These waiver requirements have meant
that, from the start, methanol blenders have expected to
operate using special low-RVP base gasolines as a base
stock. A 1.0 psi RVP allowance for methanol blends, while
it would be a substantial benefit to that industry, is
clearly not key to preserving any existing market for these
products; in fact, none are sold at this time. (Methanol
is established as a feedstock in the chemical industry but
not yet as a gasoline blend stock; fuel-grade ethanol, in
contrast, has no established market except gasohol).
However, if EPA were to select Option 2 or Option 3c
for gasohol, which would establish a permanent or temporary
volatility allowance for gasohol, it raises the issue of
whether to also provide a similar RVP allowance for
methanol blends. Increased emissions would certainly
result if methanol blends were marketed at a 1.0 psi higher
RVP than gasoline.
-------
7-10
References (Chapter 7)
1. Letter from B. Direnfeld, Ad Hoc Ethanol
Committee, to C. Gray, EPA, dated June 26, 1986.
2. "The Effect of Gasoline Volatility Control on
Selected Aspects of Ethanol Blending," Jack Faucett
Associates, November 4, 1985.
3. "The Effect of Gasoline Volatility Control on
Methanol and Ethanol Blend Usage," Draft Final Report, Jack
Faucett Associates, September 12, 1986 .
4. "The Impact of Volatility Regulations on the
Fuel-Grade Ethanol Industry," Sobotka & Co., Inc., March,
1987 .
5. Letter from B. Direnfeld, Ad Hoc - Ethanol
Committee, to C. Gray, EPA, dated August 15, 1986.
-------
Appendix 7-A
Effect of Gasohol RVP Allowance on Chicago VOC Inventory
In Chapter 7, the statement is made that if gasohol (90
percent gasoline and 10 percent ethanol) is granted a 1.0 psi
RVP allowance over the controlled volatility level of gasoline,
the result could be an increase of approximately two percent in
the Chicago emissions inventory, relative to the case where
gasohol must meet the same volatility limit as gasoline. This
appendix documents the derivation of that estimated increase.
Chicago was chosen as the area to examine because it is a
densely populated city in violation of the ozone NAAQS and its
gasohol market fraction was the greatest (in September 1985) of
the 61 urban non-California ozone non-attainment areas examined
in the volatility analyses.
The first step in deriving these estimates is the
calculation of motor vehicle non-methane hydrocarbon (NMHC)
emission factors. Separate emission factors (EFs) were modeled
for exhaust, evaporative, and refueling
EPA's m-house version of MOBILE3. The
MOBILE3 model was used to model EFs on
A more detailed discussion of the use
model NMHC EFs on city-specific basis is
(Environmental Impact). The remainder of this discussion will
focus on the EFs modeled for the Chicago area
NMHC emissions, using
m-house version of the
a city-specific basis,
of in-house MOBILE3 to
contained in Chapter 3
The exhaust, evaporative, refueling, and total NMHC EFs
t^O modeled for Chicago are presented in an attachment to this
memorandum. The city specific data inputs used in the
{jrf calculation of these EFs were temperatures (the maximum and
minimum temperatures recorded in Chicago on the date that the
1982-84 ozone design value was recorded, July 15, 1983); in-use
fuel volatility, expressed m terms of Reid Vapor Pressure
(RVP); and the market share of gasohol (gasoline/ethanol
blend). In-use RVP data from July 1985, when the fuel
volatility survey was conducted, and gasohol market share data
from September 1985, which was the most recent available at the
time of this analysis, were used in the modeling of EFs for all
of the projection years.
The temperature, volatility, and gasohol market fraction
data used to model Chicago area EFs are summarized below:
Temperature:
Max 95°F (35°C)
Min 77°F (25°C)
(July 15, 1983)
In-use RVP:
11.9 psi (July 1985)
-------
7-A-2
• Gasohol market
fraction: 29,9% (Sept 1985)
In modeling the NMHC EFs for each city, the effective
volatility (RVP) of the fuel in a given vehicle using gasohol
was assumed to be 0.2 psi greater than that of straight
gasoline, as a result of commingling of gasohol and gasoline in
the vehicle fuel tank. This 0.2 psi offset assumes that
gasoline and gasohol are required to meet the same volatility
(RVP) limit. If it is assumed that gasohol will be permitted a
1.0 psi allowance (exceedance) from the controlled volatility
level of gasoline, which would permit the blending of gasohol
from the same gasoline pool, then the total offset between the
RVPs of gasoline and gasohol would be L.2 psi. The Chicago"
area EFs presented in the attachment thus include two sets of x
EFs for each RVP control level: one labeled "offs 1.2," I
reflecting the case where gasohol has a 1.0 psi allowance above'
the RVP limit for gasoline, and the other labeled "offs 0.2,"
reflecting the case where gasohol is required to meet the same
RVP limit as gasoline.
All of the EFs shown in the attachment are for "all
vehicles," meaning that they incorporate emission rates for all
regulated vehicle classes (gasoline and diesel light-duty
vehicles and trucks and heavy-duty vehicles), appropriately
weighted by their fractions of vehicle miles traveled (VMT).
These EFs do not include emissions from off-highway vehicles
(e.g., construction equipment, boat engines), which are not
regulated in terms of emissions and are not included in MOBILE3.
Since other factors determining total emissions from
mobile sources (such as fleet composition or VMT mix) would not
be affected by a possible RVP allowance for gasohol, comparison
of the Chicago area total NMHC EFs calculated with and without
this allowance provides a good indication of the emissions
increases that could occur if such an allowance were
implemented. By combining this information with the size of
the regulated mobile source inventory as a fraction of the
total NMHC inventory, an estimate of the impact of the
allowance on the total inventory was obtained.
The cases examined were for projection years 1988 and
2010; 1988 is the earliest year that RVP limits could be
imposed, while 2010 is the latest year used in the analyses.
The control RVP levels examined were 10.5 and 11.5 psi for
1988, and 8.0, 9.0, 10.5 and 11.5 psi for 2010.
Inventory projections were run for the Chicago area for
each of seven projection years and eight control RVP levels.
From these projections, the split of the total Chicago area
NMHC inventory between regulated mobile sources and all other
sources was determined. Multiplying the fractional increase m
the emission factor by the fraction of the total inventory
-------
7-A-3
represented by the regulated mobile sources yielded the
estimated inventory increases. This is summarized in Table
7-A-l.
The greatest increase (2.2 percent) is seen in 1988 at a
control RVP level of 11.5 psi. The much lower estimates of
this increase in 2010 are primarily due to the fact that mobile
sources will be a much smaller fraction of the total inventory
as the full effects of existing mobile source emission control
programs are realized.
-------
7 -A-4
Table 7-A-l
Chicago EF and Inventory Data
Total Emission
Factor (q/mi)
I ncr
On-Highway
Mobile Sources
as
Fraction of
Increase in
Total Inventory
RVP
No
1.0 psi
in EF
Total Inventory
A1lowance
Year
< psi)
A1lowance
A1lowance
(M
(M
Co)
1988
10.5
3 .35
3 . 50
4.5
28.8
1.3
11.5
3.85
4.12
7.0
31.9
2 . 2
2010
8.0
1. 15
1. 20
4 . 3
13.4
0.6
9.0
1. 19
1.24
4.2
13.6
0 . 6
10.5
1.27
1.32
3.9
14.3
0.6
11.5
. 33
1.41
6.0
14.9
0.9
A-US GOVERNMENT PRINTING OFFICE 1987-744-622
------- |