Diesel Particulate Study
(Draft)
October 1983
Emission Control Technology Division
Office of Mobile Sources
Office of Air and Radiation
U.S. Environmental Protection Aqency
-------
Diesel Particulate Study
(Draft)
October 1983
Emission Control Technology Division
Office of Mobile Sources
Office of Air and Radiation
U.S. Environmental Protection Aaency
-------
Table of Contents
Page
INTRODUCTION 1
SUPPORTING TECHNICAL ANALYSES
Chapter
1. Technology 1-1
2. Emissions Impacts 2-1
3. Air Quality Impact and Population Exposure 3-1
4. Visibility Assessment 4-1
5. Cancer Risk Assessment . 5-1
6. Non-Cancer Health Effects 6-1
7. Soiling Effects 7-1
8. Economic Impact 8-1
9. Cost Effectiveness 9-1
10. Sensitivity 10-1
-------
INTRODUCTION
I. Purpose of the Study
EPA's study of the costs and benefits of the control of
diesel particulate emissions and their regulation has been
underway for some time. Emission standards for 1982 and later
diesel-powered light-duty vehicles and light-duty trucks
(light-duty diesels) were promulgated in 1980. Similar
standards for diesel-powered heavy-duty diesels were proposed
in 1981.
The pertinent data on both the costs and benefits of
diesel particulate control have been constantly changing over
time. This is particularly true of the last two to three years
since the time of the rulemakings mentioned above. Emission
control technology has been constantly evolving, changing both
baseline emission rates and the ability and costs of further
control. In addition, a number of cancer-related health
studies on diesel particulate have been completed in the last
two years, allowing an assessment of benefits in this area
which was not previously possible. Data and projections in
other key areas have also been changing, resulting in a need
for EPA to reexamine its regulatory position.
Recent regulatory activity by EPA has reflected these
changing circumstances. In the light-duty area, EPA has
proposed and will soon promulgate a delay of the more
stringent, 1985 standards until 1987, leaving in place the
current 1982 standards through 1986. This action is based on
the fact that a new control technology could not be applied
fleet-wide for the 1985 model year, but will require two
additional years of effort. This new technology is referred to
as a trap-oxidizer and produces substantial reductions (i.e.,
greater than 50 percent) in diesel particulate emissions. In
the heavy-duty area, EPA has announced its intention to
repropose its particulate standard along with the NOx standard
proposal, because of the interelationship between NOx and
particulate control. This combined proposal should enable this
interaction between the two pollutants to be better assessed
and facilitate a more orderly standard setting process.
The purpose of this study is to provide a comprehensive
assessment of the costs and benefits of the control of diesel
particulate emissions and to recommend a regulatory strategy
for their control. As such, this study expands, updates and
combines the Regulatory Analyses supporting the light-duty
diesel (LDD) particulate final rule and the heavy-duty diesel
-------
(HDD) particulate proposed rule.[1,2] The study will identify
current and future diesel particulate emissions and exposure
levels, assess the health and welfare impact of diesel
particulate, and estimate the costs of controlling diesel
particulate emissions to various levels. The study will then
integrate these aspects of diesel particulate control and
develop, evaluate and recommend a regulatory control strategy.
Four regulatory scenarios are examined, covering a wide
range of technological stringency. The least stringent control
scenario (the relaxed scenario) would require no further
control from LDDs and only very modest reductions from HDDs,
representing the least stringent degree of control
conceivable. The next most stringent scenario (the
intermediate scenario) would require the application of
advanced non-trap technology. By their nature, these
techniques are quite cost effective and this scenario
represents a modest degree of control that should be available
at low cost. The third scenario (the base scenario) consists
of the current trap-based standards (i.e., the 1985 LDD
particulate standards and that proposed for 1986 HDDs). This
scenario provides more control than that achievable through
non-trap technology, but will be more costly and less cost
effective than the second scenario due to the use of trap
technology. The fourth scenario (the stringent scenario)
represents the greatest degree of control presently
conceivable. Nearly all vehicles would be equipped with traps
under this scenario. These scenarios are summarized in Table i.
As an averaging concept has already been promulgated for
compliance with the 1985 LDDV and LDDT trap-based particulate
standards, the flexibility it provides will be presumed here
for the base and stringent scenarios. As it is likely, but not
certain, that a similar program will be proposed for HDDVs,
both averaging and non-averaging situations will be examined
for those scenarios requiring control (i.e., base and
stringent).
A factor that must be considered when assessing the
ability to control diesel particulate is the level of the
applicable NOx standard. In general, as NOx emissions are
reduced, engine-out levels of particulate increase. Thus,
under a stringent NOx standard the technically feasible level
of particulate control (without the use of aftertreatment
devices) will be higher than under a lenient NOx standard.
Because there is currently some doubt as to what the NOx
standards will be in the years covered by this study, three
different LDDV and LDDT standards were evaluated: 1) 1.0, 2)
1.5, and 3) 2.0 g/mi. As the NOx standard for LDDTs is
-------
Table 1
Emission Control Scenarios
Relaxed Scenario:
LDDV (g/mi)
LDDT (g/mi)
HDDV (g/BHP-hr)
Intermediate Scenario:
LDDV (g/mi)
LDDT (g/mi)
HDDV (g/BHP-hr)
Base Scenario:
LDDV (g/mi)
LDDT (g/mi)
HDDV (g/BHP-hr)
Stringent Scenario:
LDDV (g/mi)
LDDT (g/mi)
HDDV (g/BHP-hr)
Particulate*
Current Levels(NA)
Current Levels(NA)
0.6(NA)
0.25-0.30(A)
0.30-0.35(A)
0.40(A)
NOx
0.20(A)
0.26(A)
0.25(NA &
0.08(A)
0.105(A)
0.10(A)
A)
1.0, 1.5, 2.0
1.2, 1.7, 2.3
1.0, 1.5, 2.0
1.2, 1.7, 2.3
1.0, 1.5, 2.0
1.2, 1.7, 2.3
1.0, 1.5, 2.0
1.2, 1.7, 2.3
(A) means averaging program available, (NA)
. averaging program available.
means no
-------
directly influenced by that for LDDVs, this study will also
evaluate the effect of the three LDDT NOx standards equivalent
to those for LDDVs: 1.2, 1.7, and 2.3 g/mi.
The question of the appropriate NOx standard for HDDs is
dealt with in a more straightforward fashion than LDDs, since
the level of the standard will be set by EPA. While Section
202(a) (3) (A) (ii) of the CAA requires a NOx standard of 1.7
g/BHP-hr, this level is not feasible for HDDs. Thus, EPA must
set a revised NOx standard under the requirements of Section
202 (a) (3) (B), which are very similar to the requirements
specified for the HDD particulate standard in Section
202(a) (3) (A) (iii) . Thus, under all scenarios, the HDDV NOx
standard is treated as a variable and identified in much the
same way as the particulate standard.
II. Organization of the Study
The study has been segregated into two parts. This
introduction and an evaluation of control options constitute
the main body of the study, while the bulk of the technical
analysis follows under the heading, "Supporting Technical
Analyses".
In addition to describing the context, purpose and
organization of the study, this introduction describes the
control scenarios evaluated and the diesel sales projections
used throughout the analysis. The following "Evaluation of
Control Options" summarizes the costs and environmental
benefits of the various diesel particulate control scenarios
and then goes on to compare and evaluate their relative
strengths and weaknesses.
The supporting technical analysis is contained in ten
chapters. The first seven chapters address the benefits of
control, including vehicular emissions (Chapter 1), nationwide
and urban emissions (Chapter 2), air quality and exposure
(Chapter 3) , carcinogenic risk (Chapter 4) , visibility (Chapter
5), non-carcinogenic health risk (Chapter 6) and soiling
(Chapter 7) . Two chapters address the cost (Chapter 8) and
cost effectiveness (Chapter 9) of control, respectively. The
last chapter (Chapter 10) addresses the sensitivity of the
technical results to key policy assumptions made throughout the
study.
The primary technical analyses (i.e., that contained in
Chapters 1 through 9) will evaluate only the relaxed and base
control scenarios because these two scenarios are considered
the most likely to occur and all technical concepts associated
with the stringent scenario, such as control technology, are
-------
also contained in the base scenario. The NOx standards of the
main analysis will be 1.5 and 2.3 g/mi for LDDVs and LDDTs,
respectively, because they are the current standards and
certification test data is available under these levels. As
the methodology used in Chapter 1 to adjust particulate
emissions for different NOx emissions standards is subject to
some error, this will minimize the use of such adjustments in
the main analysis. The impact of other LDDV and LDDT NOx
standards on the relaxed and base particulate control
scenarios, as well as that of the stringent control scenario
under all NOx standards, will be evaluated in Chapter 10,
Sensitivity. As little firm data is available concerning the
technology associated with the intermediate sceanario, it will
be addressed primarily in the options analysis in the main body
of the study and somewhat in the emissions analysis of Chapter
2 of the supporting analysis.
-------
t
CHAPTER 1
TECHNOLOGY
I. Introduction
The major reductions in diesel particulate emissions
available from engine modifications have already been achieved,
with the possible exception of electronic control of the fuel
injection system. Further major reductions will need to be
accomplished through the use of trap-oxidizer systems.
Under the current light-duty diesel vehicle (LDDV) and
light-duty diesel truck (LDDT) particulate standards of 0.6
gram per mile (g/mi), no traps are necessary. Since heavy-duty
dieseis (HDDs) are not currently subject to a particulate
standard, traps are not found on current HDDs either. However,
the more stringent particulate standards of the base scenario
will require traps on many dieseis. (This chapter investigates
each manufacturer's need for trap-oxidizer systems under the
LDDV, LDDT, and HDD particulate standards of the base scenario,
as well as determining the non-trap particulate emission levels
which would occur under less stringent particulate standards of
the relaxed (non-trap) scenario.
This chapter is divided into three sections, each in turn
addressing LDDVs, LDDTs and HDDs. The section addressing LDDVs
is the most detailed, as the methodology for all three sections
is therein described. The latter sections only reference this
methodology.
The LDDV section itself consists of five parts. The first
simply describes the source of the engine-out LDDV particulate
levels used in the analysis. The second addresses the
NOx/particulate trade-off issue and establishes NOx/particulate
relationships to be used in adjusting particulate emission
levels to varying NOx levels. While these relationships will
have only a limited use here in addressing the base
scenario--most LDDVs are at NOx levels near those appropriate
to comply with the base scenario's 1.5 g/mi NOx standard — they
will be of significant use in addressing the sensitivity of the
results of this chapter to varying LDDV and LDDT NOx standards
(see Chapter lff>) . The third part of the LDDV section estimates
the equivalent "standard" levels that each LDDV engine
configuration could meet without traps and the fourth part
converts these engine configuration levels into corporate
average non-trap standards achievable by each manufacturer (the
relaxed scenario). The fifth and final part will then
determine the percentage of LDDVs which will require traps
under the base scenario (0.2 g/mi particulate and 1.5 g/mi NOx
standards with corporate averaging).
-------
1-2
II. Light-Duty Diesel Vehicles
A. Current Levels of NOx and Particulate Emissions
The most convenient and accurate source of current LDDV
engine-out particulate levels is the new-vehicle certification
program. The first model year in which LDDV manufacturers had
to certify to the current 0.6 gram per mile (g/mi) particulate
standard was 1982. However, some manufacturers chose to test
for particulates in the 1981 model year and then carryover the
results for the 1982 model year thereby spreading out the new
emissions testing program over two years. Thus, certification
test results for LDDV particulate are primarily available for
the last two years (ie., the 1982 and 1983 model years), with
some data being available from the 1981 model year as well.
All of the 1983 model year LDDV engine families were
subdivided into configurations on the basis of transmission
type and inertia weight class. The available 1981-83 NOx and
particulate test data were then obtained for each of these
configurations. These data included emission tests plus fuel
economy tests during which emissions were also measured. Both
manufacturer tests and EPA tests were included.
A review of the test results of configurations for which
testing had been done for both the 1982 and 1983 model years
did not show a clear pattern of change from one year to the
next, although there was a modest trend for both NOx and
particulate to improve with the more recent data. Therefore,
it was concluded that only the most recent (1983) test results
should be used here when available. However, in the cases
where 1983 engine configurations were carried over from 1982
and no 1983 data were available, the 1982 model year
certification test results were used.
These most recent test results for each configuration were
then examined and outliers excluded before determining the mean
for each configuration. In general, outliers were test results
greater than 140 oercent or less than 60 percent of the mean of
the rest of the test results for that configuration. This
range may have been somewhat greater or smaller depending on
the observed spread and total number of tests. The remaining
test results for each configuration were averaged and the
resultant means used as the current level of NOx and
particulate emissions for each configuration. These
engine-configuration means are shown in Table 1-1.
B. The NOx/Particulate Tradeoff
Having established the current NOx and particulate
emission levels, an estimate of how the particulate emission
level would change if the NOx emission level were increased or
decreased was made. Such an analysis was primarily n'ecessary
-------
Table 1-1
Actual, Certification LDDV
Particulate and NOx Emission Levels
Manufacturer
General Motors
Volkswagen
Engine
Family
Z90
Z90
ZK7
ZK7
ZT8
ZT7
ZT7
ZT7
ZT7
AAO
AAO
AAO
JAO
JAO
AZ8
AZ8
AZ8
RA5
BZX
BZX
Trans .
M5
L3
L3
L3
L3
L3
L3
L4
L4
M4
M5
A3
M5
A3
M5
M5
A3
S4
A3
M5
Inertia Weight
Class (Ib)
2,500
2,500
3,000
3,500
3,500
4,000
4,500
4,000
4,500
2,250
2,250
2,250
2,500
2,500
2,250
2,500
2,500
2,250
2,750
2,750
Displacement
(liters)
1.8
1.8
4.3
4.3
4.3
5.7
5.7
5.7
5.7
1.6
1.6
1.6
1.6
1.6
1.6
1.6
1.6
1.6
1.6
1.6
Particulate
LMT (g/mi)
NOx
LMT (g/mi)
.17
.13
.22
.25
.21
.32
.37
.37
.40
.16
.19
.18
.19
.17
.22
.20
.29
.18
.18
.22
1.11
1.01
1.04
1.10
1.23
1.21
1.14
1.11
1.18
.90
1.02
1.01
1.02
1.10
1.12
1.10
1.14
1.02
1.22
.1.19
-------
Table 1-1 (cont'd)
Actual, Certification LDDV
Particulate and NOx Emission Levels
Manufacturer
Nissan
Mercedes-Benz
Isuzu
Audi
Peugeot
Volvo
Engine
Family
AF8
AF8
AF8
AF8
AF8
AFO
AFO
501
501
508
CD7
CD?
CD7
CD7
BZ7
BZ7
CZ3
. BAX
AAl
AA1
BA3
BA3
AY2
AY2
TBO
TBO
Trans.
M4
M5
M5
A3
A3
M5
L4
M4
A4
A4
M4
M5
M5
A3
M5
A3
A3
M5
M5
A3
M4
A3
M4
A3
M5
A3
Inertia Weight
Class (Ib)
2,250
2,250
2,500
2,250
2,500
3,500
3,500
3,500
3,500
4,000
2,500
2,500
2,750
2,750
2,750
2,750
3,000
2,500
3,500
3,500
3,500
3,500
3,500
3,500
3,500
3,500
Displacement
(liters)
1.7
1.7
1.7
1.7
1.7
2.8
2.8
2.4
2.4
3.0
1.8
1.8
1.8
1.8
1.6
1.6
2.0
1.6
2.3
2.3
2.3
2.3
2.4
2.4
2.4
2.4
Particulate
LMT (g/mi)
NOx
LMT (g/mi)
.17
.20
.23
.24
.23
.22
.24
.42
.38
.43
.19
.17
.18
.16
.22
.17
.19
.21
.28
.30
.32
.40
.29
.27
.29
.23
.82
.94
1.00
.89
.92
1.16
1.32
1.11
1.15
1.26
1.09
1.21
1.17
1.29
1.19
1.21
1.23
1.08
1.04
1.01
.87
.98
1.37
1.31
1.17
1.19
-------
1-5
so that the particulate emission level of each configuration
under the various NOx standards being considered in the
sensitivity analysis could be estimated. However, it is also
useful here, since many engine configurations are emitting NOx
well below the levels required by a 1.5 g/mi standard and some
adjustment of their particulate levels would appear appropriate.
Assuming that only injection timing retard or EGR is used,
the general shape of a NOx/particulate tradeoff curve is known
to be (NOx emissions in the x dimension and particulate
emissions in the y dimension) : 1) negative in slope at all
points, 2) steeply sloped at low NOx levels, and 3) gently
sloped to flat at high NOx levels. Furthermore, it is
generally known that the curve shifts outward (jfTeTp upwards and
to the right) with increasing engine displacement'. Figure 1-1
shows generalized NOx/particulate tradeoff curves and
illustrates this shifting effect of engine displacement.
Ideally, the specific tradeoff curve would be known for each
engine family/configuration. However, such curves are not
available. Therefore, an approximate method was developed for
predicting particulate emission levels from known NOx levels.
First, in order to account for the shifting of the curve
that occurs with changes in engine displacement, the 1983 model
year engine families were divided into the following three
groups: small engines (1.6 to 1.8 liters), medium engines (2.0
to 2.8 liters) and large engines (3.0 to 5.7 liters). The NOx
and particulate emission levels were then plotted for each
configuration within each engine size group.
The NOx emission levels for the small engine group ranged
from 0.80 to 1.29 g/mi. The distribution of points appeared to
have a slightly negative slope which regression of the data
confirmed. The emission levels of one configuration (VW,
engine family AZ8, A3 transmission, 2500 Ibs.) were excluded
from the regression because the NOx/particulate combination was
well outside the range of all of the other values including
other values for that same engine family. The slope of the
regression line was -0.033. This slope is quite small, .as will
be seen later when compared to those for the larger engines,
and is generally in line with what would be expected for small
engines. [1] Therefore, it was ~ used to predict changes in
particulate emission levels resulting from changes in NOx
emission levels below an absolute NOx level of 1.35 g/mi. It
was assumed that no further reduction in particulate would
occur for NOx emission levels greater than 1.35 g/mi (i.e., the
slope was considered to be zero) . [1]
A NOx emission level of 1.35 g/mi was chosen as the
reference point to change slopes for the small engine group
(and the other two groups) for two reasons. First, the great
majority of current LDDVs have NOx emissions less than 1.35
g/mi. Since the slopes obtained by the regression of the data
-------
1-6
Figure 1-1
Generalized Shape of NOx/Particulate Tradeoff
Curves Illustrating the Shifting Effect of Engine Displacment
O
•ft
4J
i-l
(0
Large Engines
Small Engines
NOx
-------
1-7
are most appropriate within the distribution of data, it was
decided to limit the applicability of the regressions to this
level. Second, 1.35 g/mi is approximately the engineering
objective (or low mileage target (LMT)) for the NOx standard of
the base scenario (1.5 g/mi). (The NOx standard of 1.5 g/mi
minus a 10 percent safety margin and divided by a deterioration
factor (DF) of 1.000 (which is typical for diesel NOx
emissions) yields a LMT of 1.35 g/mi.)
The plot of the emission levels for the medium engine
group, whose NOx values ranged from 0.87 to 1.37 g/mi, appeared
to have a greater negative slope than the small engine group.
Regression of the data confirmed this, showing a slope of
-0.201. As this slope appeared reasonable for engines of this
size, based on the limited information available on current
NOx/particulate tradeoff curves[1] and the fact it was larger
than the slope for the small engines, this slope was used to
predict the change in particulate emission levels for changes
in NOx emission levels below 1.35 g/mi NOx. Again, two
configurations' emission levels (M-B, 2.4L, both transmissions)
were not used in the regression because they definitely
appeared to be outliers. The slope of the tradeoff curve for
NOx emission levels greater than 1.35 g/mi was somewhat
arbitrarily reduced by one-half to -0.100, since it is known
that the curve becomes flatter at higher NOx levels, but by an
unknown magnitude.
For the large engines, the slope of the tradeoff curve for
NOx values below 1.35 g/mi was not based on a regression of the
data, but was simply estimated to be -0.400 based on known
tradeoff curves for large, albeit older, engines.[1] This was
necessary because there were only 8 data points for the large
engines and no correlation existed. The slope for NOx values
greater than 1.35 g/mi was also estimated using engineering
judgment and was set at -0.100. At first a slope of -0.200 was
estimated, based on the judgment that this slope should be
steeper than that for the medium-size engines. However, this
produced some unrealistically low particulate values at the
higher NOx values being examined in the NOx sensitivity
analysis, so -0.100 was chosen instead.
C. Engine Configuration's Low-Mileage Targets and
Standard Levels
Using the slopes of the tradeoff curves determined above
and the data in Table 1-1, the particulate low-mileage target
(LMT) at 1.35 g/mi NOx was calculated for each configuration.
The particulate standard level for each configuration was then
calculated for these particulate LMTs. This was done by
multiplying the particulate LMT by the appropriate 50,000 mile
deterioration factor (DF) and the appropriate safety margin.
Both factors are explained below.
-------
1-8
The particulate DF used for each configuration was the
certification DF for the 1983 model year except in three
instances. The three exceptions were engine families* with DFs
much greater than the other 18 engine families. Fifteen of the
21 total engine families had particulate DFs less then 1.10.
Another three engine families had particulate DFs between 1.10
and 1.15. The last three engine families had particulate DFs
greater than 1.24. It was concluded that the manufacturers of
these last three engine families could lower the DFs to at
least the 1.15 level if a more stringent particulate standard
required them to do so. Therefore, for the purposes of this
study, a DF of 1.15 was assumed for each of those three engine
families.
The safety margins necessary for calculating the
particulate standard levels from each particulate LMT were
determined using the methodology developed for past EPA
rulemakings.[2] That methodology requires a coefficient of
variation (COV) for production-line vehicles and the number of
prototype vehicles tested before a manufacturer fixes its
design. Results from EPA's Selective Enforcement Audit (SEA)
testing program[3] indicate that the LDDV particulate COV is
slightly less than 0.13. Also, the number of prototype
vehicles to be built and tested was presumed to equal the
maximum considered in the methodology (seven), since the engine
technology exists today and manufacturers will have more than
sufficient data upon which to base their LMTs. Thus, the
safety margin as interpolated from the table[2] would be seven
percent. However, since available SEA test data on LDDVs is
limited and the particulate COV may increase somewhat with more
stringent NOx and/or particulate standards, a somewhat larger
safety margin of 10 percent was used for this study.
The particulate standards achievable by each configuration
are shown in Table 1-2. An industry-wide, non-averaging,
non-trap, non-technology forcing particulate standard can be
determined by simply identifying the configuration with the
highest particulate standard listed in Table 1-2. Thus, for
the NOx standard of 1.5 g/mi, such a particulate standard would
be 0.43 g/mi (M-B, 3.0L engine).
It should be noted that this highest emitting
configuration, as well as the next three highest emitting
configurations, seem to be technology outliers. Three out of
four of these configurations are Mercedes-Benz (M-B) vehicles.
When the emissions of these M-B vehicles are compared to those
of other similarly sized vehicles, one finds that the M-B
Engine families, rather than configurations, are
considered here because DFs are only determined on an
engine family basis and are applied to all configurations
within that engine family.
-------
Table 1-2
Achievable Non-Trap, LDDV Particulate
Standards Under the 1.5 g/mi NOx Standard
Part. Std.
Assuming a
Engine Inertia Weight Displacement 1.5 g/mi
Manufacturer Family Transmission Class (Ib) (liters) NOx Std. (g/mi)
General Motors Z90 M5 2,500 1.8 .20
Z90 L3 2,500 1.8 .15
ZK7 L3 3,000 4.3 .12
ZK7 L3 3,500 4.3 .17
ZT8 L3 3,500 4.3 .19
ZT7 L3 4,000 5.7 .33
ZT7 L3 4,500 5.7 .36
ZT7 L4 4,000 5.7 .35
ZT7 L4 4,500 5.7 .42
Volkswagen AAO M4 2,250 1.6 .17
AAO M5 2,250 1.6 .21
AAO A3 2,250 1.6 .20
JAO M5 2,500 1.6 .21
JAO A3 2,500 1.6 .19
AZ8 M5 2,250 1.6 .27
AZ8 M5 2,500 1.6 .24
AZ8 A3 2,500 1.6 .26
RA5 S4 2,250 1.6 .20
BZX A3 2,750 1.6 .20
BZX M5 2,750 1.6 .24
-------
Table 1-2 (cont'd)
Achievable Non-Trap, LDDV Particulate
Standards Under the 1.5 g/mi NOx Standard
Part. Std.
Assuming a
1.5 g/mi
Manufacturer Family Transmission Class (Ib) (liters) NOx Std. (g/mi)
Nissan AF8 M4 2,250 1.7 .20
.24
.28
.29
.27
.22
.28
Mercedes-Benz 501 M4 3,500 2.4 .41
.37
.43
Isuzu CD7 M4 2,500 1.8 .22
.20
.21
.19
Audi BZ7 M5 2,750 1.6 .24
.19
.19
.25
Peugeot AAl M5 3,500 2.3 .24
.26
.25
.36
Volvo AY2 M4 3,500 2.4 .36
.32
.32
.25
Engine
Family
AF8
AF8
AF8
AF8
AF8
AFO
AFO
501
501
508
CD?
CD7
CD7
CD7
BZ7
BZ7
CZ3
BAX
AAl
AA1
BA3
BA3
AY2
AY2
TBO
TBO
Transmission
M4
M5
M5
A3
A3
M5
L4
M4
A4
A4
M4
M5
M5
A3
M5
A3
A3
M5
M5
A3
M4
A3
M4
A3
M5
A3
Inertia Weight
Class (Ib)
2,250
2,250
2,500
2,250
2,500
3,500
3,500
3,500
3,500
4,000
2,500
2,500
2,750
2,750
2,750
2,750
3,000
2,500
3,500
3,500
3,500
3,500
3,500
3,500
3,500
3,500
Displacement
(liters)
1.7
1.7
1.7
1.7
1.7
2.8
2.8
2.4
2.4
3.0
1.8
1.8
1.8
1.8
1.6
1.6
2.0
1.6
2.3
2.3
2.3
2.3
2.4
2.4
2.4
2.4
-------
1-11
vehicles emit significantly more particulate. Thus, it appears
that M-B has not yet implemented the kinds of combustion
chamber and injection modifications that have been made by
others (e.g., General Motors). Presumedly, M-B could do this
if it became necessary. The fourth configuration is a General
Motors vehicle powered by their 5.7 L engine (L4 transmission,
4500 Ibs.). It has been rumored that this engine will be
eliminated sometime in the next few years, primarily due to
market considerations, as well as to the fact that this is
their highest emitting engine for both NOx and particulate. If
these four configurations were excluded from consideration
here, the non-averaging, non-technology forcing particulate
standard could be 0.36 g/mi, 15 percent lower than the 0.43
g/mi level mentioned above.
D. Non-Trap "Averaging" Standards
The previous discussion presented the methodology used to
estimate particulate LMTs and standard levels for each
configuration under a NOx standard of 1.5 g/mi. From those
results an industry-wide, non-averaging, non-trap,
non-technology forcing particulate standard could be selected.
That standard was based on the assumption that every LDDV would
need to be at or below the standard (i.e., non-averaging). In
this situation, most vehicles could increase their particulate
emissions up to the level of the worst-case vehicle and still
be in compliance. While we would not expect such a situation
to occur to this extreme, it is possible that NOx control and
fuel economy incentives could lead to increased particulate
emissions if the particulate standard allowed it.
One way to significantly increase the probability that
industry-wide particulate emissions would not increase beyond
present levels and yet still set a non-technology forcing
standard would be to implement a corporate average standard
(see the introduction to the study for an explanation of
emissions averaging). Under this approach the non-trap,
non-technology forcing standard would be numerically lower than
that determined in the previous discussion since each
manufacturer could "average" its high emitters with its low
emitters. Because of this, it becomes more difficult for a
manufacturer to increase the emissions of its low emitters
since these emissions are factored into its corporate average
emission level and are no longer irrelevent. Thus, while
averaging has been considered in the past only for trap-based
particulate standards, it also has a benefit for non-trap
standards.
Table 1-3 shows the currently achievable non-trap,
non-technology forcing particulate standard for each
manufacturer under averaging. These standards were calculated
by sales weighting the achievable particulate standards for
each manufacturer's LDDV configurations listed in Table 1-2.
Sales for each configuration were obtained from the
-------
Table 1-3
Achievable Non-Trap Particulate
Standards (under "averaging")
Assuming
a 1.5 g/mi
NOx Standard
Manufacturer (g/mi)
General Motors .29*
Volkswagen .20
Nissan .26
Mercedes-Benz .42
Isuzu .20
Audi .20
Peugeot .26
Volvo .29
This level becomes 0.16 g/mi if GM's 5.7-liter engine is
discontinnued and its sales are replaced by their
4.3-liter engine.
-------
1-13
manufacturers' 1983 estimated Federal sales required by the
fuel economy program known as Corporate Average Fuel Economy
(CAFE).* If the worst-case manufacturer's (i.e.,
Mercedes-Benz) particulate averaging standard level became the
averaging standard for the industry, then as Table 1-3 shows,
the particulate averaging standard would be 0.42 g/mi. This
level is not significantly lower than the 0.43 g/mi
non-averaging standard of the previous section due to what
appears to be excessively high emission levels of the
worst-case manufacturer's engines. If this worst-case
manufacturer is treated as a technology outlier, then the
corporate average for General Motors and Volvo would set the
industry-wide, particulate averaging standard at 0.29 g/mi.
This level is 33 percent lower than the non-averaging standard
of the previous section (0.43 g/mi). A non-trap,
non-technology forcing, particulate averaging standard of 0.29
g/mi would moderate the risk that manufacturers of small LDDVs,
which are low particulate emitters, might substantially
increase particulate emissions from these vehicles.
It is interesting to note what would happen to GM's
corporate average particulate level if it discontinued
production of its 5.7-liter engine. This could happen if the
long-range trend towards increased fuel economy eliminated the
"big" cars of today whereupon the need for the 5.7-liter engine
would also be eliminated. Assuming that the vehicles which
would have had the 5.7-liter engine received instead GM's
4.3-liter engine, GM's average particulate standard level would
drop from 0.29 g/mi as shown in Table 1-3 to 0.16 g/mi.
Furthermore, since GM's estimated sales comprise about 60
percent of the total LDDV estimated sales, lowering GM's
average particulate standard level by this 45 percent would
lower total LDDV particulate emissions substantially. However,
the non-trap, non-technology forcing, averaging particulate
standard based on the second highest emitter^ would remain at
0.29 g/mi ((VolvoT
E. Determination of the Percent of Trap-Equipped
Vehicles
Thus far, this analysis has been concerned with only those
particulate standards achievable without the use of
trap-oxidizer systems. It will now consider the use of traps
as a particulate control strategy. Here the focus of the
analysis will differ from that of the previous section.
Instead of determining achievable particulate standards under
*Theseprojections are confidential and are not presented
here. The presentation of the resultant corporate
emission average does not divulge the pertinent
information contained in the projections (i.e., absolute
sales) .
-------
1-14
various scenarios which assume some percent usage of traps (in
the previous case, zero), this discussion will assume a
particulate standard of 0.20 g/mi (the base scenario) and then
determine the percentage of the LDDV fleet requiring traps in
order to achieve this standard. Emissions averaging will be
assumed to apply as the Agency expects to soon finalize a
particulate averaging program in conjunction with the 0.2 g/mi
standard (proposed in 46 FR 62608).
Two types of traps were considered for compliance with the
0.20 g/mi particulate standard. One is the wire mesh type and
the other is the ceramic type. EPA's report[4] on the
feasibility of trap oxidizers indicated that both types appear
to have good durability characteristics. The ceramic type of
trap was tested by Southwest Research Institute for EPA[5] and
the wire mesh type was tested at this same facility for
Johnson-Matthey, Inc. [6] These testing programs indicated that
the deterioration for both types of traps was negligible and,
therefore, a DF of 1.00 was used here. The EPA/[4] jFeporlT)
discusses each of the two traps in detail and concTudes that
the efficiency of the ceramic trap is about 70-90 percent while
the efficiency of the wire mesh trap is about 50-80 percent.
As the durability test of the ceramic trap referred to above
showed an 85 percent efficiency, that figure will be used
here. For the wire mesh trap, 65 percent will be used as a
reasonable mean efficiency for a typical trap. This analysis
will use these percent efficiencies to determine the tail pipe
emission levels from engine-out emission levels. (For
simplicity, mixing use of both trap types was avoided.)
The methodology used to calculate the percentage of
trap-equipped vehicles for each type of trap and manufacturer
is straightforward. First, each configuration's estimated
sales was multiplied by that configuration's non-trap standard
level taken from -Table 1-2. These results were then summed to
obtain the total number of vehicle-grams per mile (veh-g/mi)
from which each manufacturer would begin its control efforts.
Next, each manufacturer's total estimated sales were multiplied
by 0.20 g/mi to give the total veh-g/mi that the manufacturer
would be allowed under each particulate averaging standard.
The difference between the two figures is the amount of control
each manufacturer needs to achieve. It was assumed that a
manufacturer would put traps on its highest emitters first
because the g/mi reduction achieved is highest for those
vehicles.
To determine the number of veh-g/mi saved by putting a
trap on a given configuration, that configuration's particulate
LMT was first multiplied by one minus the trap efficiency and
then multiplied by that configuration's particulate DF. This
result was then transformed into a new particulate standard
level by adding a safety margin of 10 percent or 0.02 g/mi,
whichever was greatest, because 0.02 g/mi was considered the
-------
1-15
minimum acceptable safety margin. The new particulate standard
level was then multiplied by the estimated sales for that
configuration to give the new total veh-g/mi emitted. The
difference between the total veh-g/mi without traps and the
total veh-g/mi with traps was counted as controlled veh-g/mi.
Calculations were made for each configuration until the
controlled veh-g/mi equalled or exceeded the amount of veh-g/mi
that the manufacturer needed to control in order to meet the
0.2 g/mi particulate standard. For the most part, only enough
traps were assumed installed to just meet the standard.
However, in a few instances where the percentage of traps on a
given configuration approached 80 percent it was assumed that
the whole configuration would be equipped with traps.
The results of these calculations are shown in Table 1-4.
Assuming ceramic traps, 22 percent of all LDDVs would require
traps to comply with the 0.20 g/mi standard. Assuming wire
mesh traps, this figure increases to 30 percent.
III. Light-Duty Diesel Trucks
The methodology used to estimate the non-trap,
non-technology forcing, particulate standards for each LDDT
configuration was the same as that used for LDDVs. The current
particulate emission test levels for small LDDTs (i.e., engine
displacements from 1.6 to 2.3 liters), which were obtained from
certification test results, were used to calculate the
particulate standard levels shown in Table 1-5. The NOx
emission levels of the majority of these configurations were
between 1.35 and 1.7 g/mi. Since the NOx/particulate tradeoff
curve for small LDDV engines was flat in this region, no
adjustment was made to the small LDDT certification values in
deriving the particulate LMTs. For the full-size LDDTs (i.e.,
engine displacements of 6.2 liters) the current particulate
emission test levels were adjusted to their equivalent at 2.05
g/mi NOx using the same NOx/particulate tradeoff curve (slope
of -0.100) as that used for large LDDVs. (A LDDT NOx level of
2.05 g/mi under a 2.3 g/mi NOx standard is equivalent to the
1.35 g/mi NOx level for LDDVs. Also, the -0.1 slope curve was
the only one needed since the certification NOx emission levels
of the full-size LDDTs were all above 1.5 g/mi.) As shown in
Table 1-5, the industry-wide, non-trap, non-technology forcing
particulate standard without averaging would be 0.40 g/mi.
Table 1-6 presents each LDDT manufacturer's non-trap,
non-technology forcing, particulate standard under the
averaging concept. These levels were calculated using the same
methodology as was previously described for LDDVs. The highest
average particulate level is for Mitsubishi at 0.39 g/mi. Note
that this level is well above that for GM (0.28 g/mi), which
only produces full-size LDDTs. Thus, if Mitsubishi were
considered controlling, there is very little difference between
the non-averaging and the averaging non-trap standard for LDDTs.
-------
Table 1-4
Percentage of LDDV Sales Requiring Traps Under Various
Particulate Standards (assumes "averaging")
1.5 g/mi NOx Standard
Manufacturer
0.20 g/mi
Particulate
Standard with
Ceramic Trap
0.20 g/mi
Particulate
Standard with
Wire Mesh Trap
General Motors
Volkswagen
Nissan
Mercedes-Benz
Isuzu
Audi
Peugeot
Volvo
Industry-wide
Sales-Weighted
Percentage
26.8
0
25.6
55.5
0
2.2
30.3
34.4
22.3
36.2
0
33.4
79.6
0
2.9
39,
44,
5
2
30.2
-------
Table 1-5
Achievable Non-Trap, LDDT Particulate
Standards Under the 2.3 g/mi NOx Standard
Manufacturer
Small LDDTs:
Ford
Isuzu
Nissan
Mitsubishi
Toyota
Volkswagen
Toyo Kogyo
Engine
Family Transmission Class (Ibs.)
Inertia Weight Displacement
AG5
CD3
AF9
FDD
BBS
FF9
PA2
VA9
KK9
M4
M4
M4
M5
M5
M5
M5
M5
M5
M4
M5
M5
M5
M5
3,000
2,750
3,000
3,000
3,000
3,000
3,500
3,000
3,000
2,250
2,250
3,500
4,000
3,000
(liters)
2.2
2.2
2.2
2.2
2.2
2.3
2.3
2.2
2.2
1.6
1.6
1.6
1.6
2.2
Part. Std.
Assuming a
2.3 g/mi
NOx Std. (g/mi)
.29
.28
.26
.25
.35
.39
.38
.17
.25
.26
.38
.27
.33
.29
Full-Size LDDTs:
General Motors Z40
M4
L4
M4
L4
M4
L4
L4
4,500
4,500
5,000
5,000
5,500
5,500
6,000
6.2
6.2
6.2
6.2
6.2
6.2
6.2
.32
.35
.26
.28
.40
.26
.36
-------
Table 1-6
Achievable Non-Trap Particulate
Standards Under "Averaging1
Manufacturer
Small LDDTs:
Ford
Isuzu
Nissan
Mitsubishi
Toyota
Volkswagen
Toyo Kogyo
Assuming a 2.3
g/mi NOx Standard
.29
.25
.35
.39
.19
.31
.29
Full-size LDDTs:
General Motors .28
-------
1-19
As in the LDDV case, the percentage of LDDTs requiring
trap-oxidizer systems under the base scenario (0.26 g/mi
particulate standard) was determined. The methodology used to
determine this percentage was the same as for LDDVs except that
small and full-size LDDTs were considered separately. This was
done because the ratio of sales of small to full-size LDDT
sales is expected to change significantly by the mid-to-late
1980s. A study[7] by Jack Faucett Associates (JFA) projects
that in 1987, 86.5 percent of all new LDDT sales will be
full-size while only 13.5 percent will be small.
Manufacturers' LDDT sales estimates for the 1983 model year
indicate that currently full-size LDDTs represent about 55
percent of all LDDT sales. Thus, a substantial change is
expected to occur over the next several years. Therefore, the
percent of traps required by each LDDT-size group was weighted
according to the findings by JFA and then combined into a
single LDDT percentage.
Table 1-7 presents the percentage of sales for each
manufacturer that would require ceramic traps under the 0.26
g/mi particulate standard. For simplicity these calculations
were not done for the wire mesh trap, as the effect of using
wire mesh traps instead of ceramic traps was estimated in
Section II.E. for LDDVs and, given present data, the ceramic
trap appears to have advantages over the the wire mesh trap in
terms of cost and trapping efficiency. If the percentages of
wire mesh traps required per manufacturer were desired, they
could be easily approximated by applying the ratio of the
percent of LDDVs which would require wire mesh traps to the
percent of LDDVs which would require ceramic traps (see Section
II.E.)
From Table 1-7, the industry-wide percentage of sales that
would require ceramic traps under the base scenario is
estimated to be 7.6 percent.
IV. Heavy-Duty Diesels
A. Current Emission Level and Non-Trap Standards
Currently there is no particulate standard for heavy-duty
diesel engines (HDDEs). Therefore, there are no certification
test data from which to determine the current levels of HDD
particulate emissions. However, there has been a substantial
amount of HDD particulate testing over SPA's new transient
cycle by both EPA and the industry. Table 1-8 contains
particulate and NOx emission data from manufacturers'
production and development tests,[8] the EMA/EPA HDD
"round-robin" testing program,[9] and EPA's original diesel
transient baseline[10] (for those engines for which more recent
data are not available) . Although data are not available for
every HDD engine family, a large majority of sales is
represented. Sales weighting the data in Table 1-8 indicated
-------
Table 1-7
Percentage of LDDT Sales Requiring Traps Under Various
Particulate Standards (assumes "averaging")
0.26 g/mi
Part. Std. with
Ceramic Trap
Small LDDTs:
Ford 12.3
Isuzu 0.0
Nissan 31.9
Mitsubishi 40.1
Toyota 0.0
Volkswagen 15.4
Toyo Kogyo 11.5
Full-Size LDTs:
General Motors 6.9
Industry-wide 7.6
Sales-Weighted
Percentage
-------
Table 1-8
Low-Mileage, Transient Emissions
From Current Heavy-Duty Diesel Engines
Manufacturer/
Engine
Caterpillar
3208 DINA
3208 DIT
3406 DITA
3406 PCTA
3306 DITA
3306 PCTA
Cummins
NTC 290
NTC 350
NTC 350 (Big Cam)
NTCC 240
NTCC 400
NH 250
VTB-903
Daimler-Benz
OM 344A
OM 362A
Detroit Diesel
8V-71N
8V-71TA
6V-92TA
8V-92TA
8.2-T
International Harvester
DT-466B
DTI-466B
Mack
ETAZ-676
ETSX-676
ETSZ-676
Particulate
(g/BHP-hr)
0.65
0.59
0.52-0.71
0.37
0.73
0.50
0.59
0.58-0.70
0.40
0.77
0.85
0.52-0.83
0.67
0.81
0.45
0.79
0.35-0.43
0.55-0.67
0.46
0.43
0.53
0.67
0.31-0.36
0.58
0.63-0.69
0.59
NOx
(g/BHP-hr
7.8
10.0
7.9-8.
5.4
9.0
4.8
8.3
7.2-9.
6.8
4.8
5.3
6.8-6.
5.2
5.1
6.7
5.7
6.7-7.
5.8
7.8
5.0-5.
5.7
4.2
5.6-5.
5.2
5.2
6.9
)
4
0
9
6
9
7
-------
1-22
an average particulate emission level of around 0.60-0.65
g/BHP-hr. After allowing for some deterioration (these engines
were almost entirely new), it is estimated that today's HDDs
emit at an average of 0.7 g/BHP-hr in-use.
While the 0.7 g/BHP-hr level is appropriate for today's
engines, future HDDs should be able to reach somewhat lower
particulate levels with relatively minor engine modifications
and recalibrations. The impetus to control HDD particulate
(other than the particles constituting "smoke" at certain
extreme engine operation modes) has not yet occurred since
there has been no particulate standard. With a standard,
however, some reduction in particulate emissions should occur.
For example, in its comments to the HDD particulate NPRM,[11]
Caterpillar recommended a future non-trap standard of 0.6
g/BHP-hr, including DF and safety margin. For the purposes of
this analysis, this level will be used as the non-trap,
non-technology forcing, HDDE particulate standard to be
implemented sometime in the 1987-88 timeframe. Also, for the
purposes of this analysis, we have assumed that this standard
would be implemented in 1988. Thus, without the use of
trap-oxidizers, HDDEs will be projected to emit at 0.7 g/BHP-hr
through 1987 and at 0.6 g/BHP-hr thereafter.
B. Standard Level With Traps
The trap-based HDDE particulate standard of the base
scenario is 0.25 g/BHP-hr. This level was proposed by the
Agency in its HDDE particulate NPRM (46 FR 1910) . The
percentage of HDDEs that would require traps under this
standard is 100 percent because it was proposed as a
non-averaging standard and all HDDEs currently emit
substantially above 0.25 g/BHP-hr. (The effect of averaging
will be considered later in this section.)
The 0.25 g/BHP-hr standard requires a 60 percent reduction
in particulate emissions from the 0.6 g/BHP-hr non-trap level
mentioned above. Both the ceramic trap and the wire mesh trap
have efficiencies greater than 60 percent. Under the base
scenario without averaging, it has been assumed that
manufacturers would only apply traps of the required efficiency
regardless of the type of trap used. This is to say that even
if ceramic traps were applied, there would be sufficient
impetus to reduce efficiency below that achievable (e.g., to
increase regeneration intervals and reduce backpressure and
fuel economy penalties) if the standard were more stringent,
that only the efficiency actually necessary, with a reasonable
safety margin, would be applied. This efficiency has been
assumed to be 65 percent. Applying this 65 percent efficiency
to the engine-out emission standard level of 0.6 g/BHP-hr,
results in tailpipe emissions of 0.21 g/BHP-hr under the base
scenario. This is somewhat lower than the required 0.25
g/BHP-hr, but is appropriate because it is believed that HDD
-------
1-23
manufacturers will desire a somewhat larger safety margin due
to the variety of HDD application and the absence of averaging.
If averaging were implemented along with the 0.25 g/BHP-hr
standard for HDDEs, the percentage of vehicles requiring traps
would drop from 100 to about 70 percent. In this case, we have
assumed that manufacturers would utilize the full 85 percent
efficiency of the ceramic trap in order to take full advantage
of averaging.
-------
1-24
References
1. "Light-Duty Diesel NOx-HC-Particulate Trade-Off
Studies," In: Diesel Combustion and Emissions: Proceedings
of SAE Congress and Exposition, p. 86, Wade, W. R., SAE Paper
No. 800335, February 1980.
2. "Regulatory Analysis and Environmental Impact of
Final Emission Regulations for 1984 and Later Model Year
Heavy-Duty Engines," U.S. EPA, OANR, QMS, ECTD, SDSB, pp.
184-86, December 1979.
3. This Data is Publicly Available From the U.S. EPA
Selective Enforcement Section, Manufacturer's Operations
Division, Office of Mobile Sources.
4. "Trap-Oxidizer Feasibility Study," U.S. EPA, OANR,
OMS, ECTD, SDSB, March 1982.
5. "Light-Duty Diesel Organic Material Control
Technology Investigation Program," EPA Contract No. 68-03-2873,
Monthly Progress Report No. 34, August 10, 1982.
6. Letter from B. E. Enga, Johnson-Matthey, Inc., to
Anne M. Gorsuch, Administrator, U.S. EPA, Regarding the 1985
Light-Duty Diesel Particulate Standards, January 25, 1982 (EPA
Docket A-81-20, II-D-75) .
7. "The Impact of Light-Duty Diesel Particulate
Standards on the Level of Diesel Penetration in the Light-Duty
Vehicle and Light-Duty Truck Markets," Jack Faucett Associates,
For U.S. EPA, Contract No. 68-01-6375.
8. This Data Was Submitted by Heavy-Duty Diesel Engine
Manufacturers As Comments to EPA's Heavy-Duty Diesel
Particulate NPRM (46 FR 1910) and Can Be Found In EPA Public
Docket No. A-80-18.
9. "EMA/EPA Heavy-Duty Diesel Engine Cooperative Test
Program," EPA Public Docket No. A-80-18, November 1982.
10. "Emissions .From Heavy-Duty Engines Using The 1984
Transient Test Procedure, Volume 2 - Diesel," U.S. EPA, OANR,
OMS, EPA-460/3-81-031, July 1981.
11. This Data is Contained In Caterpillar Tractor
Company's Comments to the Heavy-Duty Diesel Particulate NPRM
(46 FR 1910). Caterpillar's Comments Can Be Found In EPA
Public Docket No. A-80-18.
-------
CHAPTER 2
EMISSIONS IMPACTS
I. Introduction
This chapter assesses the impact of the base and relaxed
scenarios on total nationwide and urban diesel particulate
emissions in 1995 as compared to those in 1980 and 1986. The
base scenario assumes particulate standards of 0.20 g/mi, 0.26
g/mi, and 0.25 g/BHP-hr for light-duty diesel vehicles (LDDVs),
light-duty diesel trucks (LDDTs) and heavy-duty diesel engines
(HDDEs), respectively. The relaxed scenario assumes
non-technology forcing, non-trap particulate standards for all
three vehicle classes (i.e., LDDVs and LDDTs will continue to
emit at current particulate levels, which are well below the
current standard of 0.6 g/mi, while HDDEs will emit at a level
of 0.6 g/BHP-hr beginning in 1988). Under both scenarios, the
current NOx standards for LDDVs and LDDTs (i.e., 1.5 and 2.3
g/mi, respectively) are assumed to remain in effect. The HDDE
NOx standard is not identified per se, but must be of such
stringency as to allow a non-trap particulate standard of 0.6
g/BHP-hr to be met.
The first section of this chapter estimates 1980, 1986 and
1995 particulate emission factors by vehicle type and model
year under the two control scenarios. The second section
calculates nationwide and urban emissions for both control
scenarios by combining these emission factors with vehicle
miles traveled (VMT), breakdowns by model year, diesel sales
fractions, and nationwide and urban VMT projections. The third
section compares some of these results with those of previous
EPA analyses.
II. Emission Factors
The initial step in determining nationwide and urban
diesel particulate emissions is to estimate emission factors
for the vehicles of each model year which comprise the 1980,
1986 and 1995 fleets. Generally speaking, emission factors are
the average emission rates (in g/mi) that vehicles of a certain
type and age are expected to emit during in-use operation.
Emission factors usually must be determined through in-use
testing because owner problems such as tampering, improper
maintenance, and abuse can substantially change actual emission
levels from certification test levels. However, studies[1,2]
have shown that in-use particulate emissions from diesel
engines remain at certification test levels (with appropriate
allowance made for normal deterioration) throughout the life of
the vehicle (i.e., the owner-related problems mentioned above
do not appear to significantly influence diesel particulate
-------
2-2
emissions) . Therefore, the diesel particulate emission factors
estimated for this study are derived from current certification
data in the case of LDDVs and LDDTs and from manufacturer and
Agency test data in the case of HDDs. These data sources are
fully described in Chapter 1.
A. Relaxed Scenario
1. Light-Duty Diesel Vehicles and Light-Duty Diesel
Trucks
The projected post-1980 LDDV and LDDT emission factors
under the relaxed scenario are easily determined, since it is
assumed that these vehicles will continue to emit at their
current levels. These current levels have already been
determined in Chapter 1 and are simply the achievable half-life
particulate standard levels shown in Table 2 of that chapter.
As discussed in Chapter 1, the achievable particulate standard
level is the current certification test level multiplied by the
50,000-mile deterioration factor (DF) and a 10 percent safety
margin (to account for production variability). Since the
lifetime of a typical LDDV or LDDT is about 100,000 miles, the
half-life standard level can be viewed as the average emission
rate over the life of the vehicle. That is, for the first
50,000 miles of its life, the vehicle will emit below the
standard level and for the second 50,000 miles the vehicle will
emit above the standard level.
Weighting these emission levels by the projected 1983
sales of each configuration yields fleet average emission
factors of 0.27 g/mi for LDDVs and 0.28 g/mi for LDDTs. These
emission factors will be applied to each and every model year
represented in the 1995 calendar-year fleet. Strictly
speaking, this would not be the case since older vehicles
generally have higher emissions due to more deterioration and
vice versa. However, the 50,000 mile deterioration factors for
LDDVs and LDDTs are less than 1.1 on the average (i.e., a 10
percent increase in 50,000 miles). Thus, while the emission
factor for newer vehicles is slightly overestimated
(deterioration at this point is less than average) , the
emission factor for older vehicles is slightly underestimated,
and the net result is virtually the same as if each model
year's vehicles were assigned slightly different emission
factors based on the deterioration occurring between individual
model years.
Pre-1980 model year vehicles generally emitted higher
levels of particulate than those of later years. Emission
factors for these years were estimated from the historical
emission levels and sales of these vehicles[3] and are shown
below:
-------
2-3
Model Year LDDV LDDT
1980 0.5 0.5
1979 0.8 0.9
1978 0.7 0.9
1975-77 0.5 0.5
1971-74 0.5
2. Heavy-Duty Diesels
Estimating emission factors for HDDEs is much more
complicated than estimating emission factors for LDDVs and
LDDTs/ because HDDE emissions are measured in terms of grams
per brake horsepower-hour (g/BHP-hr) and not g/mi, as only the
engine is tested and not the entire vehicle. Because vehicle
emissions (in g/mi) can vary widely at a constant g/BHP-hr
engine emission level, due to widely varying vehicle weights
and sizes, the conversion of g/BHP-hr emission rates to g/mi
equivalents in order to obtain HDD emission factors is not a
simple process.
The general equation relating engine emission rate and
vehicle emission rate is as follows:
Vehicle emission factor (g/mi) = [engine emission rate
(g/BHP-hr) x diesel fuel density (7.1 Ib/gallon)]/[engine
brake-specific fuel consumption (Ib fuel/BHP-hr) x vehicle
fuel economy (miles/gallon)].
It was determined in Chapter 1 that the engine emission
rate under the relaxed scenario would be 0.7 g/BHP-hr for 1987
and earlier HDDs and 0.6 g/BHP-hr for 1988 and later HDDEs.
This leaves two factors still to be determined: vehicle fuel
economy and engine brake-specific fuel consumption (BSFC).
a. Heavy-Duty Diesel Fuel Economy Estimates
The fuel economy of heavy-duty diesel vehicles (HDDVs),
like that of other vehicle types, is expected to increase in
the future. This necessitates the use of projections and
prevents the sole use of current HDDV fuel economy data.
Present and future HHDV fuel economies were estimated for
four vehicle subgroups based on an analysis of data from
various sources. (This analysis is contained in Reference 4.)
The HDDV subgroups are defined by gross vehicle weight rating
(GVWR) as follows:
Class IIB = 8,500 up to 10,000 Ibs.
Classes III-V = 10,001 to 19,500 Ibs.
Class VI = 19,501 to 26,000 Ibs.
Classes VII and VIII = 26,001 Ibs. and up.
-------
2-4
Current fuel economies for Classes IIBf III-V, and VI were
derived from fuel consumption modeling results published by the
Energy and Environmental Analysis, Inc. (EEA) for the U.S.
Department of Energy. The EEA estimates were not used directly
because the fuel consumption values are based on total VMT and,
hence, are more indicative of highway fuel consumption rather
than urban fuel consumption. This latter parameter is the most
important here since the objective of this study is primarily
to evaluate the environmental impact of particulte emissions in
urban areas. Therefore, the EEA estimates for Classes IIB,
III-V, and VI were reduced by 20 percent to represent urban
fuel economies.
The current fuel economy for Classes VII-VIII was taken
from test results collected by Southwest Research Institute
(SwRI) under contract to EPA. These data were obtained using
urban test cycles and, hence, are already representative of
urban fuel consumption. For comparison, the EEA value for
Classes VII-VIII is generally about 25 percent higher than the
SwRI estimate.
Future fuel economy improvements for the four HDDV
categories were derived from the above-mentioned EEA modeling
results. As before, the values for Classes IIB, III-V, and VI
were reduced by 20 percent to reflect urban fuel consumption.
For Classes VII-VIII, the EEA estimates still appeared to
remain more representative of highway fuel usage rather than
urban fuel usage even if they were reduced by 20 percent. This
is explained in that many of the expected fuel economy
improvement technologies for these larger vehicles should be
more beneficial during highway cruising than during the
stop-and-go driving which is characteristic of urban areas. To
account for this difference, the largest overall increase of
the other three categories (i.e., 15 percent improvement from
1980 to 1991) was also used to represent the fuel economy
improvement for category VII-VIII.
The HDDV fuel economy estimates are shown in Table 2-1.
b. Heavy-Duty Diesel Brake-Specific Fuel Consumption
The second factor of the vehicle-emission equation which
needs to be estimated is HDDE brake-specific fuel consumption
(BSFC). As with HDDV fuel economies, estimates of BSFC for the
four weight categories were based on an analysis of data from
various sources. (This analysis is contained in Reference 4.)
The important factors which are used to identify future fuel
consumption improvements are the: 1) engine fuel-saving
technologies, 2) urban fuel economy gains for each technology,
and 3) market penetration of each technology.
-------
2-5
Table 2-1
HDDV Fuel Economies (mpg)
Model Year
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980 +
Class IIB
13.1
13.1
13.0
13.0
13.0
13.0
12.8
12.7
12.5
12.3
12.2
12.0
11.8
11.7
11.6
11.4
Classes III-V
10.6
10.4
10.2
10.1
9.9
9.8
9.7
9.7
9.6
9.6
9.5
9.4
9.4
9.4
9.3
9.2
Class VI Classes VII-VIII
7.6
7.6
7.6
7.6
7.6
7.6
7.5
7.5
7.4
7.4
7.4
7.3
7.2
7.1
7.0
7.0
5.12
5.10
5.09
5.07
5.06
5.04
4.98
4.92
4.85
4.79
4.73
4.67
4.62
4.56
4.50
4.45
-------
2-6
Table 2-2 presents HDDE BSFC by model year.
c. Heavy-Duty Diesel Emission Factors
Having estimated fuel economies and brake-specific fuel
consumptions for each of the four HDD groups, the emission
factors (g/mi) for each group by model year were calculated
using the vehicle-emissions equation and are shown in Table 2-3.
These HDD emission factors, like those for LDDVs and
LDDTs, all include half-life (or average) deterioration and the
fact that newer vehicles have slightly lower emissions, and
older vehicles slightly higher emissions, is ignored. This is
again very acceptable, since deterioration of HDD particulate
emissions should be very low (about 15 percent over the life of
the vehicle).
/
B. Base Scenario
The base scenario differs from the relaxed scenario only
in the fact that some vehicles in the base scenario are
equipped with trap-oxidizers. Thus, except for any unique
features of trap-oxidizers which affect in-use emissions, the
methodology used here is the- same as that described above for
the relaxed scenario. That is, certification data with average
deterioration and an appropriate safety margin are assumed to
adequately represent in-use emissions. (Emission factors for
calendar years 1980 and 1986 do not need to be readdressed
since the base-scenario standard does not take effect until
1988.)
The one feature of trap-oxidizers which may affect this
relationship between certification and in-use emissions is the
possibility of trap failure. Trap-oxidizer systems are not
currently being used on any vehicles, and therefore, there are
no data on their reliability in-use. Limited data on
trap-oxidizer system durability has been generated by
experimental testing programs.[5] These programs have
demonstrated that traps can physically undergo repeated
regeneration cycles over 50,000 miles of vehicle operation and
still maintain their initial trapping efficiencies. However,
these test programs involved only a few vehicles and somewhat
controlled operating conditions. It is possible that when put
into general use, some failures of trap-oxidizer systems -will-
occur .
The reasons for failure of a trap-oxidizer system can be
divided into two general categories: 1) failure of the
electronic control system used to regenerate the trap, and 2)
physical failure of the trap due to unforeseen operating
conditions. Electronic control systems consisting of
-------
2-7
Table 2-2
HOPE Fuel Consumptions (Ibm fuel/BHP-hr)
Model Year
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980+
Class IIB Classes III-V
0.408
0.408
0.411
0.411
0.411
0.411
0.415
0.415
0.418
0.418
0.421
0.424
0.424
0.424
0.427
0.430
0.387
0.393
0.399
0.403
0.411
0.411
0.415
0.415
0.418
0.418
0.421
0.424
0.424
0.424
0.427
0.430
Class VI Classes VII-VIII
0.406
0.406
0.406
0.406
0.406
0.406
0.411
0.411
0.416
0.416
0.416
0.421
0.426
0.430
0.430
0.430
0.390
0.391
0.391
0.392
0.392
0.393
0.397
0.401
0.405
0.410
0.412
0.416
0.420
0.423
0.427
0.430
-------
2-8
Table 2-3
Relaxed Scenario HDDV Emission Factors (g/mi)
Model Year
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980 +
Class IIB Classes III-V
0.797
0.797
0.797
0.797
0.797
0.797
0.802
0.808
0.951
0.967
0.968
0.977
0.993
1.002
1.003
1.014
1.038
1.042
1.047
1.047
1.047
1.058
1.058
1.058
1.239
1.239
1.243
1.247
1.247
1.247
1.252
1.256
Class VI Classes VII-VIII
1.381
1.381
1.381
1.381
1.381
1.381
1.382
1.382
1.614
1.614
1.614
1.617
1.620
1.628
1.651
1.651
2.133
2.136
2.140
2.143
2.148
2.151
2.155
2.159
2.530
2.531
2.550
2.558
2.561
2.577
2.587
2.597
-------
2-9
microprocessors and central processing units (CPUs) have come
into widespread use on light-duty vehicles since 1980. These
control systems, used in conjunction with three-way catalysts,
are necessary to attain the 1981 emission standards for many
vehicles. Recent testing of in-use vehicles by EPA's Emission
Factor Testing Program[6] has generated data on the failure
rate of these electronic control systems. That data indicates
that 1.5 to 2.0 percent of 1-year old light-duty vehicles of
1981-82 vintage are gross emitters of HC and CO. Since it is
reasonable to assume that the reason for the gross emissions is
failure of the electronic control system, it can be concluded
that the failure rate for electronic control systems for these
model year vehicles was about 1.5 to 2.0 percent per year. It
should be noted that these results are based on a limited
number of vehicle tests and could be subject to change in the
future.
This failure rate should be adjusted to account for the
fact that this electronic control technology is relatively new
and that, for the purposes of this study, trap-oxidizer systems
will not be required before 1987. The industry has five more
years to reduce the failure rate of electronic control
systems. Therefore, the failure rate for 1987 and later
electronic control systems used on trap-oxidizers is estimated
to be 1.0 percent per year.
The other general category of trap-oxidizer system
failure, as mentioned above, is the occurance of unforeseen
operating conditions. Manufacturers will design trap-oxidizer
systems to withstand almost every in-use condition they can
foresee. However, it is still possible that certain operating
conditions will occur which prevent proper regeneration of the
trap, thus, leading to eventual trap failure. Therefore, a
failure rate of 0.5 percent per year will be used in this
analysis for this second type of trap-oxidizer system failure.
Adding the electronic control system failure rate to the
unforeseen operating conditions failure rate yields an overall
failure rate of 1.5 percent per year for LDDVs with traps.
This overall failure rate will also be used for LDDTs and
MDV/LHDVs because their annual mileages and lifetimes are
similar to those for LDDVs. HDDVs, however, while having
approximately the same lifetime as these other vehicles, are
driven, on the average, substantially more miles per year.
Therefore, the 1.5 percent per year failure rate was adjusted
for HHDVs to reflect the greater (factor of four) annual number
of miles by these vehicles. In doing this, the 1.0 percent per
year electronic failure rate was held constant since these
types of failures were assumed to be primarily due to factors
such as time and transients in engine compartment temperature,
which here depend more on time than annual vehicle mileage.
-------
2-10
The 0.5 percent per year failure rate due to the occurence of
unforeseen operating conditions, on the other hand, was assumed
to be partially dependent on annual mileage and was doubled.
Thus, the trap failure rate used for HDDVs was 2.0 percent per
year.
Having determined the trap-oxidizer system failure rates
for the different vehicle types, these failure rates can be
combined with the basic methodology used to estimate the
emission factors under the relaxed scenario to estimate
emission factors under the base scenario. The results of this
combination are shown in Table 2-4.
The 1995 emission factors for LDDVs and LDDTs of model
years 1978 through 1986 and for HDDs from 1978 through 1987 are
the same as those under the relaxed scenario. This occurs
because the more stringent particulate standards of the base
scenario do not become effective until 1987 in the case of
LDDVs and LDDTs and 1988 in the case of HDDs.
When the new standards do become effective, it is again
assumed that vehicles will emit, on the average, at their
applicable standard levels, except for the effect of trap
failure. These applicable standard levels are 0.20 g/mi for
LDDVs, 0.26 g/mi for LDDTs, and a 60 percent reduction from the
relaxed-scenario levels identified in the previous section for
HDDs. To these levels must be added the effect of trap
failure. This is done according to the following equation:
Model Year Emission Factor = Standard Level + (Vehicle
Age) x (Trap Failure Rate) x (Fraction of Vehicles with
Traps) x (Difference between Non-trap Emissions and
Standard Level)
Some of the terms in the above equation deserve some
elaboration. Vehicle age is assumed to 0.5 years for 1995 model
year vehicles and one year greater for each preceding model
year. Trap failure rate is 1.5 percent for LDDVs, LDDTs and
MDV/LHDVs and 2.0 percent for HHDVs. The fraction of vehicles
with traps is included in the above equation because the trap
failure rate should only be applied to vehicles with traps.
This figure is 0.223 for LDDVs, 0.076 for LDDTs and 1.00 for
all HDDVs.
The difference between non-trap emissions and standard
level is included to account for the fact that the vehicle
emissions should simply revert back to their non-trap levels if
the trap should fail. The standard level is subtracted because
emissions up to this level have already been taken into account
by the first term on the right hand side of the equation
(standard level). In the case of HDDVs, the non-trap emissions
-------
2-11
Table 2-4
Base Scenario Emission Factors (g/mi)
Vehicle
Model Year
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978+
LDDV
.200
.201
.202
.202
.203
.204
.204
.205
.205
.270
.270
.270
.270
.270
.270
.5
.8
.7
LDDT
.260
.260
.260
.260
.261
.261
.261
.261
.261
.280
.280
.280
.280
.280
.280
.5
.9
.9
HDV
Class
IIB
0.336
0.343
0.350
0.357
0.364
0.371
0.380
0.390
0.951
0.967
0.968
0.977
0.993
1.002
1.003
1.014
1.014
1.014
HDV
Classes
III-V
0.437
0.448
0.459
0.468
0.477
0.492
0.501
0.510
1.239
1.239
1.243
1.247
1.247
1.247
1.252
1.256
1.256
1.256
LHDV
0.581
0.593
0.605
0.618
0.630
0.642
0.654
0.667
1.614
1.614
1.614
1.617
1.620
1.628
1.651
1.651
1.651
1.651
HHDV
0.901
0.928
0.954
0.981
1.008
1.034
1.061
1.089
2.530
2.531
2.550
2.558
2.561
2.577
2.587
2.597
2.597
2.597
-------
2-12
are simply those occurring under the relaxed scenario, because
all HDDVs were assumed in Chapter 1 to emit at the same level
(i.e., the non-trap level of trap-equipped vehicles is the same
as the emission level of vehicles without traps). However, a
distribution of vehicle emissions was determined in Chapter 1
for LDDVs and LDDTs and traps were placed on the highest
emitting vehicles first. Thus, the non-trap levels for
trap-equipped LDDVs and LDDTs (0.392 g/mi and 0.334 g/mi,
respectively) are higher than the non-trap levels of the
relaxed scenarios (0.270 g/mi and 0.280 g/mi, respectively).
III. Nationwide and Urban Emissions
The next step in determining nationwide and urban
emissions is to combine the "model year" emission factors
generated in the previous section for each vehicle type into a
single, weighted calendar-year emission factor for each vehicle
type. This is done by multiplying each model year's emission
factor by that model year's fraction of calendar-year VMT and
the diesel sales fraction for that model year, and then
summming across all model years. The result is an emission
factor that is appropriately weighted by both the number of
diesels on the road, relative to total vehicles, and by their
age. In other words, the 1995 weighted emission factor is now
on a total (i.e., gasoline and diesel combined) VMT basis for
that vehicle type.
The breakdown of VMT by model year [7] are shown in Table
2-5 for LDDVs, LDDTs, and HDDVs. It should be noted that the
VMT breakdown shown for HDDV Classes IIB, III-V, and VI is that
given in the reference for gasoline-fueled HDVs and the VMT
breakdown shown for HHDV Classes VII-VIII is that for
diesel-powered HDVs. This is appropriate because at the time
the referenced study was performed, the great majority of
gasoline-fueled HDVs were in Classes IIB-VI and nearly all HDDs
were in Classes VII and VIII. However, the use of the
historical Class IIB-VI breakdown here does assume that the
dieselization of this class will not alter this breakdown.
The diesel sales fractions for each model year are shown
in Table 2-6 for LDDVs and LDDTs, and in Table 2-7 for HDDVs.
Two sets of projections are used in this study. The first is a
"best estimate" projection and is based on a continuation of
present conditions, including the absence of a major oil
crisis. This results in moderate growth of diesel sales. The
second set is a "worst case" projection, which could be
realized if another oil crisis were to occur. Here, the rate
of diesel sales is substantially higher than under the best
estimate projections. The term "worst case" refers to the
degree of environmental impact which would occur due to diesel
particulate emissions.
-------
2-13
Table 2-5
Total VMT for LDDVs, LDDTsy and HDDVs
Model Year
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
LDDV
.091
.124
.108
.080
.100
.107
.088
.067
.059
.050
.038
.026
.021
.015
.009
.006
.003
.001
LDDT
.159
.137
.108
.072
.096
.098
.068
.050
.035
.035
.032
.021
.022
.019
.014
.011
.007
.005
Classes
IIB, III-V*
Class VI
.201
.161
.124
.084
.090
.083
.059
.041
.029
.028
.024
.017
.015
.012
.009
.007
.005
.003
HDV
Classes
VII-VIII
.247
.188
.102
.058
.093
.080
.056
.038
.029
.028
.020
.015
.015
.011
.007
.005
.003
.001
These VMT fractions are used for each HDV subgroup
separately.
-------
2-14
Table 2-6
Diesel Fraction of Total Sales for LDDVs and LDDTs
Best Estimate
Model Year
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
1972
1971
1970+
LDDV
.115
.115
.114
.114
.113
.113
.100
.090
.080
. .073
.066
.060
.053
.046
.061
.034
.028
.009
.004
.003
.003
.003
.003
.003
.003
.000
LDDT
.339
.330
.321
.312
.303
.294
.27
.240
.210
.180
.160
.130
.100
.080
.060
.034
.028
.009
.005
.003
.002
.000
.000
.000
.000
.000
Worst Estimate
LDDV
.300
.290
.280
.270
.260
.250
.220
.190
.160
.130
.100
.070
.053
.046
.061
.034
.028
.009
.004
.003
.003
.003
.003
.003
.003
.000
LDDT
.600
.560
.520
.480
.440
.400
.360
.320
.280
.240
.200
.150
.100
.080
'.060
.034
.028
.009
.005
.003
.002
.000
.000
.000
.000
.000
-------
2-15
Table 2-7
Diesel Fraction of Total Sales for HDDVs
Best Estimate
Worst Estimate
Model
Year
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
1972
1971
1970
1969+
Class
IIB
.371
.357
.343
.329
.315
.301
.287
.273
.259
.245
.231
.179
.126
.074
.037
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
Classes
III-V
.476
.463
.449
.436
.422
.409
.396
.382
.369
.355
.342
.264
.186
.108
.054
.000
.000
.000
.000
.000
.004
.001
.003
.020
.020
.020
.000
Class
VI
.669
.645
.621
.598
.574
.550
.526
.502
.479
.455
.431
.369
.286
.214
.164
.114
.114
.078
.070
.042
.032
.016
.016
.016
.015
.016
.000
Classes
VI I -VI I I
.983
.980
.978
.975
.973
.970
.967
.965
.962
.960
.957
.947
.937
.928
.918
.91
.89
.88
.85
.83
.73
.77
.78
.76
.75
.75
.75
Class
IIB
.895
.841
.789
.741
.694
.648
.546
.546
.503
.422
.344
.256
.126
.074
.037
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
Classes
III-V
1.000
1.000
1.000
.949
.910
.864
.764
.764
.716
.612
.510
.377
.186
.108
.054
.000
.000
.000
.000
.000
.004
.001
.003
.020
.020
.020
.000
Class
VI
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
.929
.784
.642
.513
.286
.214
.164
.114
.114
.078
.070
.042
.032
.016
.016
.016
.016
.016
.000
Classes
VII-VIII
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
.949
.958
.937
.928
.918
.91
.89
.88
.85
.83
.73
.77
.78
.76
.75
.75
.75
-------
2-16
Regarding the best estimate diesel sales fractions,
historical diesel and total sales data were used for model
years 1961-82. The LDD sales fractions for model years 1990
through 1995 are those determined in a study[8] for EPA by Jack
Faucett Associates which investigated the impact on diesel
penetration in the LDV and LDT markets of diesel particulate
standards. The LDDV and LDDT sales fractions for model years
1983-89 were obtained by linearly interpolating between the
figures for 1982 and 1990. The 1985, 1990, and 1995 HDDV sales
fractions were derived from projections made by Data Resources
Inc.,[9] with the in-between years again being obtained by
linear interpolation.
Regarding the worst case sales fractions, in-house
estimates were used to represent what is considered to be the
maximum diesel penetration in this timeframe. For model years
1961-83, the diesel sales fractions are, or course, identical
to the best estimate diesel sales fractions because they are
based on historical data.
For LDDVs, a maximum penetration rate for the 1995 model
year was projected to be 30 percent. It was also thought that
most of the increase in diesel penetration between 1984 and
1995 would occur in the first half of this time span. thus,
the LDDV penetration rate rises by three percentage points per
year from 1984 through 1990 and after which rises by only one
percentage point per year through 1995.
For LDDTs, a maximum penetration rate of 60 percent was
projected for 1995. Unlike LDDVs, however, the increase in
LDDT dieselization is likely to be more consistent with time,
due to the fact that significant dieselization is already
occurring under the best estimate projections. Therefore, a
constant increase of four percentage points per year was
projected from 1985 through 1995.
For the four classes of HDDVs, the worst case
dieselization rates were derived by estimating the year that
total dieselization would occur and then by linear
interpolation to historic levels. these years were 1997 for
Class lib, 1993 for Classes III-V, 1988 for Class VI, and 1986
for Classes VII-VIII.
The weighted emission factors (g/mi) for each calendar
year, vehicle type, control scenario, and diesel sales scenario
are shown in Table 2-8, along with estimates of total VMT
(gasoline plus diesel) and the urban fraction of VMT for each
vehicle type.
The urban/rural splits were obtained from U.S. Federal
Highway Administration data.[10] It should be noted that this
-------
Table 2-8
Weighted Emission Factors and Projected VMT
Weighted Emission Factor (g/mi)
Calendar Year 1980:
HDV
Class
LDDV LDDT I IB
HDV
Classes
III-V
HDV
Class
VI
HDV
Classes
VII-VIII
All Scenarios
Calendar Year 1986;
Best Estimate Diesel Sales
Worst Case Diesel Sales
Calendar Year 1995:
Best Estimate Diesel Sales
Relaxed Scenario
Base Scenario
Worst Case Diesel Sales
Relaxed Scenario
Base Scenario
Projected VMT (109 miles);
1930
1986
1995
Urban Fraction
of VMT (all years)
0.0059 0.0075 0.000 0.0027 0.1092
0.0135 0.0268
0.0164 0.0316
0.1262 0.2393 0.4627
0.1876 0.3564 0.6589
0.0270 0.0762
0.0209 0.0712
0.0606 0.1172
0.0465 0.1094
1,109 232.3
1,209 267.2
1,537 329.5
0.2516 0.4377 0.7968
0.1250 0.2195 0.3997
0.5585 0.9116 1.2987
0.2685 0.4447 0.6499
59.4
48.8
13.79
34.47
40.54
48.8
6.63
5.03
4.69
48.8
18.40
13.99
9.97
48.8
2.1883
2.3746
2.4133
2.1337
1.1394
2.1831
1.1659
67.9
84.0
117.9
26.9
fO
I
-------
2-18
urban/rural VMT data for HDVs was not broken according to
vehicle size but by generic type (i.e., bus, single-unit truck,
tractor-trailer combination). It was assumed that buses and
single-unit trucks were Classes IIB-VI vehicles and that
tractor-trailers were Class VII and VIII vehicles.
Nationwide emission estimates are obtained by simply
multiplying the weighted emission factors by VMT. Urban
emissions are obtained by multiplying the nationwide emissions
estimates by the urban VMT fraction. These figures are shown
in Tables 2-9 and 2-10. It should be noted that the emission
estimates in these tables for HDDV Classes IIB, III-V, and VI
have been combined into a single category labelled medium-duty
vehicle/light heavy-duty vehicle (MDV/LHDV) to ease the
presentation of the results. The subsequent discussion will
focus on the urban emission results of Table 2-10 as these are
the most pertinent with respect to human exposure to diesel
particulate emissions.
Table 2-10 has been arranged to depict a number of
effects. One, projections for calendar years 1980, 1986, and
1995 have been placed side-by-side to allow easy comparison.
Two, the effects of both the relaxed and base scenarios are
shown in 1995 to depict the effect of control. Because the
control of LDDVs and LDDTs provides so little control relative
to HDDV control, a modified base scenario has been added where
only HDDV emissions are controlled. Three, an attempt has been
made to depict the causes of the increases in total urban
emissions between 1980 and future years. Beside each emission
estimate for 1986 and 1995 is a percentage which indicates that
vehicle class1 contribution to the overall increase in urban
emissions between 1980 and that year. For example, LDDV
emissions are 24,600 metric tons per year in 1995 under the
best estimate, relaxed scenario. This is an increase of 20,700
metric tons per year from the 1980 level. The 30 percent
figure beside the 24,600 metric ton per year estimate indicates
that the 20,700 metric ton per year increase is 30 percent of
the total increase in urban emissions between 1980 and 1995,
69,300 metric tons per year.
Concerning the actual figures in Table 2-10, it can be
seen that urban emissions increase between 1980 and 1995
regardless of .the scenario chosen. The increase is smallest
for the best estimate, base scenario (57 percent) and largest
for the worst case, relaxed scenario (257 percent) . As can be
seen from the figures in parentheses, the largest contributor
to these increases are LDDVs. HHDVs contribute to the
increases under the relaxed scenarios, but actually serve to
mitigate such increases under the relaxed scenarios. Also,
while LDDVs, and LDDTs in some cases, produce the largest
emission increases, their control under the base scenario has
-------
2-19
Table 2-9
Nationwide Diesel Particulate Emissions
(metric tons per year)
Best Estimate Diesel Sales
LDDV
LDDT
MDV/LHDV
HHDV
Total
LDDV
LDDT
MDV/LHDV
HHDV
Total
1980
6,500
1,900
2,000
148,800
159,200
Worst
1980
6,500
1,900
2,000
148,800
159,200
1986
16,400
7,200
6,500
199,500
229,600
Case Diesel
1986
19,900
8,400
15,500
202,500
246,300
Relaxed
Scenario
41,500
25,000
19,300
251,300
337,100
Sales
Relaxed
Scenario
93,100
38,500
38,300
257,200
427,100
1995
Base
Scenario
32,200
23,600
9,600
133,900
199,300
1995
Base
Scenario
71,400
36,000
18,600
138,800
264,800
-------
Table 2-10
Urban Diesel Particulate Emissions
(metric tons per year)
Best Estimate Diesel Sales
1995
LDDV
LDDT
MDV/LHDV
HHDV
Total
1980
3,900
900
1,000
40,000
45,800
1986
9
3
3
53
70
,700
,500
,200
,700
,100
(24%)
(11%)
(9%)
(56%)
Worst
Relaxed
Scenario
* 24
12
9
67
113
Case
,600
,200
,400
,600
,800
(30%)
(17%)
(12%)
(41%)
19
11
4
36
71
Base
Scenario
,100 (60%)
,500 (41%)
,700 (14%)
,100 (-15%)
,400
1995
Base Scenario
Only HDD Control
24,600
12,200
4,700
36,100
77,600
Diesel Sales
1995
LDDV
LDDT
MDV/LHDV
HHDV
Total
1980
3,900
900
1,000
40,000
45,800
1986
11
4
7
54
78
,800
,100
,600
,500
,000
(24%)
(10%)
(21%)
(45%)
Relaxed
Scenario
55
18
18
69
162
,300
,800
,700
,200
,000
(44%)
(16%)
(15%)
(25%)
42
17
9
37
106
Base
Scenario
,400 (64%)
,600 (28%)
,100 (13%)
,000 (-5%)
,100
1995
(64%)
(36%)
(12%)
(-12%)
to
1
o
Base Scenario
Only HDD Control
55,300 (
18,800 (
9,100 (
37,000 (
120,400
69%)
23%)
11%)
-4%)
Figures in parentheses depict each vehicle class contribution to the overall
emissions increase over 1980 emissions (in percent). The sum of the
percentages for the four classes is 100 percent.
-------
2-21
the least effect. LDDV emissions are only reduced 22 percent
and LDDT emissions only 6 percent, as opposed to MDV/LHDV and
HHDV emission reductions of about 50 percent. Finally, the
effect of only controlling HDD emissions and avoiding further
control of LDDV and LDDT emissions is small. Overall urban
emissions only increase about 10-15 percent.
A final pertinent aspect of the urban emission estimates
of Table 2-10 is the relative contribution of each vehicle type
to overall urban emissions. Table 2-11 shows the fraction of
total urban emissions in each year being emitted by each
vehicle class. As can be seen, the relative contributions vary
depending on which situation is examined. One general
observation is that, despite its low urban VMT fraction, HHDVs
are still major contributors to urban emissions regardless of
diesel sales scenario (e.g., 31 to 45 percent under the relaxed
scenario).
IV. Comparison of Results with Previous Studies
It is also pertinent to compare the results of Table 2-10
to the projections of urban diesel particulate emissions of
previous studies. This was done for two cases: best estimate
and worst case diesel sales.
The Regulatory Analysis which accompanied the 1982
light-duty diesel particulate regulation[3] estimated
nationwide light-duty diesel particulates in the year 1990.
Two scenarios were analyzed: 1) an uncontrolled scenario where
light-duty diesel vehicles and trucks were projected to emit
1.6 g/mi particulate, and 2) a controlled scenario with a 0.6
g/mi standard for 1982-84 and a 0.2 g/mi standard for 1985 and
beyond (0.26 g/mi for light trucks). (This controlled scenario
is the same as the base scenario of this study, except here the
1985 standards have been delayed to 1987.) A range of
potential diesel penetrations was examined by applying a 4-25
percent bracket around a "best estimate" diesel sales
scenario. The LDDV NOx standard was presumed to be 1.0 g/mi
(part of the reason for the high uncontrolled particulate
emission factor).
This 1979 analysis estimated that 1990 urban emissions for
LDDVs and LOOTS would be 84,000-141,000 metric tons per year
under the uncontrolled scenario and 22,000-37,000 metric tons
per year under the controlled scenario. Extrapolating that
same methodology to 1995 (i.e., continued diesel penetration
into the in-use fleet and slightly increased total VMT), urban
emissions would have been projected to be 112,000-190,000
metric tons per year (uncontrolled) and 30,000-50,000 metric
tons per year (controlled).
-------
2-22
Table 2-11
Relative Contribution of
Urban Emissions (percent)
Best Estimate Diesel Sales
LDDV
LDDT
MDV/LHDV
HHDV
Total
LDDV
LDDT
MDV/LHDV
HHDV
1980
9%
2%
2%
87%
100%
1980
9%
2%
2%
87%
1986
14%
5%
5%
76%
100%
Worst
1986
15%
5%
10%
70%
1995
Relaxed Base
Scenario Scenario
22%
11%
8%
59%
100%
Case Diesel
1995
Relaxed
Scenario
34%
12%
11%
43%
26%
16%
7%
51%
100%
Sales
Base
Scenario
40%
17%
8%
35%
1995
Base Scenario
Only HDD Control
32%
16%
6%
46%
100%
Base Scenario
Only HDD Control
46%
16%
7%
31%
Total
100%
100%
100%
100%
100%
-------
2-23
As shown in Table 2-10, best estimate, urban emissions for
LDDVs and LDDTs for both the relaxed and base scenarios fall
within the previous estimates for the controlled scenario; both
scenarios resulting in emissions well below that for the
previous uncontrolled scenario.* Worst case urban emissions
under the relaxed scenario are greater than the upper limit for
the previous controlled scenario, but still well below that for
the uncontrolled scenario. Worst case emissions under the base
scenario are essentially equal to the upper limit of the
previous controlled scenario.
Moving to HDDVs, the Draft Regulatory Analysis
accompanying the heavy-duty diesel particulate NPRM estimated
1995 urban emissions to be 79,000-97,000 metric tons per year
(uncontrolled) and 28,200-34,600 metric tons per year
(controlled, 0.25 g/BHP-hr standard in 1986).[11] These 1980
estimates are closer to those in Table 2-10 than the previous
light-duty diesel estimates. For best estimate sales, the
current relaxed-scenario estimate is about equal to the lower
limit of the previous uncontrolled scenario estimate and is
only about 20 percent less than the upper limit of the previous
uncontrolled scenario estimate. The current base-scenario
estimate is only about 5-20 percent higher than the previous
The great majority of the difference between the estimates
for the relaxed scenario of this study and the
uncontrolled scenario of the previous study is due to the
difference in projected emission factors. The previous
study projected a uncontrolled particulate emission factor
of 1.0 g/mi while this study has estimated the current
non-trap emission factor to be about 0.27 g/mi. One
reason for this difference in particulate emission factors
is, as already mentioned, that the previous study assumed
a NOx standard of 1.0 g/mi for LDDVs (and its equivalent
for LDDTs) while this study has assumed a 1.5 g/mi NOx
standard for LDDVs and 2.3 g/mi NOx for LDDTs. The
remainder of the difference (approximately 10 percent) is
due to small differences in overall diesel sales
projections and total light-duty VMT in 1995. It should
be noted that the previous study projected nearly twice
the level of LDDV penetration as this study (20 percent
versus 11.5 percent) , but only 60 percent of the LDDT
penetration (20 percent versus the current 33.9 percent).
Thus, the net effect of the two differences is very small.
-------
2-24
controlled scenario estimate. The results for the worst case
sales scenarios are similar.*
The information presented above is summarized in Table
2-12 (best estimate sales) and Table 2-13 (worst case sales).
The mid-points of the emission ranges contained in the previous
studies are shown in Table 2-12 (and the upper limits shown in
Table 2-13) , because the mid-points represented what was then
EPA's best estimate of diesel penetration and the upper limits
represented what was then EPA's worst case estimate of diesel
penetration.
Both tables are organized in a hierarchical fashion, with
those scenarios yielding the highest urban emission estimates
located near the top and those yielding the lowest estimates
near the bottom. Also shown (in parentheses) are the degrees
of emission reduction from the original uncontrolled emission
estimate compared to that provided by the original controlled
emission estimate.
As can be seen from Table 2-12, the base scenario provides
about the same control as that estimated for essentially the
same controls 3-4 years ago. On the other hand, while
emissions under the relaxed scenario are 60 percent greater
than those under the base scenario, the relaxed scenario still
provides 74 percent of the original reduction projected for the
trap-based particulate standards.
Two alternate scenarios are also shown in Table 2-12. One
is the base scenario with further controls placed only on HDDVs
(i.e., relaxed scenario for LDDV and LDDTs). This scenario
still provides nearly the same control (only 4 percent less)
than that originally projected for the base-scenario
standards. The other is labelled "Intermediate Control
Scenario," and consists of the relaxed scenario for LDDVs and
LDDTs and an intermediate 0.4 g/BHP-hr standard for HDDVs.
(Intermediate standards were not considered between the
relaxed- and base-scenario standards for LDDVs and LDDTs
The difference between the current relaxed-scenario
estimate and the previous uncontrolled estimate is
primarily due to: 1) the current analysis presumes a
decrease in engine-out HDDE emissions from 0.7 g/BHP-hr to
0.6 g/BHP-hr in 1988, and 2) vehicular emissions in the
current study are projected to decrease with future
increases in HDDV fuel economy. The difference between
the current base-scenario estimate and the previous
controlled estimate is due to the more detailed fuel
economy and fuel consumption estimates that are used in
this study.
-------
2-25
Table 2-12
Comparison of Current Urban Emission Estimates
to Those of Previous Studies - Best Estimate Sales
Scenario
Original 1979-80
Analyses (Uncon-
trolled)
Relaxed Scenario
Intermediate Con-
trol Scenario[l]
Base Scenario (HDD
Control Only)
Base Scenario
Original 1979-80
Analyses (Controlled
Total 1995
Urban Emissions
(metric tons per year)
Reduction from Original
Uncontrolled Emission
Estimate Relative to
That Provided By Original
Controlled Estimate
239,000
114,000
92,000
78,000
71,000
71,000
--
74%
88%
96%
100%
100% (base)
[1] Relaxed scenario for LDDVs and LDDTs, 0.4 g/3HP-hr standard for
HDDVs.
-------
2-26
Comparison
to Those of
Table 2-13
of Current Urban Emission Estimates
Previous Studies - Worst-Case Sales
Scenario
Original 1979-80
Analyses (Uncon-
trolled)
Relaxed Scenario
Intermediate Con-
trol Scenario[l]
Base Scenario (HDD
Control Only)
Base Scenario
Original 1979-80
Analyses (Con-
trolled)
Total 1995
Urban Emissions
(metric tons per year)
287,000
162,000
137,000
120,000
106,000
85,000
Reduction from Original
Uncontrolled Emission
Estimate Relative to
That Provided By Original
Controlled Estimate
62%
74%
83%
90%
100% (base)
1] Relaxed scenario for LDDVs and LDDTs, 0.4 g/BHP-hr standard for
HDDVs.
-------
2-27
because the difference between the two sets of standards is
already very small.) This scenario provides 88 percent of the
reduction originally projected for the trap-based standards.
Thus, based on the information contained in Table 2-12, it is
possible to obtain most, if not all, of the control originally
projected with standards less stringent than the trap-based 0.2
g/mi, 0.26 g/mi and 0.25 g/BHP-hr for LDDVs, LDDTs and HDDVs,
respectively.*
As can be seen from Table 2-13 (worst case diesel sales) ,
the order of the various scenarios does not change
significantly. However, none of the current control scenarios
provides as great a reduction in emissions from the original
controlled scenario for worst case sales when compared to those
which occur for the best case sales (Table 2-12) . The base
scenario only provides 89 percent of the originally projected
control and the relaxed scenario provides only 61 percent of
that control. The two alternate scenarios fall in between.
This difference from the results of Table 2-12 is due primarily
to the increased severity of the worst case diesel penetrations
of this study as compared to those of the previous studies.
It should be remembered that the present analysis assumes
NOx standards of 1.5 and 2.3 g/mi for LDDVs and LDDTs,
respectively. The effect of 1.0 and 1.2 g/mi NOx standard
for LDDVs and LDDTs, respectively, which were assumed in
previous analysis, is addressed in Chapter 10.
-------
2-28
References
1. "Characterization of Particulate Emissions from
In-Use Diesel Vehicles," Gibbs, R., et al., SAE Paper No.
801372, October 1980.
2. "A Study of Exhaust Emissions from Twenty High
Mileage Oldsmobile Diesel Passenger Cars," U.S. EPA, OANR, QMS,
ECTD, TEB, March 1980.
3. "Regulatory Analysis of the Light-Duty Diesel
Particulate Regulations for 1982 and Later Model Year
Light-Duty Diesel Vehicles," U.S. EPA, OANR, OMS, ECTD, SDSB,
February 1980.
4. Letter and Supplement from C. L. Gray, U.S. EPA,
ECTD, to T. Young, Engine Manufacturers' Association, July 19,
1983.
5. "Trap-Oxidizer Feasibility Study," U.S. EPA, OANR,
OMS, ECTD, SDSB, March 1982.
6. (Internal Memo from T. Darlington to P. Lorang.
This reference to be constructed at later date.)
7. "Compilation of Air Pollutant Emission Factors:
Highway Mobile Sources," U.S. EPA, OANR, OMS, ECTD, TEB,
EPA-460/3-81-005, March 1981.
8. "The Impact of Light-Duty Diesel Particulate
Standards on the Level of Diesel Penetration in the Light-Duty
Vehicle and Light-Duty Truck Markets," Jack Faucett Associates
for U.S. EPA, EPA Contract No. 68-01-6375, January 17, 1983.
9. "U.S. Long Term Review," Data Resources, Inc.,
Summer 1982.
10. U.S. Federal Highway Administration data as
contained in "MVMA Motor Vehicle Facts and Figures '82," Motor
Vehicle Manufacturers Association of the U.S., Inc., Public
Affairs Division, 1982.
11. "Draft Regulatory Analysis - Heavy-Duty Diesel
Particulate Regulations," U.S. EPA, OANR, OMS, ECTD, SDSB,
December 1980.
-------
CHAPTER 3
AIR QUALITY IMPACT AND POPULATION EXPOSURE
I. Introduction
In an attempt to place the impact of the urban emission
estimates of the previous chapter in a better perspective for
assessing both health and welfare impacts, this chapter
estimates the air quality impact of and population exposure to
diesel particulate emissions in 1995 under the various diesel
sales and control scenarios outlined in Chapter 1. This is
accomplished in four sections.
The first section outlines and uses a methodology for
deriving nationwide average diesel particulate emission factors
for urban areas in 1995. These scenario-specific nationwide
average diesel particulate emission factors become the primary
input to the following three sections.
The second section of this chapter uses atmospheric lead
monitoring data as a surrogate to estimate atmospheric levels
of diesel particulate in 1995 under the various scenarios.
This analysis will provide diesel particulate ambient air
concentrations at one or two particular monitor locations in a
large number of U.S. cities. These 1995 ambient diesel
particulate concentrations are then compared both to each other
and to 1980 levels.
The third section is concerned with a similar analysis of
four types of localized areas which are particularly sensitive
to motor vehicle emissions. These microscale areas include
urban expressways, street canyons and enclosed spaces such as
parking garages and roadway tunnels.
While yielding estimates of diesel particulate
concentrations in particular locations, neither the urban nor
localized air quality analyses address overall population
exposure as people move from location to location within an
urban area. This type of approach is needed in order to
estimate the actual exposure of individuals to these
concentrations as well as to make an assessment of the cancer
risk resulting from exposure to diesel particulate. Therefore,
the fourth section of this chapter will apply a CO exposure
model which was developed by the Office of Air Quality Planning
and Standards (OAQPS) for use in deliberating alternative CO
National Ambient Air Quality Standards (NAAQS). CO
concentrations, like lead, are almost entirely motor vehicle
related and can be used as a surrogate for diesel particulate.
Those sources of CO which are not motor vehicle related, such
as indoor sources, are removed from the model for this analysis.
-------
3-2
It should be remembered that these methodologies utilize
the ambient measurement of other pollutants (lead and CO,
respectively) to estimate the future year (1995) concentrations
of diesel particulate. None of them are based on actual
measurements of urban levels of diesel particulates. As with
any indirect analysis method, the absolute accuracy of this
methodology is not well known because direct measurements of
the pollutant of interest in urban areas cannot be made.
However, these are the best approaches currently available for
nationwide estimations and that are likely to be available in
the near future.
II. Nationwide Diesel Particulate Emission Factors
The first step in estimating either annual average ambient
particulate levels in U.S. cities or the nationwide average
urban population exposures is to derive fleet-wide urban diesel
particulate emission factors for urban areas for each of the
four scenarios. To do this, the procedures outlined in Chapter
2 are repeated to determine the average diesel particulate
emission factor for each vehicle class and scenario as shown in
Table 3-1 (reproduced from Table 2-5 of Chapter 2). The
emission factors for each vehicle category in a particular
scenario are then combined according to the weighting of their
1995 urban VMT, which can be derived from the projected VMT
data in Table 3-1.
Table 3-2 again shows the particulate emission factors for
each vehicle category/scenario, the derived urban VMT breakdown
for 1980 and 1995, and the fleet-wide, urban particulate
emission factors for each scenario. Also shown is a breakdown
of each vehicle class1 contribution to urban emissions under
each scenario.
III. Urban Lead-Based Air Quality Analysis
Since diesel particulate is not easily distinguishable
from other carbonaceous particulate, air quality monitoring
data are not presently available for diesel particulate,
especially under the conditions expected to exist in 1995.
Thus, any method for estimating diesel particulate air quality
impacts must use some measurable surrogate in the ambient air
that is directly relatable to automobile emissions. Various
studies in the past have used such substances as lead or CO to
provide a link between vehicle emissions and air quality. Once
this link is established, then vehicle emissions of the
surrogate substance are related to diesel particulate
emissions, resulting in an estimate of diesel particulate air
quality impacts.
-------
3-3
Table 3-1
Weighted Emission Factors and Projected VMT
LDDV LDDT MDV/LHDV
Weighted Emission Factor
(g/mi)
Calendar Year 1980;
All Scenarios
Calendar Year 1995;
Best Estimate Diesel
Sales
0.0059 0.0074 0.0676
HHDV
1.913
Relaxed Scenario
Base Scenario
Worst-Case Diesel Sales
Relaxed Scenario
Base Scenario
0.0272
0.0205
0.0606
0.0460
0.0760
0.0711
0.1170
0.1092
0.4130
0.2020
0.8088
0.3864
1.6589
0.8499
1.7025
0.8753
Projected Nationwide
VMT (IQJ miles);[2]
1980
1986
1995
Urban Fraction of VMT
(all years)
1,109
1,209
1,537
59.4
232.3
267.2
329.5
48.8
38.2
43.5
55.2
48.8
67.9
84.0
117.9
26.9
-------
3-4
Table 3-2
Derivation of National Average
Diesel Particulate Emission Factors (g/mi)
Urban VMT
Breakdown (%)
1980 1995
1980
Emission Factor (g/mi)
Best Estimate
Sales
Relaxed
Base
Worst Case
Sales
Relaxed
Base
LDV
LDT
MDV/LHDV
HHDV
81.4
14.0
2.3
2.3
80.6
14.2
2.4
2.8
0.0059
0.0074
0.0676
1.9130
0.0272 0.0205
0.0760 0.0711
0.4132 0.2022
1.6589 0.8499
0.0606 0.2460
0.1170 0.1092
0.8088 0.3864
1.7025 0.8753
Fleet-Average 0.0506
Urban-VMT
Weighted
Emission Factor
Vehicle Class Contribution to Urban Emissions %:
LDV
LOT
MDV/LHDV
HHDV
0.0891 0.0554 0.1324 0.0863
9.5%
2.0%
3.1%
85.4%
24.6%
12.1%
11.0%
52.3%
29.9%
18.2%
8.7%
43.2%
36.9%
12.5%
14.5%
36.0%
43.0%
18.0%
10.7%
28.4%
-------
3-5
One methodology, which has been used in the past by GM and
EPA, uses lead as a surrogate for diesel particulate. [1] This
type of analysis uses historical data from urban sites in the
national urban lead monitoring network as an index of mobile
source pollutant levels. An estimate is made of the fleet's
automotive lead emission factor which caused the observed
ambient lead levels, and is compared to the expected diesel
particulate emission factor. Very generally speaking, if
diesel particulate emissions in 1995 are expected to be twice
automobile lead emissions in 1975, for example, then ambient
diesel particulate concentrations in 1995 can be expected to be
twice the 1975 ambient lead concentrations. In this case, 1975
monitoring data was chosen over more recent data to avoid, for
the most part, the errors associated with estimating the
leaded/unleaded vehicle mix.
The basic mathematical expression of this methodology is:
Cm) . E(D)1995 S(D) x ^1995 „ .....
MUJ1995 E(Pb) * S(Pb) * VMT,q_, X C(PD)1975
Where: iy/:3 iy/b
C(D)i995 = ambient concentration of diesel particulate (ug/m3)
£(0)1995 = Fleet-average diesel particulate emission factor
in 1995 (g/mi)
E(Pb)i975 = Fleet-average emission factor for lead in 1975
(g/mi)
S(D) = Dispersion factor for diesel particulate emissions
.S(Pb) = Dispersion factor for lead emissions
VMTX = Total urban vehicle miles travelled in year x
C(Pb)i975 = Urban ambient lead concentrations in 1975 (ug/m3)
Previous work by EPA has resulted in the development of
acceptable estimates for the fleet-average lead emission
factor, E(Pb)]Q75, the diesel particulate dispersion factor
S(D), and the lead dispersion factor S(Pb).[l] Automotive lead
emission factors were calculated for calendar year 1975 based
on the lead content in gasoline (1.9 g/gallon) , light- and
heavy-duty vehicle average fuel economy (13.5 mpg for LDV and
8.7 mpg for HDV) and the urban VMT breakdown by vehicle
class[1] and resulted in 1975 fleet-average lead emission
factor of 0.11 g/mi. The diesel particulate dispersion factor
S(D) was considered to be essentially 1.00 due to their small
size. The lead dispersion factor was determined to be 0.43
-------
3-6
based on measurements showing that 57 percent of the lead
emissions from motor vehicles are larger than 9 microns, which
is the estimated cut-off between: 1) those particules which
disperse, and 2) those which settle out on the ground soon
after emission and do not contribute to ambient lead
concentrations at the fixed site monitors.[1]
The VMT growth estimates of the previous analysis will be
revised here, due to significant changes in these projections
over the past 2-3 years. The nationwide VMT and urban
estimates presented in Table 3-1 show urban VMT growth to be 40
percent between 1980-95. Since the Energy and Environmental
Analysis projections [2] do not go back to 1975, a second
DOE-sponsored study was used to derive the urban VMT growth
between 1975-80. This Oak Ridge National Laboratory study
estimated VMT growth to be 14 percent between 1975-80. [3]
Combining these two figures yields an overall VMT growth
between 1975-95 of 60 percent.
The use of the factors mentioned above results in the
following general equation:
= E(D) g/mi 1.0 x T 6n x rfph>
1995 0.11 grams lead X 0.43 X 1>6° X C(Pb)1975
mile
or
C(D) = 33.82 E(D) X C(Pb)
At this point it is necessary to consider one additional
factor in order to make the lead ambient concentrations fully
compatible with mobile source modeling. This factor is used to
correct the lead air quality data for the fact that only an
estimated 89 percent of the total nationwide lead emissions are
due to mobile sources.[1] Therefore, the ambient lead
concentrations were multiplied by 0.89 to account for the
possible contribution of non-automotive sources. This
adjustment is probably conservative, though to an unknown
degree, because only lead monitors in areas of no known large
stationary source of lead were chosen for this analysis and the
majority of the non-automotive sources of lead emissions reside
in a few identifiable areas which have been excluded from this
analysis.
The final version of the equation, after application of
the above adjustments, used to convert the lead ambient air
monitoring data to an estimate of urban diesel particulate
concentrations is provided below:
-------
3-7
C(D) ug/m3 = 33.82 E(D)1995 X C(Pb)1975 X 0.89
= 30.10 E(D)i995 x C(pb)l975
Since each of the four scenarios in this analysis has a
specific average diesel particulate emission factor (E(D))
associated with it, four discrete conversion factors are
produced relating urban ambient concentrations of lead to
diesel particulate levels. As an example, using the fleet-wide
urban diesel particulate emission factor for best estimate
sales and the relaxed scenario from Table 3-1 results in a
factor of 2.68. This means that 1995 urban diesel particulate
concentrations are projected to be 2.68 times larger than 1975
urban lead levels.
Table 3-3 presents the lead-based estimates of diesel
particulate concentrations for each scenario in for 28 cities
included in the National Air Surveillance Network (NASN) for
lead in 1975. These monitor stations were selected from a
larger lead data base as they were known to be in areas having
no large stationary sources of lead emissions, and to be above
12 meters in height in order to best represent large scale
average urban lead concentrations. Table 3-4 presents the
range of concentrations of diesel particulate for each scenario
as a function of city size.
For the best estimate sales and relaxed particulate
standards scenario, the ambient air diesel particulate
concentrations range from a low of 1.15 ug/m3 for the city of
Kansas City, Kansas to a high of 7.18 ug/m3 in Los Angeles.
The other scenarios show similar ranges with the highest
projected concentration occurring in the worst case sales,
relaxed standards scenario, as expected (10.67 ug/m3) . In
comparing the best estimate sales scenarios it can be seen that
the base scenario will result in an estimated 38 percent
reduction in the 1995 ambient diesel particulate concentrations
compared to the relaxed scenario. This could constitute as
much as a 2.70 ug/m3 reduction (Los Angeles) or as little as
a 0.43 ug/m3 reduction (Kansas City, Kansas) in diesel
particulate levels.
This same methodology can also be applied to 1980 diesel
particulate emissions to show the change in estimated ambient
diesel particulates between 1980-95. Using the 1980 diesel
particulate emission factors from Table 3-1 and a VMT growth
rate of 14 percent (as indicated previously for 1975-80) ,
results in a conversion factor of 1.08 which in turn result in
the diesel particulate urban concentration estimates in Tables
3-3 and 1-4. As can be seen, ambient diesel particulate levels
in 1995 will increase over 1980 levels under all scenarios.
-------
3-8
Table 3-3
Lead Based Ambient Diesel Particulate
Concentrations (uq/m ^
1995
Best Estimate Sales
City
1980
Population Greater Than
Houston
Los Angeles
New York
Philadelphia
2.68
3.43
1.34
1.72
1.58
Population Between 500,
Boston
Denver
Kansas City ,
New Orleans
Phoenix
Pittsburgh
San Diego
St. Louis
1.17
1.22
1.02
2.05
2.69
1.09
1.45
1.51
Population Between 250,
Atlanta
Birmingham
Cincinnati
Jersey City
Louisville
Oklahoma City
Portland
Tucson
Yonkers
1.34
1.56
1.03
1.32
1.23
2.12
1.30
1.03
0.96
1.48
Population Between 100,
Kansas City, KA
Mobile
New Haven
Salt Lake City
Spokane
Trenton
Waterbury
0.77
0.55
1.23
1.47
1.26
0.75
1.13
2.41
Relaxed
1,000,000
5.59
7.18
2.81
3.59
3.29
000 and 1,000,000
2.47
2.55
2.15
4.29
5.63
2.28
3.02
3.16
000 and 500,000
2.81
3.27
2.17
2.76
2.57
4.44
2.74
2.17
2.01
3.10
000 and 250,000
1.61
0.97
2.17
2.60
2.21
1.31
1.99
4.25
Base
3.50
4.48
1.75
2.24
2.05
1.54
1.59
1.34
2.68
3.51
1.42
1.88
1.97
1.75
2.04
1.35
1.72
1.60
2.77
1.71
1.35
1.26
1.94
1.01
0.72
1.60
1.92
1.64
0.97
1.47
3.14
Worst Case Sales
Relaxed
8.32
10.67
4.18
5.33
4.89
3.66
3.78
3.19
6.38
8.37
3.39
4.50
4.69
4.18
4.86
3.22
4.10
3.83
6.61
4.06
3.22
2.99
4.62
2.39
1.71
3.83
4.57
3.90
2.31
3.51
7.49
Base
5.43
6.96
2.73
3.47
3.19
2.38
2.46
2.07
4.15
5.45
2.20
2.93
3.06
2.73
3.16
2.10
2.68
2.49
4.31
2.64
2.10
1.94
3.01
1.55
1.11
2.49
2.99
2.55
1.51
2.29
4.88
-------
3-9
Table 3-4
Average Lead Based Ambient Diesel Particulate
Concentrations by Population (ug/m^)
1995
City Size Grouping Best Estimate Sales Worst Case Sales
(Population) 1980 Relaxed Base Relaxed Base
Greater than 1.27-3.02 2.65-6.33 1.66-3.95 3.95-9.40 2.57-6.13
1,000,000
500,000-1,000,000 0.95-2.10 1.98-4.38 1.23-2.74 2.95-6.53 1.92-4.25
250,000-500,000 1.00-1.67 2.09-3.50 1.30-2.18 3.09-5.20 2.01-3.39
100,000-250,000 0.62-1.78 1.29-3.72 0.81-2.32 1.92-5.53 1.24-3.60
Ranges are average values plus and minus one standard derivation(s).
-------
3-10
For example, between 1980-95, diesel particulate concentrations
in Los Angeles would increase 3.75 ug/m3 under best estimate
sales and the relaxed scenario, versus 1.05 ug/m3 under best
estimate sales and the base scenario.
Another characteristic difference between the present year
(1980) urban ambient diesel particulate projection and the 1995
projections are that the proportion of LDDs versus HDDs and
hence their impact on air quality are substantially different.
LDDs produce only 12 percent of total diesel particulate
emissions in 1980 and about 36 to 61 percent in 1995 depending
on which scenario is chosen. For a city such as Los Angeles,
this translates to an increase in urban ambient LDD particulate
concentrations of 1.74 ug/m3 and 2.23 ug/m3 for base and
relaxed standards with best estimate sales, respectively. Of
course, by analogy the impact of heavy diesel vehicle
categories (MDV/LHDV and HHDV) on urban air quality (1995
versus 1980) is proportionately less than the overall fleet,
though in absolute terms still increasing.
Prior analyses have been performed by EPA on the impact of
diesel particulate on urban air quality. The most pertinent
studies are those done for the regulatory analyses for the
light-duty diesel particulate standards and the HDD particulate
standards.[4,5] In both of these regulatory analyses an
identical lead-based air quality projection was made for diesel
particulate. The only differences between these projections
and the present study would be the diesel particulate emission
factors used and the year for which the projections were made.
The more recent heavy-duty analysis will be used as the primary
comparison to the present study.
The analysis of the urban air quality impact resulting
from diesel particulates, as calculated in this report, .is
approximately 45 percent lower than the previous analysis based
on a comparison between the midpoint of the previous range of
ambient concentrations and the best estimate relaxed scenario
in this analysis. For example, the urban ambient level of
diesel particulate in this study was estimated to be 7.18
ug/m in Los Angeles under the best estimate relaxed scenario
while the corresponding value in the previous analysis for the
uncontrolled fleet was approximately 12.9 ug/m3. The primary
reason for this difference is the fact that non-trap emission
levels for LDDs are well below those projected three years ago
and that the relaxed-scenario standard for HDDs includes a
slight degree (15 percent) of control.
IV. Microscale Air Quality Analysis
Certain very specific localized areas are known to be
affected by motor vehicle emissions to a greater extent than
-------
3-11
urban areas as a whole (and the locations of the lead
monitors). Among these localized areas (hereafter called
microscale areas) are urban expressways, street canyons,
roadway tunnels, parking garages and residential garages. In a
previous effort by EPA designed to evaluate potential hazards
due to unregulated pollutants emitted from motor vehicles, a
set of ambient air dispersion models and model parameters were
developed and validated.[6]
These models, while mathematical in nature, were validated
based on known concentrations of CO in these microscale areas.
As such, these models can be considered accurate for the exact
geographical and meteorological situations being examined.
However, as the relationship between diesel particulate and CO
emission factors may differ under specific conditions, these
models can only be considered to be good estimates when applied
to diesel particulate modeling. However, this approach is the
best assessment available for localized estimates of diesel
particulate concentrations.
This work identified a set of typical and severe
situations for each of these microscale areas, differing by
vehicle traffic volume, windspeed, and other factors
influencing ambient concentrations. The results of the earlier
work allow calculation of the ambient air concentration for any
of these microscale areas (in either the typical or severe
situations) based only on the pollutant emission factor. If a
pollutant is assumed to be evenly distributed within the
microscale and of low short-term reactivity, then the pollutant
emission factor is multiplied by the conversion factor (one for
each microscale area situation) and thereby converted directly
into an ambient concentration at the specific microscale
location.
Table 3-5 presents the various selected microscale areas
and their corresponding conversion factors. These factors
represent the ambient concentrations of any pollutant estimated
to occur in each of these microscale areas for a vehicle
emission factor of 1 g/mi.
The particular values listed in Table 3-5 for the typical
situations were selected to be reasonably representative of the
desired types of areas. The concentrations represented by the
severe situation for each scenario would be expected to occur
only a small percentage (1 percent) of the time on a nationwide
basis. However, in a given specific area, the severe case
could occur much more frequently. For example, the severe
expressway situation used a segment of the Santa Monica freeway
in Los Angeles, which is a 10-lane freeway with a 200,000
vehicle/day traffic count. The windspeed and direction were
-------
3-12
Table 3-5
Summary of Microscale Situation Concentrations
Microscale
Conversion Factor
Situation (ug/m^ per g/mi)
1. Roadway Tunnel
Typical - Lowry Hill, Minnesota 1,123
Severe - Baltimore Harbor Tunnel 2,856
2. Street Canyon (sidewalk receptor)
Typical - 4 lane canyon, 800 vehicles/hr., 42
8 raph windspeed
Severe - 6 lane canyon, 2400 vehicles/hr., 282
2 mph windspeed
3. On Expressway (Wind: 315 deg. relative,
2.2 mph)
Typical - San Antonio 1-410, 124
Severe - Los Angeles 1-10, 506
4. Beside Expressway
100 meters away-downwind 105
-------
3-13
typical of this location. While this kind of traffic flow is
severe for most urban expressways in the nation (impossible for
most), it is a definite regular occurrence for this expressway
and other busy large expressways in large metropolitan areas.
Thus, while the severe situation would not be expected to occur
frequently on urban expressways in general, there is a real
possibility of frequent occurrence in the few very busy freeway
segments in large cities.
Table 3-6 presents the results of the microscale area
calculations for the four scenarios. The range of localized
diesel particulate concentrations in Table 3-6 constitute an
estimate of the levels which might be expected in these areas
in 1980 and 1995. These levels are not to be construed as
anything like average urban levels or average personal
exposures. In fact, the overall population exposure
contributed by these very high, short-term levels is probably
relatively small. However, to the extent that the population
is exposed as they pass through these microscale areas in their
day to day activities, these localized area diesel particulate
concentrations could constitute an impact on their health or
welfare.
For example, high localized concentrations of diesel
particulate on an expressway (30-70 ug/m^ in 1995) may be
reflected in reduced short-term visibility or increased
short-term odor which may impact on the health and welfare of
the commuting public. However, current levels of diesel
particulate in an identical situation could already be
approximately 25 ug/m^.
An examination of Table 3-6 shows the wide variety in
potential localized area concentrations of diesel particulate.
These projected levels range from as low as 2 ug/m-* for a
typical street canyon under the best estimate sales and base
standards scenario to as high as 378.0 ug/m^ for a severe
roadway tunnel under the worst case sales and relaxed standards
scenario. The lowest levels of this range roughly correspond
to the overall urban area concentrations presented in Table
3-3. This finding is consistent with the fact that some of the
fixed site monitors used in the lead ambient monitor network
are sited in locations such as street canyons (on top of tall
buildings) or near expressways and concentrations in these
localized areas, under typical conditions, may approach the
overall urban area averages.
An analysis of the overall differences between the relaxed
and base scenarios yields the same percentage differences as
those found for urban emissions in general in Chapter 2. These
differences can be translated to an increase in localized diesel
-------
3-14
Table 3-6
Microscale Diesel Particulate Concentrations (ug/m
1995
I.
II.
III.
Roadway
Tunnel
Typical
Severe
Street
Canyon
Typical
Severe
1980
57
145
2
14
Best Estimate
Relaxed
100
254
4
25
Sales
Base
62
158
2
16
Worst Case
Relaxed
149
378
6
37
Sales
Base
97
246
4
24
On Expressway
Typical
Severe
6
26
11
45
7
28
16
67
11
44
IV. Beside
Expressway 59 6 14
-------
3-15
particulate concentrations of from as low as 2 ug/m^ for a
typical street canyon situation to as high as 96 ug/m^ for a
severe roadway tunnel situation. Comparing the increases in
these projected 1995 localized diesel particulate
concentrations to the concentrations which may be occurring now
(1980) results in the observation that, for the severe roadway
tunnel situation, present levels of diesel particulate may be
expected to be on the order of 145 ug/m^, which can be
compared to the projection for the best estimate sales relaxed
control scenario in 1995 of 254 ug/m^, or the projection for
the best estimate sales base control scenario of 158 ug/m3.
V. Population Exposure Analysis
A. Introduction
The population exposure estimates used in this report are
based on a general air pollutant exposure model, called the
NAAQS Exposure Model (NEM), developed by OAQPS for the
evaluation of relative population exposures under alternative
NAAQS [7]. The NEM is an activity pattern model that simulates
a set of population groups called cohorts as they go about
their day-to-day activities. Each of these cohorts are
assigned to a specific location type during each hour of the
day. Each of several specific location types in the urban area
are assigned a particular air quality value based on fixed site
monitor data. The model computes the hourly exposures for each
cohort and then sums these values over the desired averaging
time to arrive at average population exposures and exposure
distributions. Thus, the model simulates pollutant
concentrations in urban areas by relating pollutant
concentrations in urban locations to fixed-site monitor levels
and simulates the activities of people by relating the
population to a fixed set of cohorts with defined activity
patterns.
For example, a certain fraction of the total urban
population might be assigned to an office worker cohort with a
home-work-home activity pattern. This cohort would experience
a consistent set of microenvironments, such as, home,
transportation, and office in a normal days activity. Each of
these microenvironments would have an associated pollutant
concentration related to the fixed-site monitor level for the
specific time of day and date. The fixed-site monitor levels
are adjusted to correct for the differences that typically
exist between the monitored locations and the microenvironment
location. These adjustments are general enough to account for
multiplicative and additive types of correlations between the
monitors and locations.
-------
3-16
A unique feature of the model is that it separates
concentrations, people, places, and time (all of the important
ingredients of exposure) into discrete elements. The
concentrations are broken up into values determined by the
precision of the fixed site monitors (e.g., for CO into whole
integers of ppm). The people or urban population of an area
are separated into cohorts that are assumed to have specific
activity patterns and hence exposures. The places are
separated into a discrete number of areas which are assumed to
have identical pollutant concentrations over a given period of
time. Thus, the definition of places may be influenced by the
type of pollutant studied and its emission sources. Time is
separated into the smallest unit of measure which is desired.
Since this methodology was designed to be used with the NAAQS,
which are based, at a minimum, on a 1-hour time period, one
hour is the shortest time period considered by the general
model. Longer averaging times, such as 24 hours or a year, can
be obtained by calculating the appropriate averages from the
1-hour exposures.
The general NEM approach has been used by OAQPS to develop
specific models for a number of criteria pollutants, such as
CO, S02, N02 and particulate.[8] These pollutants have
been studied by applying the specific pollutant data base from
the appropriate EPA monitoring program and by designing the
place designations to those most appropriate to the pollutant.
Place or location designations in the NEM are determined as
exposure districts or exposure neighborhoods, depending on
whether the pollutant of interest is a point or dispersed
source of emissions, respectively. The exposure district
approach is more geographical in nature and relies on the fact
that pollutants which are primarily emitted by large point
sources can be adequately characterized by exposure districts
of fairly large areas. In contrast, pollutants with emission
sources which are dispersed thoroughout an urban area
(including mobile sources) are best characterized by
considering exposure neighborhoods with common exposure
patterns. These neighborhoods may be spread out in a random
fashion throughout the urban area.
The general NEM modeling approach using monitoring data in
four cities (Chicago, Los Angeles, Philadelphia, and St. Louis)
was applied to the criteria pollutants mentioned above
resulting in four average exposures and exposure
distributions. A limited set of 24 total monitors in four
cities was used because of the extensive computer time required
to run the NEM. A rough nationwide exposure extrapolation has
been developed by OAQPS, which involves relating each of the
large urban areas in the country to the most similar of the
four modeled NEM cities.[8]
-------
3-17
Direct measurements of diesel particulate levels in
overall urban areas have not been made because of the
difficulty in distinguishing diesel particulate from other
carbonaceous particulate. The use of a surrogate approach to
relate diesel particulate to some other pollutant which can be
readily measured in urban areas appears to afford the best
chance of obtaining reasonable estimates of diesel particulates
as was done for urban air quality estimates in the previous
section of this report. Of the criteria pollutants which have
been assessed by OAQPS using the NEM methodology, CO appears to
have the most desirable characteristics as a surrogate for
diesel particulate or any other mobile source pollutant. CO is
the only criteria pollutant which is chiefly emitted by mobile
sources. The dispersion characteristics of CO are also very
similar to those of diesel particulate, since diesel
particulates are very fine and disperse essentially like a gas.
The CO NEM is designed to provide an overall estimate of
CO population exposure related to different values of the CO
NAAQS. While the selection of CO and the NEM methodology is
the best basis for a mobile source assessment of population
exposure, a number of modifications to the NEM CO assessment
are necessary and/or desirable in order to provide the best
estimate of diesel particulate exposure.
One modification to the standard NEM methodology involves
the removal of all indoor sources of CO, such as gas stoves,
from the air quality inventories. This change allows for a
more precise estimate of automotive sources, particularly
indoors, where the contribution of outdoor emissions is still
present without the confounding presence of indoor sources.
This is the most important modificaton to the model in order to
allow a reasonable estimate of automotive exposure to CO and,
via an appropriate conversion, to diesel particulate.
Other desirable modifications to the NEM, which are
planned for the near future but which are not available at the
present time for this report, include an effort to correct the
model for a suspected underestimation of mobile source
microscale area contributions, and an effort to design a
national extrapolation procedure expressly for mobile sources.
The current version of the NEM methodology is expected to be
slightly low relative to the likely changes which will result
from the ongoing work but is certainly adequate in its present
form for the purposes of this document.
The NEM based exposure estimation methodology used in this
report provides both an average CO exposure and CO exposure
distribution for the four cities in the data base. The average
CO exposure results are used to develop the nationwide exposure
-------
3-18
estimates for diesel particulate in 1995. The exposure
distribution form of the methodology is not essential for the
uses of this report and will not be presented here. However,
for the sake of completeness and because the distributions do
present information on exposure ranges of diesel particulate
which may be interesting in placing the exposures in
perspective, the exposure distributions for diesel particulate
will be presented later in this chapter.
B. Past Exposure Efforts
Before discussing the details of the diesel particulate
exposure estimate derived in this report, it may be useful at
this point to compare the NEM methodology to the general
methodologies used in previous EPA assessments of mobile source
pollutant exposures. Two different assessments have been used
in the past: 1) one based on a methodology by Pedco for a
previous EPA diesel particulate risk assessment and, 2) one
based on a methodology by SRI for an EPA benzene risk
assessment.
The Pedco exposure assessment used an Air Quality Display
Modeling (AQDM) approach wherein the urban area to be modelled
was broken up into a set of geographical grids where the grid
population and grid pollutant concentrations were combined into
an exposure for each grid. [9] The Pedco approach used TSP to
derive the original predicted concentrations of particulate and
adjusted these predictions based on TSP monitor levels.
However, since the emission pattern and ambient distribution of
TSP may be very different from diesel particulate due to the
large contribution of non-mobile sources to TSP emissions, this
is and was thought to be a source of possible error in the
Pedco assessment. No effort was made to simulate different
activity patterns such as is done by the NEM model. The Pedco
approach was applied to only one city, Kansas City, and this
single result was extrapolated nationwide. The Pedco approach
was very valuable at the time of its development as a coarse,
but usable first estimate of the population exposure to diesel
particulate, but the present NEM model is a more precise
approach and yields a more accurate result. It is not possible
to directly compare the results of these two approaches (NEM
and Pedco) because of the different emission factors used.
However, it is estimated that the current NEM assessment
results in exposures which are roughly a factor of 10 higher
than the Pedco assessment. At the time of the preparation of
the Pedco report, and its subsequent use in the EPA preliminary
risk assessment for diesels, it was thought that the Pedco
assessment might be low, primarily because Kansas City was not
thought to be the most typical urban area with respect to
automobile emissions. Thus, while this factor is significant,
-------
3-19
it is not unexpected or unusual in our view, but rather
indicates that the current NEM approach represents a more
correct and precise exposure assessment for mobile sources.
The SRI modeling approach used for the EPA benzene risk
assessment estimated the mobile source contribution using an
area wide dispersion model called the Hanna-Gifford dispersion
model.[10] This approach is very simplistic, requiring only an
estimate of an urban area's vehicle registrations, VMT, area
size, average annual wind speed, and vehicle emissions. A
relatively limited examination of population activity patterns
was used by SRI to estimate the influence of the many different
sources of benzene, but only the area-wide dispersion-based
averages were used for the automobile contributions.
Comparison of the exposures calculated by the SRI benzene
assessment to the NEM exposures results in the finding that the
NEM based exposures are roughly a factor of five higher than
the SRI estimates for a comparable emission factor. The SRI
report states that the automobile contribution to the benzene
assessment are probably underestimated because of the fact that
the area-wide model used may not adequately reflect the high
localized concentrations believed to occur around
automobile-use areas. [10] The relatively close agreement
between the SRI and NEM exposure assessments and the intuitive
logic discussed earlier on why the NEM should be higher leads
to the conclusion the NEM model used in this report is the most
valid approach currently available.
C. Average Nationwide Diesel Particulate Exposures
Table 3-7 presents the NEM based average CO concentrations
for the four cities used in NEM program. The average CO
exposure concentrations for each city in Table 3-7 are combined
in order to provide the desired nationwide average exposure. A
simple method to use and the one used by the CO NEM and in this
report groups each of the large urban areas in the country
(populations greater than 200,000) with one of the four modeled
cities according to overall urban characteristics including
populations, vehicle use patterns, etc.[8] Under this
nationwide extrapolation a large portion of the population (43
percent) is grouped under Chicago. While this nationwide
extrapolation is reasonable and valid as an estimate, it is the
least precise part of this assessment. Thus, it is one of the
areas that ECTD is continuing to investigate as part of the
ongoing mobile source exposure estimate project. ECTD intends
to use a large bank of CO monitor data, perhaps from the EPA
SAROAD data base, selected with a view toward mobile source
contributions, to provide an extrapolation to the nationwide
situation.
-------
3-20
Table 3-7
Average Total CO Exposure in Four Cities
City
Chicago
Los Angeles
Philadelphia
St. Louis
Overall
CO ppm
(Annual avg.)
1.8 ppm
3.0
1.3
2.0
2.12
Population
of City As
Used in CO NEM
2,363,014
7,716,895
2,933,790
1,221,461
Associated Total
Urban Population
In Cities 200,000
1970
38,894,395
26,339,249
10,553,523
17,350,712
93,137,849
-------
3-21
The national population in Table 3-7 as mentioned
previously (Column 4) counts only persons in urban areas with
populations greater than 200,000. While it is desirable to
extend this analysis to the population in all urban areas
including those with populations less than 200,000, the
likelihood that the exposures in smaller urban areas would be
lower than any of the four NEM cities prevents this from being
precisely done. Therefore, we have limited our analysis to the
populations in the large urban areas without considering the
exposures of rural or small urban areas.
The aforementioned nationwide extrapolation to the NEM-CO
average output results in the calculation of an overall average
nationwide concentration (based on CO) of approximately 2.1
ppm. This total adjusted national average (2.1 ppm) can then
be manipulated into a diesel particulate national average by
ratioing CO and diesel particulate emission factors and
multiplying the result by 2.12 ppm. The national average CO
emission factor for 1978 (the same year as the CO NEM data
base) is estimated to be 62.3 g/mi.[ll] However, since future
year VMT is expected to increase by about 45 percent, the
diesel particulate emission factor should be adjusted upward by
a factor of 1.45 for 1995. [2] The 1975 diesel particulate
emission factors are the same as those used in the air quality
analyses (see Table 3-2).
Table 3-8 presents national average diesel particulate
concentrations for each of the four main scenarios. These
values will be used in Chapter 5 to estimate the diesel
particulate cancer risk.
The total population exposure (from Table 3-8) for the
best estimate sales and relaxed standards scenario is estimated
to be 61 percent higher than the exposures calculated for the
corresponding base case standards. However, separating the
light- and heavy-duty components of these exposures, individual
contributions to this increase in exposure with relaxed
standards are 14 percent for LDVs (LDVs and LDTs) and 86
percent for HDVs (HDV/LHDV and HHDV) . These data can be
interpreted as meaning that the bulk of the increase in
exposure with best estimate sales and relaxed standards can be
attributed to HDVs with a comparatively small contribution from
LDVs.
If the worst case sales projections are used to derive
relationships between the relaxed and base scenarios above,
then the overall population exposure is increased 54 percent
with LDDs contributing 28 percent of the increase, and HDVs 27
percent.
-------
3-22
Table 3-8
Total National Diesel Particulate Exposure in 1995
Annual Average Diesel Particulate
Exposure (ug/m^)
Best Estimate Sales Worst Case Sales
Relaxed Base Relaxed Base
LDV
LOT
MDV/LHDV
HHDV
Total 4.98 3.09 7.40 4.82
-------
3-23
D. Exposure Distribution for Diesel Particulate
In addition to the national average exposure derived in
the previous section, this mobile source model can be used to
identify a distribution of exposures among discrete
concentration ranges. A manipulation of this information in a
manner analagous to the previous discussed average exposure can
be used to provide a national average diesel particulate
exposure distribution. For convenience this distribution is
presented in Table 3-9 as a range of percentages for the two
cities with the lowest and highest exposures versus the diesel
particulate concentration range (dependent on the scenario).
The exposure distributions included in Table 3-9 can be
used as a relative illustration of how the total exposure is
broken down into concentration ranges. While this data is not
used further in this analysis, in the event that a non-linear
risk model is used in the future to estimate diesel cancer
risk, data such as those in Table 3-9 will be necessary for
estimating the cancer risk to individuals.
A brief inspection of the data in Table 3-9 show that
while there are distinct differences between each city's
exposure distribution, there is the common feature wherein most
of the diesel particulate exposures (95-99 percent) are in the
lowest range. The corresponding ranges of diesel particulate
exposure are from 0-10 to 0-21 ug/m^ depending on scenario.
If future interest is generated on this kind of exposure index,
a way to further break out the exposures within this lowest
range will be necessary and this effort is underway as part of
the ongoing ECTD project on developing a mobile source exposure
assessment methodology.
-------
3-24
Table 3-9
Diesel Particulate Exposure Distribution
Diesel Particulate
Best Estimate Sales
Relaxed
117-
106-117
94-106
82-94
70-82
59-70
47-59
35-47
28-35
21-28
16-21
0-16
Base
73-
66-73
58-66
51-58
44-51
36-44
29-36
22-29
17-22
13-17
10-13
0-10
Worst Case Sales
Relaxed
174-
174-174
139-157
122-139
105-122
87-105
70-87
52-70
42-52
31-42
21-31
0-21
Base
114-
102-114
91-102
79-91
68-79
57-68
45-57
34-45
27-34
20-27
16-20
0-16
Range of
Population Exposed %
High
Los Angeles
0.000654
0.000345
0.000941
0.006804
0.008965
0.036916
0.066520
0.185011
0.552863
1.292952
2.825991
95.022026
Low
Philadelphia
0.000250
0.000000
0.000263
0.001121
0.000555
0.007867
0.009422
0.020867
0.084505
0.194798
0.414451
99.265896
-------
3-25
References
1. "An Investigation of Future Ambient Diesel
Particulate Levels Occuring in Large Scale Urban Areas,"
Reiser, D., EPA-AA-SDSB-79-30, November 1979.
2. "The Highway Fuel Consumption Model: Eighth
Quarterly Report,: Energy and Environmental Analysis, Inc. for
U.S. DOE, DOE Contract No. DE-AC01-79PE-70032, July 1982.
3. "Transportation Energy Conservation Data Book", Kulp
G. et.al., ORNL Publication 5765-Edition 6, 1982.
4. "Regulatory Analysis of the Light-Duty Diesel
Particulate Regulations for 1982 and Later Model Year
Light-Duty Diesel Vehicles," U.S. EPA, OANR, OMSAPC, ECTD,
SDSB, February 1980.
5. "Draft Regulatory Analysis - Heavy-Duty Diesel
Particulate Regulations," U.S. EPA, OANR, OMSAPC, ECTD, SDSB,
December 1980.
6. "Estimating Mobile Source Pollutants in Microscale
Exposure Situations," Ingalls, M., EPA-460/3-80-021, July 1981.
7. "A General Model for Estimating Exposure Associated
with Alternative NAAQS," Biller, W., et al., June 1981.
8. "The NAAQS Exposure Model (NEM) Applied to Carbon
Monoxide," Johnson, T. , et al., Draft EPA-OAQPS Report,
December 1982.
9. "Air Quality Assessment of Particulate Emissions from
Diesel-Powered Vehicles," Pedco Environmental, Inc., March 1978.
10. "Assessment of Human Exposure to Atmospheric
Benzene," Marq, S., and S. Lee, EPA Report No. 450/3-78031.
11. "MOBILE2.5 Emission Factor Program" Unpublished
numbers, U.S. EPA, OANR, QMS, ECTD.
-------
CHAPTER 4
VISIBILITY ASSESSMENT
I. Introduction
The most obvious effect of diesel particulate, especially
in urban areas, is reduced visibility. in order to study this
effect, there must be a means of measuring the relationship
between diesel particulate and visibility levels. A method is
needed to determine the visibility impact from a specific level
of diesel particulate.
This chapter develops and applies a method for measuring
the change in visibility caused by an increase in diesel
particulate concentration. This is done on a city-by-city
basis, yielding visibility levels for four regulatory scenarios.
II. Modeling Visibility
There is no absolutely preferred method for modeling
visibility; different measuring techniques are appropriate for
various times and locations. The three types of
visibility-related indices are: 1) direct measures of human
perception, 2) measures of light intensities, and 3) measures
of visual properties of air. Using observers to measure
airport visual ranges is an example of direct human perception
measurement. This is a subjective method which is difficult to
convert to objective physical parameters. There exists no
correlation between the methods of measuring direct human
perception and diesel particulate. In measuring light
intensities, the relationship between perceived contrast and
measured physical contrast is also a subjective and complex
one. (Contrast, combined color and brightness scales, and
blue-red luminous ratios are examples of measures of
intensities.) Because both of the above methods appear to be
inadequate for relating the effect of particulate on
visibility, the third method is, by necessity, the method of
choice. The measuring of visual properties of air and airborne
particles can directly relate the particulate matter from
diesels to a reduction in visibility.
Visibility is defined as the greatest distance it is
possible to see a prominent dark object against the sky at the
horizon.[2] Middleton's Lawfl] relates contrast and light
intensity; both are reduced equally at horizontal views of
objects against the horizon. Koschmieder's Law[l] goes a step
further and relates visual range to the extinction
coefficient. The typical observer can detect an object with 2
percent contrast against the background.[1] The mathematical
formula describing Koschmieder's Law is:
-------
4-2
L =
v
3.91
bext
where Lv is the visual range, 3.91 is ln(.02) and bext is
the total extinction coefficient.
Koschmieder1s Law can be derived from the Beer-Lambert
Law, which describes the more fundamental effect of the
extinction coefficient (bext) on light intensity. The
Beer-Lambert Law is:
I = I0e -bextL
Where:
I0 = light intensity at the object being observed, and
I = light intensity at a distance L from the object.
Described very simplistically, Koschmieder1s Law simply
states that objects become invisible when the ratio of I to
I0 becomes 0.02. Substituting 0.02 for I/IO into the
Beer-Lambert Law and solving for L yields Koschmieder's Law.
As can be seen, the most important parameter in all of
these laws is the extinction coefficient (bext) • This
coefficient is the sum of four components:
1,
2,
3.
4.
scattering by gas molecules, b^g;
absorption by gas molecules, baq;
scattering by particles, bsp;
absorption by particles, bap.
Diesel particulate impacts directly on the latter two
processes. In order to gain insight into the relative role of
diesel particulate in light attenuation, each of the four
components of the extinction coefficient should be examined.[2]
A.
The
molecules
Rayleiah
values of
Gas Scatter
extinction coefficient . due to scattering by gas.
in the free atmosphere at sea level, known as
scattering, is roughly 1.5 x 10~5 meters"1;'
the extinction coefficient within a few percent of
this have actually been measured.[1] If light degradation were
due solely to gas molecula scattering, then the visibility
would be approximately 260 kilometers by the Koschmieder
-------
4-3
formula. Thus, scattering by gas molecules does not play a
major role in observed visibility degradation.
B. Gas Absorption
Nitrogen dioxide (NC>2) is the only absorbing gaseous
specie present in high enough concentrations to have a
significant effect on light absorption. In optics, N02 seems
to be important only in plumes, not in the case of a well-mixed
layer.[1] Therefore, absorption by gas molecules can be
discounted in calculating the total extinction coefficient.
C. Particle Scatter
Particles with diameters in the range of 0.1 to 1.0
micrometer scatter light with the greatest effectiveness.
Diesel particulate falls into this range. Figure 4-1 shows the
ratio of mass to scatter coefficient as a function of particle
radius. The duration of this scattering effect is prolonged
for this size range, since such aerosols generally do not
settle out by gravity and are not removed efficiently from the
atmosphere except by incorporation into clouds and subsequent
rainout. Studies show that they may persist in the atmosphere
for several days.[3]
D. Particle Absorption
The most important contributor to particle absorption is
graphite carbon; any sub-micron particles with a high carbon
content will have a significant effect on visibility. Diesel
particulate, with its 65-80 percent carbon content, falls into
this category.[4]
III. Visibility Equations
Once all the factors involved are known, the computation
of visibility levels due to a change in diesel particulate
level is straightforward. According to the Koschmieder
formula, the visual range is inversely proportional to the
total extinction coefficient. The total extinction coefficient
(bex|-) is the sum of the extinction coefficient for the base
line visibility (bo) and the extinction coefficients due to
absorption and scattering of diesel particulate.
Thus:
bext = bo +
-------
4-4
Figure 4-1 [1]
10*-=
i(H
r-1000
LO-7
o.i-=
riccoo
\
-------
4-5
particulate, b = AMC. The proportionality constant is called
the extinction efficiency, referred to as A. There are several
values, all in a close range, for the extinction efficiency for
diesel particulate; these are listed in Table 4-1. The average
value, A = 8 m^/g, is used for the concentration of diesel
particulate. (Taking into account the carbon content of diesel
particulate (approximately 70 percent), the extinction
efficiency for fine elemental carbon is 11.5 m2/g.)
Therefore, the portion of the extinction coefficient due to
diesels is the product of the increase in particulate
concentration and the extinction efficiency of diesel
particulate, and the equation for bext becomes:
bext = bo + 8[m2/g]Mc (1)
In order to compute percentage changes in visibility, the
baseline visibility extinction, bo, must be known. Baseline
visibilities were obtained from several Trijonis
reports.[5,6,7,8] Trijonis determined the existing
visibilities from cumulative frequency distributions of
quality-checked* airport observations. Figures 4-2 and 4-3
show the distribution of median visibilities for various parts
of the country. Visibility in the Northeast tends to be rather
low, with the relative humidity acting as the dominating
factor. Median visibilities range from 8-12 miles with very
small differences in metropolitan areas, urban/suburban areas,
and nonurban areas. For the Southwest, the median visibility
is 30-55 miles in large urban centers and 65-80 miles at
suburban/non-urban locations. Table A-l of the Appendix lists
median, tenth percentile and ninetieth percentile visibility at
12 northeastern locations. Table A-2 of the Appendix lists
median visibility levels at 94 urban/suburban locations
throughout the U.S. No data exists for the annual median
visibility levels of all the major U.S. cities, so estimates
must be made from the median visibilities listed above. Such
estimates are shown in Tables A-3 and A-4 of the Appendix for
major U.S. metropolitan areas and cities, respectively.
The baseline visibilities (Lvo) are related to
extinction according to the formula:
b = 3'°
0 Lvo
Telephone surveys were conducted at the airports to ensure
that each location had an adequate set of visibility
markers for estimating visual range, reliable reporting
practices, observation personnel and observation locations.
-------
4-6
Table 4-1
Extinction Efficiency (A)
for Diesel Particulate
Study A (m2/g)
Trijonis* (1982) 8.4
Klimisch (1982) 8
Roesslor and 6.8
Faxog (1978)
Vuk, et al (1976)
Pierson (1978)
The figure for Trijonis is calculated from his extinction
efficiency for fine elemental carbon, 12 +3 m^/g, using
a 70 percent carbon content in the diesel particulate.
-------
Figure 4-2
P: Based on photographic
photometry data
11; Based on neplielometry data
*; Based on uncertain extrapolation of
visibility frequency distribution
>£>
I
Figure 2 Median yearly visibilities and visibility Isopleths for suburban/nonurban areas,
-------
4-8
Figure 4-3
70
12
Figure 3 Median annual 1 PM visibilities (in miles) and
visibility isopleths for California, 1974-1976. [6]
-------
4-9
The proportionality constant of 3.0 in this equation is
applicable when using airport visibility data, as opposed to
using the 3.91 figure from the Koschmieder formula. [6] Airport
data does not adhere to the conditions for applying the
Koschmieder formula because natural objects at a great distance
are usually small (small objects need a contrast greater than 2
percent to be seen) , and natural objects are never black. The
proportionality constant of 3.0 is appropriate according to
Trijonis, et al. [6]
To simplify this formula's units, it may also be expressed
as:
. _ 18.6 x 10"4 [miles/m]
o ~ Lvo[miles] U)
where the units of extinction are inverse meters and the units
of visibility are miles. Values of bo for a number of U.S.
cities are shown in Table A-2 in the Appendix.
The new visual range caused by the addition or subtraction
of diesel particulate can now be calculated from the following
expression:
L _ 18.6 x 10"4 [miles/m] _ 18.6 x 10"4 [miles/m] (3)
v b . b + 8[m2/g]M
ext o '^ c
Where:
is determined using Equations (1) and (2) .
However, the above equation assumes that bext i-s
constant throughout the entire visual range. This is a
satisfactory assumption for the baseline situation (i.e., to
estimate bo) . However, it may not be satisfactory to assume
that the effect of diesel particulate will be constant
throughout the visual range. The ambient concentration of
diesei particulate will be relatively high in the central city
and near suburban traffic corridors and will be relatively low
outside of the city or metropolitan limits. Since visibility
may extend to areas beyond the city or metropolitan limits,
this effect must be taken into account. For those cases.. where
visibility extends beyond the affected area, this effect may be
taken into account by returning to the Beer-Lambert Law and
rederiving Koschmieder ' s Law assuming one value of bext for a
fixed distance (i.e., up to some limit) and another value of
-------
4-10
IV. Revised Visibility Equations
As described earlier, the Beer-Lambert Law describes the
reduction in light intensity as a function of distance and the
extinction coefficient of the media. For an object outside of
the affected area being viewed from within the affected area:
-b L
Where ba is the extinction coefficient existing within the
affected area, La is the distance from the observer to the
limit of the affected area and la is the light intensity of
the object at the limit of the affected area. Ia is
described by the equation:
-b (L - L )
I = I e ° a
a o ,
where L is the total distance between the object and the
observer.
Combining the two equations yields:
(-b, L - b (L - L ))
I = IQe a a ° a
Applying Trijonis1 application of Koschmieder ' s Law, ln(I/Io)
is -3.0, and solving for L (now Lv) results in:
3.0 - (ba - b0)La
" ~
where L and b are in inverse units, or
18.6 X 10~4 [miles/m] - (b, - b ) .L ,,x
L = _ a _ o _ a ( ** /
bo
where Lv and La are in miles and bo and bm are in
inverse meters.
The term b^ can be derived using Equation (1) .
Substituting this into Equation (4) yields:
T 18.6 X 10~4 [miles/m] - 8 [m2/g] ML
Jjv — _ c a
bo
-------
4-11
The terms bo and Mc can be derived from Equation (2)
and air quality models, respectively. Only La remains to be
described.
La is the typical distance between the viewer and the
limit of the affected area. Before determining this distance,
the limits of the affected area must be defined. In the actual
situation, the concentration of diesel particulate gradually
falls off until it reaches zero; in the model being used, a
constant level of diesel particulate inside the affected area
and no affect outside is assumed. The limit of the affected
area, La, must be between the point where the actual ambient
concentration of diesel particulate begins to fall off and the
point where it finally reaches zero. Therefore, the affected
area's limit, La, is where the actual diesel particulate
level is approximately half of its central city level. Two
convenient limits which could suffice are: 1) the city limit,
and 2) the metropolitan area limit. These limits (or their
nominal radii) are shown for a large number of metropolitan
areas and cities in Tables A-3 and A-4 of the Appendix,
respectively. It has been assumed that the metropolitan areas
and cities were circular to calculate a nominal radius.
The metropolitan area limits for large cities, such as Los
Angeles (36 miles) and New York City (21 miles) , appear very
reasonable as diesel particulate penetration limits (i.e., for
Los Angeles, La = 36 miles and for New York City, La = 21
miles). However, for smaller cities, such as Ann Arbor,
Michigan (15 miles) and Madison, Wisconsin (20 miles), the
metropolitan area limit appears much too large. The city
limits appear much more reasonable for these smaller cities
(i.e., 2.6 miles for Ann Arbor and 4 miles for Madison). Thus,
metropolitan area limits will be used for the largest U.S.
cities and city limits will be used for the smaller cities.
This will more closely model the size of the affected areas,
yielding a better model for the extent of the actual diesel
particulate concentration level. To be conservative, the
demarcation between the two will be made at a relatively large
city population, 1,000,000, resulting in the use of
metropolitan area limits in only six cases: Chicago, Detroit,
Houston, Los Angeles, New York, and Philadelphia.
Now that the limits of the affected area are established,
the typical distance between the viewer and the limit of the
affected area must be determined. This depends on both where
the viewer is located and on which direction he is viewing.
While it is conceivable that a model could be formulated to
determine the mean viewing distance based on relative
population density and shape of the affected area, etc., the
radius of the metropolitan area or city should be a sufficient
estimate of the typical distance between a viewer and the limit
of the metropolitan area. While most metropolitan areas are
-------
4-12
not circular, a reasonable approximation to the average
distances between the center and the edge can be derived from a
calculation of a nominal radius from the actual area of the
metropolitan area, assuming it is circular in shape. Such
nominal radii for the largest metropolitan areas and cities in
the U.S. are listed in Tables A-3 and A-4 of the Appendix.
V. Visibility Levels
A. Methodology
Measuring the change in visibility levels due to a change
in diesel particulate is dependent on four factors:
1. the mass concentration of the diesel particulate,
2. the extent of this concentration,
3. the extinction efficiency of diesel particulate, and
4. baseline visibilities.
For those cases where the visual range does not extend
past the limits of the affected area, the visual range can be
calculated from the following expressions:
_ _ 18.6 x 10"4 [miles/m] (6)
= ~
b . = ' + 8[m2/g]* M
ext bvQ c
Where Lv is the visual range in miles, Lvo is the baseline
visual range in miles, and Mc is the mass concentration of
the diesel particulates in grams per cubic meter.
For those cases where the visual range does extend beyond
the affected area, the visual range can be calculated from the
following expressions:
T _ 18.6 x 10~4 [miles/m] - ,8[m2/g]* ML ....
ij.. — _ c a \ i)
bo
Where Mc is the mass concentration of diesel particulate in
the affected area, La is the nominal radius of the affected
area, and
If the presence of diesel particulate is determined in
terms of the elemental carbon concentration, then 11.5
should be used instead of 8 m^/g.
-------
4-13
b = 18.6 x 10~4[miles/m]
o L
vo
where Lvo is the baseline visual range without the effect of
diesel particulate.
B. Results
As described in the previous section, three pieces of
information are needed for each city in order to project the
effect of diesel particulate emissions on its visibility: 1)
city radius, 2) baseline visibility, and 3) the ambient
concentration of diesel particulate. The baseline visibility
and the nominal radius have already been estimated for each
city and metropolitan area with 100,000 inhabitants or more and
are listed in Tables A-3 and A-4 of the Appendix.
The ambient diesel particulate concentrations in 1995 for
various cities were estimated in Chapter 3 (see Table 3-4 of
that chapter). There are four concentration values relating to
the four regulatory scenarios: 1) best estimate sales, relaxed
controls, 2) best estimate sales, base controls, 3) worst case
sales, relaxed controls, and 4) worst case sales, base
controls. However, these ambient diesel particulate
concentrations are not available for every city with 100,000
people or more. Thus, the available concentrations were
averaged according to city size and used for those cities for
which projections were not available. The four city-size
categories and their corresponding average particulate
concentrations are listed in Table 4-2.
Applying the baseline visibilities, nominal radii, and
1995 ambient diesel particulate concentrations to the
appropriate city situation (represented by Equation 6 or 7)
yields absolute visibility levels in 1995 for each of the four
scenarios. From these city-specific visibility projections,
the effect of 1995 diesel particulate concentrations on
baseline visibility can be estimated. The average visibility
reduction for each city-size category and diesel control
scenario is shown in Table 4-3.
As can be seen, the visibility impact of all scenarios is
strongly dependent on city size, with the larger cities
experiencing the larger effect. This is primarily due to the
greater diesel particulate concentrations projected for larger
cities (see Table 4-2). However, the especially large
visibility effects experienced by the cities having a
population of more than one million is also due to their larger
estimated radii. As was described earlier, the entire
metropolitan area was assumed to be affected by diesel
particulate emissions in these instances, where in the cases of
the three smaller groupings, only the city proper was assumed
to be affected.
-------
4-14
Table 4-2
Average Diesel Particulate Concentrations, ug/m 3
City Size
(Population)
More than 1,000,000
500,000-1,000,000
250,000-500,000
100,000-250,000
Best Estimate
Relaxed
4.38
2.69
2.19
1.92
Sales
Base
2.74
1.68
1.37
1.20
Worst Case
Relaxed
6.52
4.00
3.27
2.86
Sales
Base
4.25
2.61
2.13
1.86
-------
4-15
Table 4-3
Average Reduction in Visibility
Due to Diesel Particulate, Percent
City Size Best Estimate Sales Worst Case Sales
(Population)
More than 1,000,000
500,000-1,000,000
250,000-500,000
100,000-100,000
Relaxed
19.9
7.2
5.2
2.9
Base
13.5
4.6
3.1
1.8
Relaxed
27.0
10.7
7.1
4.4
Base
19.4
7.1
5.1
2.8
-------
4-16
With respect to the various scenarios, under best estimate
sales the relaxed scenario reduces visibility by 3-20 percent.
These visibility reductions are reduced to 2-14 percent under
the base scenario. Under worst case sales the visibility
reductions under both the relaxed and base scenarios are much
greater, 4-27 percent and 3-19 percent, respectively. In both
cases, the base scenario removes approximately one-third of the
visibility reduction of the relaxed scenario.
VI. Summary
A method exists to determine the visibility impact of a
specific level of diesel particulate. The necessary input data
include the ambient mass concentration of the diesel
particulate, the extent of this concentration (assumed to be
the city limit), the extinction efficiency of diesel
particulate (a value of 8 m^/g is used) and baseline
visibilities for each city.
Visibility levels in 1995 for all U.S. cities with more
that 100,000 inhabitants were projected under four regulatory
scenarios. The larger cities showed a greater reduction in
their visibility levels for each scenario. Under the best
estimate diesel sales scenario, the relaxed control scenario
resulted in a visibility loss of 3-20 percent, while the
visibility reduction under the base scenario was 2-14 percent.
Under the worst case diesel sales scenario, visibility
decreased 4-27 percent under the relaxed scenario and 3-19
percent under the base standards. In both cases, the base
scenario removed about one-third of the loss in visibility due
to diesel particulate emissions under the relaxed scenario.
-------
4-17
References
1. "Visibility Protection for Class I Areas, The
Technical Basis," University of Washington, Seattle, Prepared
for Council on Environmental Quality, Washington, D.C.,
Pb-288842, August 1978.
2. "Heavy-Duty Diesel Particulate Regulations, Draft
Regulatory Analysis," U.S. EPA, Office of Mobile Sources,
Chapter V, December 1980.
3. "A Study of Particulate Emissions from Motor
Vehicles, A Report to Congress," U.S. EPA, Office of Research
and Development, Bradow, et al., 214 Draft, Section 7.2.
4. "Characterization of Gaseous and Particulate
Emissions from Light-Duty Diesels Operated on Various Fuels,"
Southwest Research Institute, EPA-460/3-79-008, 1979.
5. "Existing Visibility Levels in the U.S.," Trijonis
and Shepland, Technology Service Corporation for U.S. EPA,
Grant No. 802815, EPA-450/5-79-010, 1979.
6. "Impact of Light-Duty Diesels on Visibility in
California," Trijonis, Final Report for California Air
Resources Board , Contract No. Al-117-32, 1982.
7. "Visibility in the Southwest - and Exploration of
the Historical Data Base," Trijonis, Atmospheric Environment,
Vol. 13, pp. 833-843, 1978.
8. "Visibility in the Northeast," Trijonis and Yuan,
Technology Service Corporation for U.S. EPA, Grant No. 803896,
EPA-600/3-78-075, 1978.
-------
CHAPTER 5
CANCER RISK ASSESSMENT
I. Introduction
Of the potential health effects associated with diesel
particulate emissions, the greatest concern has been associated
with its potential carcinogenic effects. This chapter will
examine the state of knowledge concerning the carcinogenic
potency of diesel particulate and estimate the effect of
various diesel particulate control scenarios on an individual's
lung cancer risk. The non-cancer health effects associated
with diesel particulate are examined in Chapter 6.
The first section of this chapter reviews the major
studies which have investigated the carcinogenic potency of
diesel particulate. The second section compares the results of
these studies and selects a likely range of carcinogenic
potency for diesel particulate. The third and final section
combines this carcinogenic potency with the 1995 exposure
estimates made in Chapter 3 to estimate the annual lung cancer
risk for an individual under each control scenario.
II. Review of Major Studies
The potential carcinogenicity of diesel particulate has
been examined through both human epidemiological studies and
clinical studies on animals and other lower organisms. Because
the epidemiological data base is limited, much weight has had
to be placed on the clinical studies. These clinical studies
estimate the carcinogenic potency of diesel particulate by
comparing their clinical results to the clinical results of
other cancer-causing substances for which human epidemiological
data are available. In this section, past and current
epidemiological studies will first be reviewed, followed by a
review of the comparative potency analyses.
A. Epidemiological Studies
The ideal means to determine the risk of developing lung
cancer from a given exposure of diesel particulate is to
conduct a long-term epidemiological study. Such a study would
trace the health of several well-defined groups of people who
were exposed to precisely known concentrations of diesel
particulate for known periods of time. Comparable groups that
were not exposed would also be monitored in order to detect any
differences in cancer rates. The influence of such factors as
diet, family history and smoking would be known in order to
strengthen the validity of the study's findings and reduce the
margin for error.
-------
5-2
Unfortunately, this ideal is usually not attainable in
real life and some potential error in the final results must be
accepted if an epidemiological study is to be performed at
all. This is the case with respect to epidemiological studies
of diesel particulate. Two such studies will be reviewed: 1)
the London Transit Authority Study, which was completed a
number of years ago, and 2) the U.S. Railroad Workers Study,
which is still underway.
1. London Transit Authority
Of the epidemiological studies completed to date which
specifically examine diesel emissions, the London Transport
Authority (LTA) Study is generally considered to be the most
thorough, although it too has significant deficiencies.[1] This
study initially examined the lung cancer incidence among
different groups of LTA employees between 1950 and 1954, [2] and
was later updated to include the years through 1974. [3] Among
the groups followed were diesel bus garage workers (generally
high level of exposure) and design engineers (generally low
level of exposure). Lung cancer incidences were identified
from information on the death certificates of those who were
still employed by the LTA at the time of death, ill-health
retirement records, and the records of transfers to other LTA
job categories. The study did not continue to monitor the
health of individuals once they were no longer employed by the
LTA. This is an area of potential bias since cancer typically
develops several years after initial exposure to carcinogens or
even after exposure terminates.
Other weaknesses of the study include the fact that the
extent of individual exposure to diesel exhaust was not
measured. Instead, particulate concentrations were simply
measured inside and outside of selected garages on a few
separate days during the 1950-74 observation period. Also, no
specific cohort of employees was identified and followed
throughout the study. Thus, the potential influence of such
factors as smoking habits, medical history and related
socioeconomic characteristics is not known.
The study found that the cancer incidence of the highly
exposed group was actually less than that expected based upon
Greater London lung cancer death rates in the 1950-74
timeframe. Thus, the study concluded that in regard to this
study population, no evidence existed associating lung cancer
to diesel engine exhaust.
However, this study has been analyzed independently by Dr.
Todd Thorslund of EPA's Carcinogen Assessment Group and Dr.
Jeffrey Harris, a member of the Analytical Panel of the
-------
5-3
National Academy of Sciences (NAS) Diesel Impacts Study
Committee. Both found that the potential errors involved in
the LTA study could have resulted in a sizeable underestimation
of the carcinogenic potency of diesel particulate.[1,4]
(
Based on the analyses by Thorslund and Harris, it is
possible that significant excess cancer deaths could result in
the general population even though the LTA study showed no
excess cancer deaths in the diesel particulate exposed group.
Thus, the many design flaws of the LTA work disqualify it from
further consideration in this study.
2. U.S. Railroad Workers
Another epidemiological study is currently being conducted
by Harvard University to evaluate the possible carcinogenic
effect of diesel exhaust in U.S. railroad workers. Data for
the study come from the U.S. Railroad Board. Components of the
study include: 1) a retrospective cohort analysis of
approximately 57,000 male railroad workers, 2) a case-control
study of 300 incident lung cancer cases and matched controls of
railroad workers, and 3) actual environmental monitoring of
worker exposure to diesel exhaust. These approaches will allow
for quantitative assessment of both level and duration of
diesel exhaust exposure and consideration of the major
confounding factor (cigarette smoking), thus removing the major
study design weaknesses of the LTA study.
The retrospective cohort consists of approximately 57,000
male railroad workers, aged 40-64 in 1959, with 10-20 years of
railroad service at that time. These workers were selected
from job categories having high diesel exposure and an
appropriate sample of control exposure categories. The massive
amount of data being generated in the retrospective and
case-control studies is currently being analyzed. Air
pollutants being monitored in five round-house locations
include nitrogen dioxide, sulfur dioxide, carbon monoxide,
respirable and non-respirable particulate and its constituents,
such as sulfates, polycyclic aromatic compounds and other
organic compounds. In addition, fractions of the particulate
sample extracts will be analyzed by mutagen bioassays, such as
the Ames test. A qualitative comparison of automobile diesel
exhaust with the railroad diesel exhaust will be performed by
correlating the gas chromatography/mass spectra of the
polycyclic aromatic compounds of each.
A pilot study was undertaken to evaluate the feasibility
of this larger study. The cohort of the pilot study consisted
of approximately 2,500 male railroad workers who were between
the ages of 45 and 64, working in 1967 and who had at least 10
-------
5-4
years of railroad service. Of these workers, 69.8 percent were
in occupations exposed to high concentrations of diesel
exhaust. The risk ratio for lung cancer in diesel exposed
workers relative to unexposed workers was 1.42, or a 42 percent
increase. However, this increased risk was not statistically
significant due primarily to the small size of the cohort.[6]
The larger retrospective cohort study and the case-control
study are currently in progress and are scheduled for
completion in October 1983.
B. Comparative Potency Analyses
Due to the limited epidemiological data available,
estimations of the human lung cancer risk from diesel
particulate have been made using a comparative potency method
developed by EPA. In this comparative potency method, the
results of non-human laboratory bioassays are used to compare
the carcinogenic and mutagenic potencies of diesel particulate
(specifically, the particle-bound organics) with those of other
combustion and pyrolysis products that have been shown by
epidemiological data to cause lung cancer in humans. Estimates
of the human lung cancer risks from exposure to these
established carcinogens, based on epidemiological studies, can
then be adjusted by the corresponding estimates of their
potencies relative to diesel particulate to yield estimates of
the lung cancer risk from diesel particulate. The equation
used is given below.
Estimated _ Human Risk Bioassay Potency (diesel)
Human Risk (carcinogen) x Bioassay Potency (carcinogen)
(diesel)
The ratio of potencies obtained from the same bioassay is
referred to as the relative potency.
The human carcinogens (comparative sources) selected by
EPA were coke oven emissions, roofing tar emissions and
cigarette smoke condensate. The mobile source samples selected
included those from a HDDE (Caterpillar 3304) , three LDDVs
(Datsun Nissan 220C, Oldsmobile 350, and Volkswagen
turbocharged Rabbit), and a gasoline-fueled, catalyst-equipped
vehicle (Ford Mustang II) . The organics extracted from the
particulate emitted from these sources were used to determine
the relative potencies.
The comparative sources and the mobile source samples were
both tested in mutagenesis and carcinogenesis bioassays. The
mutagenesis bioassays selected included reverse mutation in
Salmonella typhimurium (Ames test), forward mutation in L5178Y
mouse lymphoma cells, forward mutation in Balb/c 3T3 mouse
embryo fibroblasts, forward mutation in Chinese hamster ovary
-------
5-5
cells, mitotic recombination in Saccharomyces cerevisiae, DNA
breakage in Syrian hamster embryo cells, and sister chromatid
exchange in Chinese hamster ovary cells. The carcinogenesis
bioassays included oncogenic transformation in Balb/c 3T3
cells, viral enhancement of transformation in Syrian hamster
embryo cells, and skin initiation and skin carcinogenicity in
SENCAR mice. Further details of the study design can be found
elsewhere. [7]
The potencies obtained in these bioassays, together with
epidemiological data on the comparative sources, were combined
to estimate the human lung cancer risk from diesel particulate
in three independent analyses performed by Dr. Jeffrey Harris,
Lovelace Biomedical and Environmental Research, and EPA. Each
will be discussed below. The analyses differ with respect to
the choice of bioassays selected for determination of the
relative potencies and the choice of comparative source
epidemiological data.
It should be noted that EPA did not conduct new
epidemiological studies as part of this approach, but rather
relied upon existing data. For coke ovens, the work of
Mazumdar[8] and Land[9] was used, for roofing tar emissions
Hammond1 s [10] data were applied, and that of Dell and Peto[ll]
were .used in the case of cigarette smoke. The Harris and
Lovelace analyses relied upon the same coke oven and roofing
tar data. For cigarette smoke, Lovelace used the data of
USHEW,[12] Hammond[13] and Kahn, [14] which resulted in a risk
estimate similar to that obtained by EPA. Harris did not
include cigarette smoke as a comparative source in his analysis.
1. Harris
In addition to his analysis of the London Transit
Authority Study, Harris conducted a comparative potency
analysis for the National Academy of Sciences.[1] The
comparative source emissions selected by Harris were coke oven
emissions and roofing tar emissions. Using a linear relative
model, Harris analyzed the epidemiological data for coke oven
and roofing tar emissions to obtain estimates of the
proportional increase in lung cancer incidence per unit of
cumulative lifetime exposure to coke oven emissions (0.044) and
roofing tar emissions (0.015).
Harris used data from three short-term bioassays to
estimate the relative potencies of the diesel (light-duty only)
and comparative source samples. The bioassays used were tumor
initiation in SENCAR mice by skin painting, enhancement of
viral transformation in Syrian hamster embryo cells, and
mutagenesis in L5178Y mouse lymphoma cells. The results from
-------
5-6
these bioassays can be found in Tables A-l and A-2 of the
Appendix. Harris then applied these relative potencies to his
estimates of the proportional increase in lung cancer incidence
from exposure to coke oven and roofing tar emissions to obtain
estimates of the proportional increase in lung cancer incidence
from exposure to diesel emissions.
Harris1 overall estimate was a 0.0035 percent proportional
increase in lung cancer incidence per unit exposure (i.e., one
microgram per cubic meter of diesel particulate for one year).
This is roughly equivalent to 1.4 x 10"^ incidences of lung
cancer per person per year due to a continuous lifetime
exposure of one microgram per cubic meter of diesel
particulate.*
2. Lovelace Biomedical and Environmental Research
Lovelace used two methodologies to estimate the cancer
risk from exposure to LDD particulate.[15]
The first method assumed that diesel particulate was not
more mutagenic or carcinogenic than the most potent of coke
oven, roofing tar or cigarette particulate. First, the annual
lung cancer risk per person for each of the three carcinogens
were estimated from the epidemiological studies of coke oven
workers, roofers, smokers and nonsmokers using a linear,
nonthreshold model. Then, the average concentration of each
type of particulate inhaled over a year was estimated and used
to estimate the annual unit lung cancer risk per individual for
these comparative sources. All of these figures are presented
in Table 5-1. Lovelace then assigned a figure of 1.5 x 10"^
lung cancers per person due to .a lifetime exposure of one
microgram per cubic meter of diesel particulate as an upper
estimate of the potency of diesel particulate. This figure was
based primarily on the estimated annual unit risks for coke
oven and roofing tar particulate, which are both between 1.0 x
10"6 and 1.5 x 10~6.
The second method used the bioassay data developed by EPA
to estimate the relative potencies of the LDD and comparative
source samples. Like Harris, these relative potencies were
then applied to the unit risks derived from the epidemiological
studies of the known carcinogens.
The proportional increases in lung cancer incidence
obtained by Harris were translated into absolute measures
of lung cancer incidence independently by Thorslund.[5]
-------
5-7
Table 5-1
Summary of Inhalation Exposures and Annual
Lung Cancer Risks for Surrogate Populations - Harris*
Study
Population
Coke Oven
Workers
Roofers
Smokers:
(cigarettes/
day)
1-9
10-19
20-39
40+
Urban
Nonsmokers
Rural
Nonsmokers
Average Air[a]
Concentration
of Particles
(mg/m3)
3
1
2-16
18-35
36-71
73 +
0.06
0.03
Annual Lung Cancer
Risk x 10^ (per
person, per year)
4000
1100
260
470
800
1070
70
30
Annual Risk
x 106
(per person,
per ug/m,
per year)
1.3
1.1
0.03
0.02
0.02
0.01
1.2
1.0
* Information in this table was excerpted from Reference 15.
[a] The average air concentration of particles was estimated as
the total mass of particles inhaled per year divided by all
of the air breathed per year (assumed to be 20 m3 per day
X 365 days per year).
-------
5-8
The comparative sources selected from the EPA work were
coke oven emissions, roofing tar emissions and cigarette smoke
condensate. Urban soot was also selected independently by
Lovelace as an additional comparative source. The mutagenesis
bioassays used were the Ames assay, forward mutation in Chinese
hamster ovary cells (HGPRT gene locus assay), forward mutation
in L5178Y mouse lymphoma cells, and forward mutation in Balb/c
3T3 mouse embryo fibroblasts. The carcinogenesis bioassays
used were oncogenic transformation in Balb/c 3T3 cells, viral
enhancement of transformation in Syrian hamster embryo cells,
and skin initiation and skin carcinogenicity in SENCAR mice.
These bioassay data are presented in Table A-3 of the Appendix.
The overall relative potencies resulting from a comparison
of the data in Table A-3 are shown in Table 5-2, along with the
annual unit risks already presented in Table 5-1 and the
estimated annual unit risks for diesel particulate resulting
from each comparison. When only the comparative sources used
by EPA are considered (coke oven, roofing tar and cigarette
smoke condensate), the annual unit risk estimates for diesel
particulate range from 0.07 x 10"6 to 0.6 x 10~6 lung
cancers per person per year due to a constant lifetime exposure
of one microgram per cubic meter of diesel particulate (unit
exposure) . When urban soot is also considered as a comparative
source, the range increases to 0.07 x 10~6 to 3.0 x 10~6.
Based on the results of both methods, Lovelace chose 1.0 x
10~6 as being the most representative estimate for the annual
unit lung cancer risk due to diesel particulate.
3. Environmental Protection Agency (EPA)
Members of EPA's Office of Research and Development also
recently estimated the annual unit cancer risk of diesel
particulate using a comparative potency method very similar to
that used by both Harris and Lovelace. [16] The comparative
sources used in this analysis were coke oven, roofing tar and
cigarette smoke. Epidemiological data for coke oven workers,
roofing tar workers and cigarette smokers were examined using a
linear, nonthreshold model to determine the annual unit lung
cancer risk for each carcinogen. A summary of these risk
estimates can be found in Table A-4 of the Appendix.
The relative potencies of the coke oven, roofing tar,
cigarette smoke condensate and mobile source samples were
evaluated by a large number of bioassays which have already
been described. The bioassays used in the final determination
of the relative potencies were the tumorigenicity bioassays
involving skin initiation and skin carcinogenicity in SENCAR
mice, the Ames bioassay, the L5178Y mouse lymphoma cell
-------
5-9
Table 5-2
Lung Cancer Risk
From Exposure to Diesel Exhaust Based Upon
Relative Potencies of Surrogate Substances - Lovelace*
Surrogate
Exposure
Coke Oven
Emissions
Roofing Tar
Vapor
Cigarette
Smoke
Condensate
Urban Soot
Selected
Diesel Lung
Cancer Risk
Median
Relative
Potency
(surrogate
to diesel)
0.3
0.4
Annual
Cancer Risk
(per person
per ug/m3)
1.3
1.1
0.02
1.2
Estimated Risk of
Diesel Particlate
(per person
per ug/m^)
0.3
0.6
0.07
3.0
1.0
Information on this table was excerpted from Reference
-------
5-10
mutagenesis bioassay, and the sister chromatid exchange (SCE)
bioassay in Chinese hamster ovary cells. The results from
these tests are given in Tables A-5 and A-6 of the Appendix.
It should be noted that the mobile source and comparative
source samples were also evaluated in a number of additional
bioassays. The bioassays used in this analysis (and those
selected in the Harris and Lovelace studies) were selected for
their ability to produce dose-related effects and the strength
and relevance of the end point being measured. The relative
potencies are shown in Tables A-7 and A-8.
Two steps were subsequently followed to determine the lung
cancer risks for the diesels and the gasoline vehicle. First,
the relative potencies in the mouse skin tumor initiation assay
(Table A-7) were used to obtain the annual unit risk estimate
for the Nissan particulate from the annual unit risks for the
coke oven, roofing tar and cigarette smoke condensates (Table
A-4). Second, the annual unit risks for the other diesel
particulates were obtained by multiplying the annual unit risk
of the Nissan particulate by the net relative diesel potencies
of Table A-8, which were based on the Ames, lymphoma and SCE
bioassays.
The annual unit lung cancer risk estimates resulting from
these calculations are shown in Table 5-3. Since the organics
extracted from the particulate were used in the bioassays, the
risk estimates were calculated in terms of organics and then
converted in terms of particulate. For the three LDDVs, the
annual risk estimates per person range from 0.26 x 10~" to
0.46 x 10~6 due to lifetime exposure to one microgram per
cubic meter of particulate. It is interesting to note that
emissions from the Caterpillar HDDE had about one-tenth the
potency of the LDDVs. The reason for this is not clear at this
time, but it may be due to the fact that the Caterpillar
particulate had been stored for more than a year before use.
Due to the unexplainable nature of this difference, the risk
estimate for the Caterpillar particulate will not be used
further.
III. Choosing a Value or Range of Values
A summary of the risk estimates obtained from the three
comparative risk studies is shown in Table 5-4. It should be
obvious from the preceding discussion and a comparison of the
figures in Table 5-4 that there is no concensus among the
scientific community as to the carcinogenic potency of diesel
particulate. The three potency studies differ at nearly all
possible points: 1) the estimated annual unit cancer risks of
the known carcinogens, even though, for the most part, the same
epidemiological data are used, 2) the non-human bioassays
-------
5-11
Table 5-3
Unit Lung Cancer Risk Estimates
for Diesel Particulate - EPA*
Unit Risk Estimates
(annual risk/ug/m3)
Diesel Source Organics Particulate
Nissan[a] 0.58 x 10~5 0.46 x 10~6
Volkswagen Rabbitfb] 0.17 x 10"5 0.30 x 10~6
Oldsmobilefb] 0.16 x 10~5 0.26 x lO'6
Caterpillar[b] 0.87 x 10'7 0.024 x 10'6
* This table was excerpted from Reference 16 in which
lifetime risks were presented. These risks have been
converted to annual risks by dividing by the median
lifespan (76.2 years).
[a] Based on average relative mouse skin tumor initiation
activity (Table A-7).
[b] Based on average relative activity in the mouse lymphoma,
SCE, and Ames Bioassays (+MA) (Table A-8).
-------
5-12
Table 5-4
Summary of Lung Cancer Risk Estimates
Annual Risk x
Comparative Potency (per person per
Analysis particulate)
Harris 1.4
Lovelace 1.0[a]
EPA 0.26-0.46[b]
[a] When the EPA comparative sources were used, the risk
estimates obtained by Lovelace range from 0.07 x 10~6 to
0.6 x 10~6. When urban soot was also considered by
Lovelace as a comparative source, the risk estimates range
from 0.07 x 10~° to 3.0 x 10"*6. Lovelace chose 1.0 x
10~6 as being most representative.
[b] Since the heavy-duty Caterpillar sample is not considered
representative, the range of risk estimates is restricted
to the range of risk estimates obtained for the light-duty
vehicles.
-------
5-13
selected for actual derivation of the relative potencies, and
3) the relative weightings given to those bioassays selected.
In addition, all of these studies rely on the assumption
that the relative carcinogenic potencies of diesel emissions
and the related environmental emissions are preserved across
human and non-human biological systems. Although this
assumption has not been proven correct, it is the best one that
can be made until a reliable epidemiological study focusing on
exposure to diesel exhaust is performed.
It should also be noted that all of the comparative
potency analyses discussed used a linear, nonthreshold
dose-response model to extrapolate cancer incidence to lower
doses. While this has been the most widely used model in the
past, others are gaining more use presently. Figure 5-1
depicts two typical examples of other models: an infralinear
model and a linear, threshold model.[17] Since all the
exposures simulated in the non-human laboratory tests are very
high to demonstrate effects with small number of specimens, the
results must be extrapolated downward to lower, more realistic
doses. Examining Figure 5-1, the point common to all three
models can be taken to be the result of the high-dose
bioassay. Then, as can be seen, both the infralinear and
linear, threshold model result in lower, low-dose risks than
the linear, nonthreshold model. Because of this, depending on
which model is correct, the use of the linear, nonthreshold
model could overestimate the cancer risk at lower doses. To
date, however, the linear, nonthreshold model has been the only
one applied to diesel particulate emissions.*
Because of the lack of consensus among the various
studies, this study will use the range of risk estimates
obtained from the comparative potency analyses of Harris,
Lovelace and EPA. Referring to Table 5-4, the range of risk
estimates selected for this analysis is 0.26 x 10~° to 1.4 x
10~6 lung cancers per person per year due to a constant
lifetime exposure of one microgram per cubic meter of diesel
particulate.
IV. Estimated Risk Based on Projected Diesel Exposure
The range of potency estimates for diesel particulate
derived in the previous section can be combined with the
There is one additional model, the supralinear model,
which actually results in a higher, low-dose risk than the
linear, nonthreshold model;[17] however, its application
to diesel particulate would be the furthest from being
established of all of the other models.
-------
en
-H
(-1
-------
5-15
projected scenario-specific diesel particulate exposures in
1995 derived in Chapter 3 to yield scenario-specific estimates
of the individual lung cancer risk due to diesel particulate in
1995. After this has been done, these individual lung cancer
risk estimates will be compared to cancer and accidental risks
from other sources.
A. Scenario-Specific Individual Lung Cancer Risks
The population exposures to diesel particulate from both
light- and heavy-duty vehicles in 1995 were derived in Chapter
3 for four scenarios: 1) best estimate diesel sales with the
relaxed control scenario, 2) best estimate diesel sales with
the base control scenario, 3) worst case diesel sales with the
relaxed control scenario, and 4) worst case diesel sales with
the base control scenario.
The potency estimates of Table 5-4, based on the linear
nonthreshold extrapolation model, only require the annual
average individual exposure to obtain estimates of annual
average cancer risk per individual. The projected nationwide
annual average exposure levels for individuals living in urban
areas in 1995 for each scenario, expressed in terms of
micrograms per cubic meter per year, are repeated from Chapter
3 in Table 5-5. These exposure estimates are then simply
multiplied by the range of individual potencies, expressed as
lung cancer risk per micrograms per cubic meter per year, to
obtain the range of estimated individual lung cancer risk in
1995 due to diesel particulate exposure under each scenario.
The resultant individual lung cancer risks in 1995 for
each scenario are also shown in Table 5-5. Individual lung
cancer risks in 1995 due to exposure to particulate from both
light- and heavy-duty diesels range from 0.8 x 10~" to 6.7 x
10~6 under the base control scenarios and 1.3 x 10~6 to
10.4 x lO"^ under the relaxed control scenarios.
As can also be seen from Table 5-5, the relative
contribution of LDD emissions is much greater assuming worst
case diesel sales than best estimate sales. Also, the
individual lung cancer risk is reduced by roughly 35 percent
under the base scenario relative to the relaxed scenario. The
effect of the base scenario is greatest with respect to the HDD
contribution.
3. Comparison of Diesel Cancer Risk with Other Risks
To place these estimated cancer risks in perspective, they
can be compared to current (generally 1981) individual risks
from other sources. The other individual risks provided for
-------
5-16
Table 5-5
Individual Diesel Cancer Risk Projections in 1995
Scenario
Projected Individual
Diesel Particulate
Exposure in 1995
(ug/m3-person-year)
Light-Duty
Heavy-Duty
TOTAL
Estimated Individual
Risk Based on Pro-
jected Diesel Par-
ticulate Exposures
in 1995 X 106
(lung cancer risk/
person-year)*
Light-Duty
Heavy-Duty
TOTAL
Best Estimate Sales Worst Case Sales
Relaxed
1.8
3.2
5.0
0.5-2.5
0.8-4.5
1.3-7.0
Base
1.5
1.6
3.1
0.4-2.1
0.4-2.2
0.8-4.3
Relaxed
3.7
3.7
7.4
1.0-5.2
1.0-5.2
Base
2.9
1.9
4.8
0.8-4.0
0.5-2.7
1.9-10.4 1.3-6.7
Individual lung cancer risks in 1995 were obtained by
multiplying the individual diesel particulate exposure in
1995 for each scenario by the range of potency estimates
for diesel particulate (0.26 x" 10"6 - 1.4 x 10~6
risk/person-year-ug/m3).
-------
5-17
comparison include commonplace (accidental) risks of death,
most of which would be considered involuntary (unavoidable) ,
and cancer risks from exposure to various sources. Also
included is the risk of death from lung cancer for smokers
whose deaths are attributable to smoking, along with the risk
from lung cancer for the general population whose deaths are
attributable to causes other than smoking. These risks,
expressed as individual cancer risk or probability of death per
year, are given in Table 5-6.
Accidental risks are generally applicable to the entire
U.S. population. As can be seen in Table 5-6, the aggregate
risk for tornadoes, floods, lightning, tropical cyclones and
hurricanes is within the same order of magnitude as that given
for diesel particulate. In contrast, the"risks of not wearing
seat belts, burns, drowning and motor vehicle accidents all
exceed the risk projected for exposure to diesel particulate.
The risk of a motor vehicle accident is more than an order of
magnitude greater than the maximum risk given for diesel
particulate.
In addition to the accidental risks discussed above,
cancer risks which result from dietary . and occupational
exposures are included for comparison. These cancer risks are
roughly within the same order of magnitude as that for diesel
particulate. (The risk from lung cancer will be discussed
separately.) Exposures to many of the cancer risks given in
Table 5-6, including the risk from diesel particulate, can be
applied across the general U.S. urban population or a vast
majority of it. Exposures to the other cancer risks such as
arsenic, or being a frequent airline passenger, can only be
applied to a selected segment of the population. For example,
only 2.82 million people, or roughly 1 percent of the
population are exposed by virtue of their occupation to
atmospheric arsenic.[22] Thus, the number of people exposed to
arsenic is far less than those exposed to diesel particulate
and the other cancer risks whose exposures can be applied
across the general U.S. population. The number of people
exposed to each source should be taken into consideration when
making direct comparisons of risk.
In some cases, risks resulting from certain occupational
exposures far exceed those risks presented in Table 5-6. For
example, exposures to arsenic results in an individual annual
risk of respiratory cancer as high as 180 x 10~6 for those
few workers exposed near cotton gins. [22] For ethylene
dibromide, cancers can result from both dietary and
occupational exposures. The risk from dietary exposures to
ethylene dibromide is given in Table 5-6. The occupational
risks of cancer resulting from inhalation of ethylene dibromide
-------
5-18
Table 5-6
Comparison of Risks from Various Sources*
Sources of Risk
Diesel Particulate:
Relaxed Scenario
Base Scenario
Commonplace Risks
Motor Vehicle Accident[18]
Not Wearing Seat Belts[19]
Drowning[18]
Burns[18]
Tornados, Floods, Light-
ning, Tropical Cyclones
and Hurricanes[20]
Cancer Risks
Natural Background Radi-
ation (sea level)[20]
Average Diagnostic Medical
X-Rays in the United
States[20]
Frequent Airline Passenger
(4 hours per week
flying)[20]
Four Tablespoons Peanut
Butter Per Day (due to
presence of aflatoxin)[20]
Ethylene Dibromide[21]
One 12-Ounce Diet
Drink Per Day[20]
Arsenic[22]
Miami or New Orleans
Drinking Water (due
to presence of
chloroform)[20]
Lung Cancers:
For Smokers Due to
Smoking[23]
For General Population
Due to Causes Other
Than Smoking[23]
Estimated Risk
(risk/person-year)
1.3 x 10~6 - 10.4 x 10~6
0.8 x 10~6 - 6.7 x 10~6
222.0 x 10~6
112.0 x 10~6
26.0 x 10~6
21.0 x 10~6
2.0 x 10~6
20.0 x 10~6
20.0 x 10-6
10.0 x 10~6
8.0 x 10~6
4.2 x 10~6
2.6 x 10~6
1.7 x lO'6
1.0 x ID'6
419.0 x 10'6
73.9 x ID"6
Exposed
Population
Urban U.S,
Entire U.S.
Entire U.S.
General U.S
Entire U.S.
General U.S,
Entire U.S,
Widespread
Limited
Fairly
Widespread
Widespread
Widespread
1% of U.S.
Southern
U.S., Urban
Entire U.S.
In some cases, an average lifetime of 76.2 years
assumed to convert a liftime risk to an annual risk.
was
-------
5-19
vapor can be as high as 5.2 x 10~3 for citrus warehouse
laborers.[21]
The risk of lung cancer for smokers whose deaths are
attributable to smoking, along with the risk from lung cancer
for the general population whose deaths are attributable to
causes other than smoking, are also included in Table 6 for
comparison. The maximum lung cancer risk given for diesel
particulate is roughly 2.5 percent of the lung cancer risk for
smokers whose deaths are attributable to smoking, and 14
percent of the lung cancer risk for the general population
whose deaths are attributable to causes other than smoking.
The analogous figures for the minimum diesel exposure are 0.2
percent and 1.1 percent, respectively. As can be seen, smoking
is the primary contributor to lung cancer deaths in the U.S.
(85 percent).
-------
5-20
References
1. "Potential Risk of Lung Cancer from Diesel Engine
Emissions," Harris, J., National Academy Press, Washington,
D.C., 1981.
2. "The Health of the Worker," British Journal of
Industrial Medicine, Raffle, P., Vol. 14, pp. 73-80, 1957.
3. "Trends in
j. -.nenus JLII Lung Cancer in London in Relation to
Exposure to Diesel Fumes," In: Health Effects of Diesel Engine
Emissions: Proceedings of an International Symposium, Waller,
Vol. 2, EPA-600/9-80-057b, 1980.
R.
4. Answer to the Posed Question: Are the Results
Obtained in the London Transit Worker Study Sufficient to
Dismiss Any Concern Regarding the Potential Cancer Hazard for
the U. S. Population in the Future, Due to Diesel Engine
Exhaust?, EPA Memo From Todd Thorslund, Carcinogen Assessment
Group to Michael Walsh, Mobile Source Air Pollution Control,
January 29, 1981.
5. "A Suggested Approach for the Calculation of the
Respiratory Cancer Risk Due to Diesel Engine Exhaust,"
Presented at the EPA Workshop on the Evaluation of Research in
Support of the Carcinogenic Risk Assessment for Diesel Engine
Exhaust, Thorslund, T. W., February 24-25, 1981.
6. "Lung Cancer and Occupational Exposures to Diesel
Exhaust: A Pilot Study of Railroad Workers," Schenker, M. B.,
T. Smith, A. Munoz, S. Woskie, and F. Speizer, Draft Submitted
for Publication, 1982.
-_,----- _- . - _ — ___^_ . _
Diesel and Related Environmental Emiss
Sample Generation, Collection and P
International, Lewtas, J., R. L. Bradow,
Harris, R. B. Zweidinger, K. M. Cushing,
Albert, Vol. 5, pp. 383-387, 1981.
"Mutagenic and Carcinogenic Potency of Extracts of
1 Related Environmental Emissions: Study, Design,
neration, Collection and Preparation," Environ.
-» =s 1 T ^ T.I 4- -a e? T "D T n*-=s/^^r.T D U T i -I TM-T a »- e? T3 F\
R.
B.
H.
E.
Jungers,
Gill and
B.
R.
D.
E.
8. "An Epidemiological Study of Exposure to Coal Tar
Pitch Volatiles Among Coke Oven Workers," Journal of Air
Pollutant Control Association, Mazumdar, S., C. K. Redmond, W.
Sollecito, and N. Sussman, Vol. 25, pp. 382-389, 1975.
9. Presentation at
Standards, Land, C. E., 1976.
OSHA
Hearings on Coke Oven
"Inhalation of Benzo-a-pyrene and Cancer in
:ad. Sci. , Hammond, E. D. , I. J. Selikoff,
id H. Seidman, Vol. 271, pp. 161-124, 1976.
10.
Ann. NY Acad. Sci. , Hammor
Lawther, and H. Seidman, Vol
Man,"
P. L.
-------
5-21
References (cont'd)
11. "Cigarette Smoking and Brochial Carcinoma: Dose and
Time Relationships Among Regular Smokers and Lifelong
Non-Smokers," J. Epidemiol. Community Health/ Doll, R. and R.
Peto, Vol. 32, pp. 303-313, 1978.
12. "Smoking and Health: A Report of the Surgeon
General," U.S. Department of Health Education and Welfare, DHEW
Publication No. (PHS) 79-50066.
13. "Quantitative Relationship Between Cigarette Smoking
and Death Rates," Natl. Cancer Inst. Monogr., Hammond, C. E.,
Vol. 28, pp. 3, 1968.
14. "The Dorn Study of Smoking and Mortality Among U.S.
Veterans: Report on Eight and One-Half Years of Observation,"
Kahn, H. A., EPT-Demiological Approaches to the Study of Cancer
and Other Chronic Diseases, Haenszel W., Editor, Natl. Cancer
Inst. Monogr, Vol. 19, pp. 1, 1966.
15. "Potential Health and Environmental Effects of
Light-Duty Diesel Vehicles II," Cuddihy, R. G., W. C. Griffith,
C. R. Clark, and R. 0. McClellan, Lovelace Biomedical and
Environmental Research Institute, Inhalation Toxicology
Research Institute Report LMF-89, 1981.
16. "A Comparative Potency Method for Cancer Risk
Assessment: Application to Diesel Particulate Emissions,"
Albert, R. E., J. Lewtas, S. Nesnow, T. W. Thorslund and E.
Anderson, Submitted to Risk Analysis, 1982.
17. "Assessment of Technologies for Determining Cancer
Risks from the Environment," Office of Technology Assessment,
June 1981.
18. The World Almanac and Book of Facts 1983, New York,
New York, 1983.
19. "Rediscover the Safety Belt", U.S. Department of
Transportation, National Highway Traffic Safety Administration,
1983.
20. Risk/Benefit Analysis, Wilson, R. , and E. Crouch,
Cambridge, Massachusetts, 1982.
21. "Ethylene Dibromide: Position Document 2/3",
Special Pesticide Review Division, Environmental Protection
Agency, EPA/SPRD-81/74, 1980.
-------
5-22
References (cont'd)
22. "Final Risk Assessment on Arsenic," Carcinogen
Assessment Group, Environmental Protection Agency,
EPA-600/6-81-002, May 1981.
23.. Estimates for total lung cancer deaths in 1981
obtained from the National Center for Health Statistics,
Department of Health and Human Services. Percentages of lung
cancer deaths attributable to smoking and other causes obtained
from Clearinghouse on Smoking and Health, 1982 Report on
Smoking and Health.
-------
CHAPTER 6
NON-CANCER HEALTH EFFECTS OF
DIESEL PARTICULATE
I. Introduction
One of the primary concerns regarding diesel particulate
is its potential for adversely affecting human health. The
potential adverse health effects of this material can be
divided into two broad categories: 1) carcinogenic and 2)
non-carcinogenic, or non-cancer. This chapter deals
specifically with non-cancer health effects. The potential
carcinogenic effects of diesel particulate were already
discussed in Chapter 5.
Although a large amount of information documenting the
adverse health effects of inhaling particulate matter is
available in the literature, comparatively little deals
specifically with diesel particulate. However, concern over
the potentially adverse health effects of exposure to diesel
exhaust has recently increased, and has resulted in a
significant amount of new research concerning diesel
particulate and its effects on health.[1,2] Unfortunately,
because much of the diesel particulate health effects
information which is available is comparatively recent and has
not been peer reviewed by other scientists, very few conclusive
statements can be made regarding the health effects of diesel
particulate exposure. [3] Therefore, at this time, the best
approach for evaluating the non-cancer health effects of diesel
particulate is to evaluate the health effects of particles for
which established literature is available. However, before
outlining how this comparative analysis will be performed, it
is important to describe three things this analysis will not do.
First, the fact that particles in the ambient air can
cause adverse non-cancer health effects will not be established
here. It has long been recognized that exposure to various
forms of particulate matter can cause a wide variety of adverse
non-cancer health effects. These effects have been well
documented in the literature and total suspended particulate
matter was among the first airborne pollutants to have a NAAQS
established by EPA in 1971.
Second, in evaluating the documented non-cancer health
effects of particulate matter, the focus will not be on any
specific types of particulate, but rather on typical ambient
mixtures of particles. Obviously, some types of particulate
affect health differently than others. For example, soluble
particles may affect health through different mechanisms than
insoluble particles. Some specific particles also are
inherently more dangerous than others (e.g., radioactive
material). However, because it is generally impossible
-------
6-2
epidemiologically to ascribe the adverse health effects of
ambient exposures to any specific component of the particulate
mixture, the effects of specific particles are less important
than the effects of typical mixtures of particles found in the
atmosphere.
Third, this comparative analysis will not be conducted
quantitatively, but qualitatively. While a few quantitative
health effects studies based on measurements of total suspended
particulate or British smoke shade are available, the
extrapolation of these results to diesel particulate could only
be based on qualitative relationships and the quantitative
results would imply a degree of precision beyond that which was
defendable.
Proceeding to the description of what will be done, the
comparative analysis will be performed on two levels:
particulate inhalation characteristics and laboratory health
effects testing. The available information on each of these
levels will be presented first for ambient inhalable
particulate and second for diesel particulate, with a
comparison of the two sets of results following on each level.
An overall asessment will then be made as to whether or not
diesel particulate should be expected to affect health
(non-carcinogenically) disproportionate to its impact on
ambient mass particulate levels.
II. Non-Cancer Health Effects of Typical Particulate Matter
A. Inhalation of Particulate Matter
One of the most easily understood determinants of adverse
health effects from inhaling particulate matter is the "body
dose." For the purposes of this chapter, the important aspects
of body dose are: 1) where particles are deposited in the
respiratory tract, and 2) how these particles are cleared from
the system by natural defense mechanisms. Therefore, some
general knowledge regarding the structure of the respiratory
tract, in addition to deposition and clearance within the
system is a prerequisite for specifically discussing the
non-cancer health effects of particulate exposure.
The principal features of the respiratory system are
depicted in Figure 6-1. The upper respiratory tract begins
with the nares (or mouth during oral breathing) and ends at the
entrance to the trachea. The lower respiratory tract is
subdivided into the conducting airways (or tracheobronchial
region) and the gas-exchange region (or alveolar region). The
tracheobronchial region consists of the trachea down to the
minute terminal bronchioles. The alveolar region includes the
partially alveolated bronchioles and finally terminates with
the alveoli themselves. (A more complete description of the
respiratory tract as it relates to particle deposition can be
found in Reference 5.)
-------
. 6-3
Figure 6-1
Diagrammatic Representation of the
Human Upper and Lower Respiratory Tract[4]
Upper respiratory tract
Anterior nares
Lower
respiratory •
tract
Trachea
Bronchus
Aveoli
-------
6-4
1. Deposition in the Respiratory Tract
As stated above, the health effects associated with
particulate matter in the respiratory tract are dependent, to a
large degree, upon where in the tract deposition takes place.
Spatial deposition within the respiratory system is primarily
determined by particle size, with the mode of breathing (nose
versus mouth) also having a substantial effect on the
disposition of large particles.
Moving through the respiratory tract, deposition in the
upper respiratory tract during nose breathing is nearly 100
percent complete for particles with diameters larger than about
10 micrometers and declines to about 10 percent for particles
with diameters less than 1 micrometer.[5] During mouth
breathing, deposition in the upper respiratory tract is less
efficient, although the vast majority of large particles are
still removed in this region.
Most particles smaller than about 10-15 micrometers enter
the lower respiratory tract and are deposited, to varying
degrees, in the tracheobronchial and alveolar regions as shown
in Figure 6-2. In the tracheobronchial region, deposition
during mouth breathing is especially high for particles with
diameters of 5-10 micrometers (up to 80 percent removal) and
tapers off to a deposition of about 5 percent for particles
with diameters of 0.1-1 micrometers. Deposition of 5-10
micrometer particles in this region during nose breathing is
considerably less due to their previous deposition in the upper
respiratory tract.
In the alveolar region, deposition is almost nonexistent
for particles with diameters greater than about 10 micrometers,
since nearly all such large particles already would have been
deposited in the upper respiratory tract and the
tracheobronchial region. Deposition in the alveolar region
during mouth breathing peaks at about 60 percent for particles
with diameters of 3-4 micrometers and declines to around 15
percent for particles with diameters between 0.1-0.2
micrometers. This peak is still present during nose breathing,
but it's level is much less (20 percent). Generally, this
information shows that particles with diameters less than about
10-15 micrometers generally penetrate deeper into the
respiratory system than larger particles.
2. Clearance of Particulate Matter From the Respiratory
Tract
Clearance is the process whereby particles are removed
from the respiratory tract. This process is described in this
section in a simplified manner. It must be noted, however,
that the mechanisms for removing particles are often complex
and the efficiencies of these mechanisms often vary
significantly between individuals, due to such factors as
-------
6-5
Figure 6-2 [6,7]
Deposition in the Tracheobronchial
and Alveolar Regions
By Indicated Particle Diameter
I I I i
Range of alveolar deposition
mouth breathing.
Range of tracheobronchial deposition
mouth breathing.
Extrapolation of above to
predicted by Reference 4.
Extrapolation to point
demonstrated by Reference
Ff-CXXXXXXXXXXXXX
0.1 0.2 0.3 0 4 0.5
PHYSICAL DIAMETER. ;:m
1.0 2.0 3.0 4.0 5.0
AERODYNAMIC DIAMETER. ;.m
Ifi.O
20 30
-I r+-
I
NOTE: Deposition is expressed as fraction of particles of a given diameter
entering the mouth (or nose).
-------
6-6
smoking, pathological abnormalities, and response to inhaled
pollutants. A more complete description of respiratory
clearance is available in Reference 2.
Particulates may be removed from the respiratory tract in
two principal ways. First, particles which are soluble in body
fluids (or the soluble coating on insoluble particles) may
dissolve in any region of the respiratory system where
deposition occurs. After dissolution, the constituents of the
particle may interact locally with cells or tissues, or they
may be absorbed into the blood and transported to other areas
of the body.
Second, relatively inert and insoluble particles may be
removed from the respiratory tract by more mechanical means.
The process is somewhat specific to the various regions of the
system; therefore, each region is discussed separately.
Clearance of insoluble particles from the anterior portion
of the upper respiratory tract takes place mainly by blowing
the nose or sneezing. In the posterior portion of this region,
the conducting airways are lined with both ciliated cells that
have hairlike projections and mucus-secreting cells. Particles
that are deposited in these conducting airways are trapped in
the mucus and are mechanically transported by cilia action to
the throat, where they are either swallowed, entering the
gastrointestinal tract, or expectorated. This clearance
mechanism is called the "mucociliary conveyor." Clearance in
the upper respiratory tract is normally rapid (i.e.,
minutes).[6]
The primary clearance mechanism in the tracheobroncial
region is also the mucociliary conveyor. As described above,
entrained particles are transported to the throat where they
may be swallowed, thereby entering the gastrointestinal tract,
or expectorated. Smaller particles, which may deposit in the
smaller airways deeper in the lung, take longer to clear than
larger particles, which tend to deposit in the larger airways.
Generally, however, clearance from the tracheobronchial region
of the respiratory system normally takes hours to days.[6]
The principal clearance route in the alveolar region is
via alveolar macrophages. These specialized cells phagocytize
(i.e., engulf) deposited particulate matter. Some macrophages
containing particulate travel to the mucociliary conveyor of
the tracheobronchial region where they are cleared through the
gastrointestinal tract. Others travel to lymph nodes and are
cleared from the body through the lymphatic system. Clearance
of insoluble particles from the alveolar region generally takes
months or years.[6]
-------
6-7
3. Related Health Concerns
There are two principal concerns associated with the
deposition of inhalable particulate in the lower respiratory
tract. First, particles deposited in this area, even if not
directly toxic themselves (e.g., inert particles), may have
hazardous materials adsorbed onto their surfaces.
Consequently, these adsorbed, hazardous materials may be
transported deep into the most sensitive areas of the lung
where they may cause localized effects or be absorbed and
circulated to other parts of the body, causing problems
elsewhere. Second, all particles deposited in this area have
relatively long residence times. As discussed previously,
clearance of particles in the tracheobronchial region may take
days, while in the alveolar region it may take years to clear
insoluble particles. These long residence times provide a
greater opportunity to generate health problems even if toxic
materials are not present. Both of these concerns have led
EPA's Office of Air Quality Planning and Standards to recommend
that a NAAQS be established for particulate matter with
diameters of 10 micrometers or less.[6] Therefore, the
particles in the ambient air which are associated with the
effects of concern have two general characteristics: 1)
chemical constituents that are soluble in body fluids, and 2)
diameters of 10 micrometers or less.
These general characteristics of typical particulate
matter that cause adverse non-cancer health effects are
important later in this analysis, since the greater the
similarity between this particulate matter and diesel
particulate, the stronger the inference that diesel particles
can also cause adverse health effects. The key points to
remember are:
1. Deposition in the respiratory system is
particle-size dependent,
2. Smaller particles with diameters less than about
10-15 micrometers are transported into the deepest portions of
the respiratory system (tracheobronchial or alveolar regions)
where they reside for long periods of time (hours to years), and
3. Within this subset of inhalable particles, some
particles are deposited in greater amounts depending on their
diameter and the heights and breadths of .these peaks are
dependent on the mode of breathing (mouth versus nose).
B. Effects of Particulate Deposition in the Lower
Respiratory Tract
As stated in the previous section, the deposition of
inhalable particulate in the tracheobronchial and alveolar
regions of the respiratory tract pose the greatest threat to
health. The effects of concern in the tracheobronchial region
include:
-------
6-8
1. Reduced lung function,
2. Aggravation of existing respiratory disease
(especially for bronchitics and asthmatics),
3. Increased infectious disease, and
4. Predisposition to the development of bronchitis.[6]
In the alveolar region the effects of concern include:
1. Reduced lung function,
2. Damage to lung tissues,
3. Increased susceptibility to infection, and
4. Aggravation or predisposition to cardiopulmonary
diseases.[6]
These effects have been observed to varying degrees in
laboratory and epidemiological studies. Because of individual
variation and limitations in analytical methodologies, it is
difficult to tell at what particulate concentrations these
effects begin or become significant. Presently, many of these
effects do not appear to have clear thresholds. [6]
The exact causes of many of the above non-cancer health
effects are not well known, but the following mechanisms or
responses are generally involved either singly or in
combination: [5,6]
1. Macrophage damage due to physical overloading with
particles or because of a toxic response to chemicals adsorbed
on particles;
2. Excess mucus secretion causing a reduction in the
flow rate of the mucociliary conveyor;
3. Structural changes in the lung tissue due to
physically or chemically induced damage;
4. Deposition of particles in excess of the lung's
clearance ability with an attendant build-up of particles; and
5. Bronchioconstriction of airways due to the
stimulation of nerves in the tracheobronchial region.
While the health effects listed above are a step closer to
overall human health than the lung functions (mechanisms) just
described, it is the list of mechanisms which will be most
useful below in assessing the relative potency of diesel
particulate. There are simply not enough data on the effect of
-------
6-9
diesel particulates on the types of health effects listed
above. While the amount of available data on the effect of
diesel particulate on lung function is also less than
desireable, it is greater than that on health effects and will
provide the basis for comparison below.
III. Non-Cancer Health Effects of Diesel Particulate Matter
A. Inhalation of Diesel Particulate
Concerns regarding the health effects of ambient exposures
to diesel particulate were first based on its physical and
chemical characteristics. The particulate matter from diesel
engines is composed of basic units which are 0.1 micrometer or
less in diameter.[8] These units form agglomerates with
diameters ranging up to a maximum of about 1 micrometer. Most
of the agglomerates, however, are significantly smaller than 1
micrometer in diameter (90 percent by mass) , with about 50
percent by mass being 0.3 micrometer or less.[8,9,10] The
small size of diesel particulate means that it is deposited in
the lower respiratory tract, where clearance may take years.
Also important is the fact that the basic particulate unit
is composed of a carbonaceous core with a wide variety of
organic compounds adsorbed onto its surface. While at least
one study specifically identified 70 organic compounds
associated with diesel particulate, [8] the great majority of
the individual compounds remains unknown. Such chemical
constituents could react locally with the cells or tissues of
the lung, or be transported to other areas of the body.
These are the same general characteristics that were
identified above for typical inhalable particles. Therefore,
based solely on the inhalation characteristics of diesel
particulate, it is logical to expect that exposure to diesel
particulate could cause the same adverse non-cancer health
effects as other inhalable particulate.
B. Effect of Diesel Particulate Deposition in the Lower
Respiratory Tract
This inhalation-based . connection between inhalable
particulate and diesel particulate has fostered research
specifically aimed at understanding the non-cancer health
effects of exposure to diesel particles. The results of this
research can be used to resolve two issues which are of
paramount concern. First, does diesel particulate actually
elicit the same adverse effects or responses that were
described above for inhalable particulate in general, as would
be expected based on the similarities between the particles?
Second, is exposure to diesel particulate disproportionately
more hazardous than would be suggested by its contribution to
the concentration of inhalable particulate suspended in the
-------
6-10
ambient atmosphere because of its deep lung deposition and
adsorbed chemicals? More specifically, is the potency of
diesel particulate and the mixture of inhalable particulate in
the ambient air significantly different, so that any increase
in diesel particulate beyond current levels would be especially
hazardous? These two questions are discussed separately
because one issue can be resolved more conclusively than the
other at this time; the first question in this section and the
second in the next.
Two types of studies which specifically deal with diesel
emissions are most useful in answering either of these
questions: epidemiological and laboratory. Before the
findings of these studies are presented, it should be noted
that most of this research has already been compiled or
reviewed in References 3, 11, and 12. Because of this, only a
brief overview of the literature will be presented here.
The epidemiological research into the non-cancer health
effects of diesel particulate exposure is extremely limited.
There are no studies which specifically evaluate diesel
particulate. Only a very few studies evaluate diesel exhaust,
and diesel particulate by association. The primary reason for
this is the lack of suitable populations available for study.[1]
Some of the studies that have been completed suggest that
occupational exposure to diesel exhaust (e.g., railroad,
transit, mining workers) results in a higher prevalence of
chronic respiratory symptoms, bronchitis, and loss of lung
function. [3] Other studies have shown no significant adverse
effects between groups of exposed and unexposed
individuals. [3] Therefore, although the available
epidemiological studies suggest that chronic exposure to diesel
exhaust, including diesel particulate, may adversely affect
health, the results are inconclusive. Because of this, no firm
conclusion regarding the health effects of diesel particulate
can be made based on this type of information. Thus, the
results of laboratory studies must be examined to better
determine the effects of diesel particulate exposure.
Most laboratory investigations of diesel particulate
exposure have been conducted at higher particulate
concentrations than normally would be encountered in the
natural environment. This is common practice in such studies
and is done to reduce the cost of such research. Because of
the high exposures used in these studies, they are very useful
in identifying the mechanisms or responses that would account
for the effects of concern that are observed in the "real
world" (e.g., bronchitis and infectious disease). However,
they are less useful for identifying health effects that will
occur at realistic exposure levels.
Most of the laboratory studies involving diesel
particulate have shown, to varying degrees, the same basic
-------
6-11
effects on lung function that were previously described for
inhalable particulate matter, including alveolar macrophage
damage, excess secretion of mucus, lung tissue damage, possible
adverse effects on the immune system, and particle build-up in
the lung.[3,11,12] This similarity of response provides strong
evidence that exposure to diesel particulate has the potential
to elicit many of the same adverse health effects which were
also previously described for inhalable particulate in
general. Therefore, the original concerns regarding diesel
particulate that were based simply on its inhalation
characteristics are supported by more recent direct evidence.
C. The Hazard of Diesel Particulate Relative to General
Inhalable Particulate
The issue of diesel particulate1s relative hazard is a
more difficult issue to resolve. As discussed above, the few
qualtitative epidemiological studies are not useful to
characterize the non-cancer health effects of diesel
particulates because their results are inconclusive. Also, the
use of very high particulate concentrations in the laboratory
studies generally precludes using this research to evaluate the
health risk of ambient exposures to diesel particles in
comparison to that associated with the ambient mixture of
particles. Nevertheless, some studies have investigated the
systemic toxicology of diesel exhaust. Such studies are
particularly useful in evaluating the concern that the organic
chemicals adsorbed on the surface of diesel particulate may
make it disproportionately more hazardous than other
particulate in the ambient mixture.
Generally, the results of these studies have not
demonstrated any significant gross toxicological effects from
exposure to diesel particulate.[8] A possible explanation for
this lack of effect is that other research has suggested that
although the organic layer of diesel particulate is soluble in
body fluids, it may be released very slowly and that enzyme
systems in the lungs may metabolize these chemical
consitituents into more innocuous substances.[1] Therefore,
this information suggests that the organic layer of diesel
particulate may not cause significant non-carcinogenic
toxicological effects.
Information concerning the efficiency with which particles
of various sizes are deposited in the lower respiratory tract
may also provide some insight into the relative hazard of
diesel particulate. It was previously stated that almost all
diesel particulate is smaller than 1 micrometer in diameter.
Figure 6-2 shows, for example, that the deposition for these
sized particles in the alveolar region during mouth breathing
is substantially less than for particles with diameters of 1-6
micrometers. (The effect is present, though less dramatic, for
nose breathing.) Therefore, particles in the ambient air which
-------
6-12
are somewhat larger than diesel particulate may pose a somewhat
greater health hazard on the basis of mass deposited in the
lower respiratory tract. This suggests that, on a mass
concentration basis, diesel particulate may not be more
hazardous than would be accounted for by its contribution to
the total ambient mixture of inhalable particulates, and that
it could be somewhat less hazardous than certain larger, though
still inhalable, particulate.
IV. Summary
Based on the available health effects and deposition
studies, there is no direct evidence that diesel particulate is
disproportionately more potent in causing non-cancer health
effects than an equivalent mass of the current ambient mixture
of particles. However, this information is so limited that it
does not provide a sufficient basis for conclusively
eliminating the concern that diesel particulate may be more
hazardous because of its chemical composition and deep lung
deposition. Therefore, the issue of diesel particulate's
relative hazard cannot be fully resolved at this time. Ongoing
research may shed more light on this issue in the future.
The following overall conclusions regarding the non-cancer
health effects of diesel particulate are possible, based on the
information summarized above.
1. Laboratory studies have shown diesel particulate
matter has the potential to cause or contribute to adverse
health effects such as reduced lung function, damage to lung
tissues, increased suceptibility to infection, aggravation of
existing respiratory disease, predisposition to bronchitis, and
aggravation of or predisposition to cardiopulmonary disease.
2. There is insufficient evidence to conclusively judge
whether diesel particulate is or is not more hazardous than the
mixture of various particles suspended in the ambient air with
diameters of 10 micrometers or less (i.e., inhalable
particulate). However, the very limited information from
health effects and deposition studies suggests that diesel
particulate may not be more hazardous.
-------
6-13
References
1. "Inhalation Toxicology of Diesel Exhaust Particles,"
In: Diesel Emissions Symposium Proceedings, McClellan, R., A.
Brooks, R. Cuddihy, R. Jones, J. Mauderly, and R. Wolff U.S.
EPA, ORD, 1981.
2. "A Subchronic Study of the Effects of Exposure of
Three Species of Rodents to Diesel Exhaust," In: Diesel
Emissions Symposium Proceedings, Kaplan, H. L. , W. F.
Mackenzie, K. J. Springer, R. M. Schreck, and J. J. Vostal,
U.S. EPA, ORD, 1981.
3. "Impacts of Diesel-Powered Light-Duty Vehicles:
Health Effects of Exposure to Diesel Exhaust," National
Research Council, 1981.
4. "Size Considerations for Establishing a Standard for
Inhalable Particles," Journal o_f the Air Pollution Control
Association, Miller, F., E. Gradner, J. Graham, R. Lee, Jr., W.
Wilson, and J. Bachman, 1979, Vol. 46, pp. 610-615.
5. "Air Quality Criteria for Particulate Matter and
Sulfur Oxides (Draft)," U.S. EPA, OAQPS, December 1981.
6. "Review of the National Ambient Air Quality
Standards for Particulate Matter: Assessment of Scientific and
Technical Information," U.S. EPA, OAQPS, January 1982.
7." American Industrial Hygiene Association," In:
Diesel Emissions Symposium Proceedings, Chan, T. and M.
Lippman, Vol. 41, pp. 399-409, 1980.
8. "EPA Studies on the Toxicological Effects of Inhaled
Diesel Engine Emissions," In: Diesel Emissions Symposium
Proceedings, Pepelko, W., U.S. EPA, ORD, 1981.
9. "Characterization of Particulate and Gaseous
Emissions from Two Diesel Automobiles as Functions of Fuel and
Driving Cycle," Hare, C. and T. Baines, SAE Paper No. 790424.
10. "Characteristics and Oxidation of Diesel
Particulate," In: Diesel Emissions Symposium Proceedings,
Trayser, D.A., L. J. Hillenbrand, U.S. EPA, ORD, 1981.
11. Diesel Emissions Symposium Proceedings, U.S. EPA,
ORD, 1981.
12. "Health Effects of Diesel Engine Emissions:
Proceedings of an Interational Symposium," Vol. 1 and 2,
Edited by Pepelko, W., R. Danner, and N. Clark, U.S. EPA, ORD,
December 1979.
-------
CHAPTER 7
SOILING EFFECTS
I. Introduction
With the increased use of diesel-powered vehicles, the
impact of diesel particulate emissions on material damage has
become a subject for investigation. The major type of material
damage associated with chemically non-reactive atmospheric
particles, such as diesel particulate, is that of soiling.[4]
This chapter will examine the effects of diesel particulate on
soiling.
In the past, the vast majority of soiling studies have
dealt with general atmospheric particulate, while little work
has been done specifically on the soiling impact of diesel
particulate. However, by considering the relative
characteristics of diesel particulate, it is possible to adapt
the findings of studies addressing atmospheric particulate
soiling to diesel particulate soiling.
The soiling damage caused by increased ambient levels of
diesel particulate can be addressed in a number of ways. One
approach would be to derive three relationships: 1) a
relationship between ambient particulate levels and the
physical phenomena of soiling (i.e., particle deposition), 2) a
relationship between soiling and cleaning frequency, and 3) a
relationship between cleaning frequency and cleaning costs. By
combining the three, a relationship between ambient particulate
levels and the cost associated with removing the soiling can be
obtained. However, with this approach intermediate
relationships are also determinable (i.e., the relationship
between particulate levels and cleaning frequency). A second
approach would be to derive a single relationship between
ambient particulate levels and the cost of soiling. This
latter methodology usually utilizes surveys of individuals'
intentions or actions to determine a "willingness-to-pay"
associated with a decrease in soiling. While the former
methodology can also utilize surveys, it is more subjectable to
scientific experimental study.
This analysis will not address any economic costs
associated with soiling due to the controversy connected with
the existing economic soiling analyses. Instead, this analysis
will only address the practical aspects of soiling (i.e.,
soiling as a function of particulate concentration and cleaning
(or other soiling remedy) frequency as a function of soiling) .
This restriction in scope has an unfortunate side effect
of placing the great majority of the research addressing
-------
7-2
atmospheric particulate soiling outside the scope of this
study. The remaining research primarily addresses the effect
of total suspended particulate (TSP) on soiling, with little
having been done on the effect of soiling on cleaning frequency
or on the soiling effects of various subclasses of TSP. No
experimental research has been conducted on soiling by diesel
particulate.
Given this, this study will take a three-step approach to
address the issue of diesel particulate soiling. First, the
physical process of soiling will be defined and described.
Second, studies addressing soiling by TSP will be reviewed to
assess the current state of knowledge in the area. Third,
soiling by diesel particulate will be compared to that by TSP
by comparing the physical and chemical properties of both types
of particulate and postulating their effect on soiling. The
goal of the entire process will be to arrive at some relative
value for the soiling effect of ambient diesel particulate to
that of TSP.
II. Description of the Soiling Process
Soiling is defined as the build-up of a layer of deposited
atmospheric particulates on an exposed surface.[1] A soiled
surface appears dirty to the eye and, as the layer of deposited
particulates increases, it will become detectable by touch.
Characteristics associated with soiling are: 1) a loss of
reflectance of visual light by an opaque material surface, or
2) a reduction in light transmission through a transparent
material.
The time interval required to transform horizontal and
vertical surfaces from a clean to a perceptibly dirty state is
generally determined by particle composition and the rate of
deposition. This process is also influenced by the location
and spatial alignment of the material, the texture and color of
the surface relative to the particle, and meteorological
variables like moisture, temperature and wind speed.[2]
The degree of soiling is determined by measuring
reflectance from an opaque surface and by measuring haze
through a transparent surface (window glass is the most common
transparent surface). The greater the original reflectance of
the surface, the more observable the soiling will be. [3] This
can easily be seen by imagining the effects of soiling on a
white-painted surface, which has a reflectance of more than 90
percent, as compared to the effects of soiling on a
black-painted surface, with a much lower reflectance.
-------
7-3
III. Atmospheric Particulate Soiling
A small number of studies have been performed relating TSP
levels to the physical rate of soiling. This section will
briefly review four such studies. The first two studies were
experimental in nature and simply attempted to determine the
relationship between particulate concentration, time, and
soiling. The third study used surveys and attempted to go one
step further by relating particle concentration to the
frequency of soiled removal (in this case, painting). The
fourth study, a literature review, identified those cleaning
tasks that would be affected by increased soiling resulting
from increased ambient particulate levels.
In the first study, Barker attempted to determine the
relationship between changes in the reflectance of a surface
and the accumulation of particles.fi] Reflectance changes of
white painted surfaces showed a first order dependence upon
total pollutant dosage as defined by the expression:
R = Rp + (R0 - Rp) exp (-KCt)
Where:
R = reflectance of the surface,
Ro = initial reflectance of the surface,
Rp = reflectance of particles,
K = deposition rate constant,
C = particle concentration,
t = exposure time.
It is interesting to note that, if soiling is defined as
the change in surface reflectance (Ro - R) rather than simply
the surface reflectance (R), the above equation becomes:
R0 - R = (R0 - Rp)(1 - exp (-KCt))
This is the equation for exponential decay, which, among
other processes, describes the decay of radioactive materials.
The change in reflectance is rapid at first and slows as time
goes on. The final reflectance of the surface approaches the
reflectance of the particulate assymptotically (i.e., very
gradually). A doubling of the particle concentration would not
affect the final reflectance of the surface, but would double
the rate of soiling. This is shown in Figure 7-1.
In a second, similar study, Beloin and Haynie exposed six
materials to particulate soiling.[4] A linear regression
analysis resulted in the following relationships for two of the
materials:
-------
Figure 7-1
Effect of Particle Concentration on Soiling Using Barker's Model
I . B0
a. 90
0.10-
0.00!
K = 0.1 month per ug/nT
R = 0
C = 1 ug/nf
o
0.0 a.a
e.0 s.0 12.0 is.0
Time (months)
1B.0 21.0 2H.0
-------
7-5
1. For acrylic paint:
R0 - R = 92.5 - R = 1.36C-345t-612
2. For white asphalt shingles:
R0 - R = 41.8 - R = .0078C1-°°7t.595
where the units of C and t are in micrograms per cubic meter
and months, respectively.
Here, soiling (R - Ro) is dependent on certain powers of
both particle concentration and time. While these
relationships appear quite different from that put forth by
Barker, they are not entirely inconsistent. First, Beloin and
Haynie were actually addressing a situation quite different
from that addressed by Harker. Beloin and Haynie1s experiments
and their correlations included a variety of particulate types,
all having different properties. Harken's relation only
applies to a single type of particulate. Second, the powers
associated with particle concentration and time in the Beloin
and Haynie equations are all essentially between zero and one,
which is what would be expected if the process described by
Harker was examined for a specific period of time. The fact
that the powers for concentration and time are not equal is
more of a question, as Barker's model implies they should be
the same. However, the fact that Beloin and Haynie included
different types of particulate in their study could be the
explanation.
To illustrate this possibility, a portion of the data from
the Beloin and Haynie study and their equation for acrylic
paint have been reproduced in Figure 2. A specific instance of
Barker's equation was then fit to the data. As can be seen,
the two relations agree very well and both describe the data
adequately. Thus, while the exact form of the relationship
between soiling and particle concentration is not known, it is
clear that atmospheric particulate does result in soiling and
that an increase in particle concentration will increase the
degree of soiling, and very likely to the same degree (i.e., a
doubling of particulate will double the soiling).
A relationship between the frequency of house repainting
and atmospheric particulate concentration was shown in the
third study by Michelson and Tourin. [5] A mailed survey of
households in the upper Ohio River Valley established a linear
relationship between repainting frequency and ambient levels of
particulate matter.
In the fourth and final study, Watson and Jackson examined
the soiling literature to determine which of 27 common
-------
Figure 7-2
Comparison of Two Soiling Models
3S.0r
aa. a
2S.0
20.0
15.0
10.0
S.0
0.0
Beloin and
Haynie
= Actual data, Beloin and Haynie
0.0 3.0 6.0 S.0 12.0 I S.0 IB.0 21.0 2H.0
Time (months)
I
cr\
-------
7-7
household maintenance and cleaning tasks would be affected by
atmospheric particulate soiling. [6] Those tasks for which
there was little or no evidence of being significantly affected
by soiling were eliminated from consideration. The eight
cleaning and maintenance tasks that would be affected by
atmospheric particulate soiling are:
Indoor Outdoor
Painting walls and ceilings Painting walls
Wallpapering Painting trim
Washing walls Washing windows
Washing windows
Cleaning Venetian blinds
No attempt was made, however, to determine the degree of the
effect that soiling had on the frequency of the performance of
these tasks; only that the effect would be significant.
Again, as was the case with the first two studies, the
usefulness of the latter two studies is limited. No
quantitative relationship between atmospheric particle
concentration and cleaning frequency can be drawn. However,
the evidence indicates that not only does suspended particulate
cause soiling, but soiling affects the performance of cleaning
and maintenance tasks. Thus, increased ambient particulate
levels will lead to increased soiling, which will have a cost
associated with its removal.
III. Diesel Particulate Soiling
The previous descriptions of atmospheric particulate
soiling refer to TSP (i.e., less than approximately 30
micrometers in diameter). Diesel particulate falls into a
subclass of TSP (fine particulate, that are less than
approximately 2.5 micrometers in diameter) and both its
physical and chemical characteristics could quite likely cause
it to have soiling properties different than those of TSP.
Unfortunately, there exist very little direct experimental data
demonstrating the relative soiling effect of fine particles or
diesel particulate to those of TSP. Because of this, it is
necessary to compare the characteristics of diesel particulate
and TSP and postulate the effect of the differences on their
overall soiling impact.
The physical and chemical properties of particulate which
most affect the degree of soiling damage appear to be
reflectance, stickiness, and size. Wallin has measured the
-------
7-8
optical reflectance of diesel particulate and found it to be
generally about 3.5 times blacker than average urban
particulate.[7] Thus, the change in reflectance due to
deposition of diesel particulate will be greater than that of
TSP because the difference between the reflectances of the
surface (Ro) and the particulate (Rp) will be greater.
This is caused by the high carbon content of diesel
particulate, which has a reflectance of almost zero. Diesel
particulate also appears to stick to surfaces more than the
average particulate due to its oily nature (i.e., heavy
hydrocarbons bound to the surface).[8]
In a report prepared for the California Air Resources
Board, Sawyer and Pitz defined a "soiling index" as the ratio
of the diesel particulate soiling to average urban particulate
soiling on the basis of equal ambient mass concentrations.[8]
They then went on to estimate the value of this index based on
the relative properties of diesel particulate and TSP.
The effect of different optical properties was taken from
Wallin's study and translated into an initial soiling index of
3 or 4 based on this single parameter. Because no experimental
data are available on the stickiness of diesel particulate in
quantitative terms, Sawyer and Pitz estimated a combined
soiling index of 5 based on the combined effects of both
reflectance and stickiness. To bracket the uncertainty, a
range from 2.5 to 7.5 for diesel soiling indices was
considered. (No effect due to different particle size was
included. While it appears in some cases that small particles
may deposit in greater amounts due to their greater diffusion
capabilities, in other cases larger particles would deposit
faster due to their greater mass. Thus, no clear preference
based on size can be determined.)
As an example of how this soiling index would be used, one
can assume an area with 75 ug/m3 of TSP present. This
concentration would have what could be called a soiling
potential of 75 ug/m3, since TSP is the base particulate
(i.e., a one-to-one correspondence between mass concentration
and soiling potential). If 5 ug/m3 of diesel particulate
were added to this atmosphere, the concentration of TSP would
become 80 ug/m3, an increase of 6.7 percent. However, using
a. soiling, index of 5 for diesel particulate, the soiling
potential with the addition of diesel particulate would be 100
ug/m3 (75 + 5*5) , an increase of 33 percent. Thus, one can
see how adding a given concentration of diesel particulate to
the atmosphere can have a much greater effect on soiling than
would be indicated by its effect on particulate mass
concentration.
-------
7-9
IV. Summary
Very little data are available on the effect of ambient
particulate on the absolute degree of soiling and the frequency
of cleaning. However, it is clear that ambient particulate,
including diesel particulate, does result in soiling and has a
cost associated with its removal. In addition, it appears that
the degree of soiling associated with diesel particulate is
greater than that of TSP on a mass concentration basis;
possibly between 2.5 and 7.5 times as great. Thus, when
relating the soiling effects of specific ambient concentrations
of diesel particulate to those of TSP, the concentrations of
diesel particulate should be increased by a factor
substantially greater than 1 to place them in the proper
perspective with the TSP concentrations.
-------
7-10
References
1. "Particulate Matter Soiling of Materials," Barker,
Final Report Draft for U.S. EPA, EPA Contract No. 68-02-3422.
2. "Review of the National Ambient Air Quality
Standards for Particulate Matter: Assessment of Scientific and
Technical Information," U.S. EPA, OAQPS, January 1982.
3. "Effects of Small Particles on Materials," Haynie,
U.S. EPA, Environmental Sciences Research Laboratory.
4. "Soiling of Building Materials," Journal of the Air
Pollution Control Association, Beloin and Haynie, 1975.
5. "Report on Study of Validity of Extension of
Economic Effects of Air Pollution Damage from Upper Ohio River
Valley to Washington, D.C.," Michelson and Tourin, Area
Environmental Health and Safety Research Association, August
1967.
6. "Air Pollution: Household Soiling and Consumer
Welfare Losses," Journal of_ Environmental Economics and
Management, Watson and Jaksch, 1981.
7. "Calibration of the D.S.I.R. Standard Smoke Filter
for Diesel Smoke," International Journal of Air and Water
Pollution, Wallin, Vol. 9, 1965.
8. "Assessment of the Impact of Light-Duty Diesel
Vehicles on Soiling in California," Sawyer and Pitz, Prepared
for the California Air Resources Board, January 1983.
-------
CHAPTER 8
ECONOMIC IMPACT
I. Introduction
A. Organization of Chapter
This chapter addresses the economic impact of the base
scenario relative to the relaxed scenario. (Full descriptions
of each scenario are given in Chapter 1.) The two basic
trap-oxidizer designs and the associated regeneration systems
are described in the remainder of the introduction.
The next two sections of this chapter examine the economic
impact of particulate control on light-duty diesel vehicles
(LDDVs) and trucks (LDDTs), and on heavy-duty diesel engines
(HDDEs). The subsections in each section deal, in order, with
estimating the cost of the hardware requirements for
particulate control, examining the economic impact on affected
vehicle and engine manufacturers, estimating the overall cost
to the consumer of particulate control, and estimating the
annual costs (for the years 1987 through 1995) and the 5-year
aggregate costs (1987 through 1991 inclusive) of these controls.
B. Description of Trap Designs
The primary component of any system for the reduction of
diesel particulate emissions is the trap-oxidizer. In addition
to the trap itself, other hardware components are required,
with the specific requirements depending on the basic
trap-oxidizer design used. Trap-oxidizers (traps) can be
broadly divided into categories on the basis of two factors:
location or placement; and filter material.
An underfloor-mounted trap occupies approximately the same
position, relative to the diesel engine, as is occupied by a
catalytic converter on a gasoline-fueled vehicle. A
close-coupled trap is located nearer to the engine, and is
usually incorporated in the exhaust manifold design. Traps are
also catalyzed or non-catalyzed, according to the presence or
absence of catalytic materials to aid in the oxidation of
accumulated particulate.
Detailed descriptions of the design and operation of each
type of trap can be found in the EPA Trap-Oxidizer Feasibility
Study.[1] For this economic analysis, the costs and economic
impact are based only on the underfloor-mounted design, since
it appears to be the preferred design of many trap-oxidizer and
diesel vehicle/engine manufacturers. The possibility of
-------
8-2
close-coupled traps being used is addressed in Chapter 10. No
clear preference for one of the two major filter materials has
yet emerged; a brief description of each follows.
Although many filter materials have been investigated for
use in traps, the current focus of development and testing is
on ceramics and alumina-coated wire mesh. Ceramic traps
utilize a non-catalytic, porous cordierite material [2(MgO) +
2(Al203) + 5(Si02)] for the substrate. This substrate is
similar in construction to the support structure used for
catalytic converters in gasoline-fueled applications, typically
consisting of a honeycomb design with parallel square channels
running the length of the unit. This trap design is being
manufactured by Corning, NGK, and other firms.
Johnson-Matthey is the primary manufacturer of traps using
alumina-coated wire mesh as the filter material. The form of
the wire mesh trap is a long cylinder with a hollow central
core. The exhaust flows radially through the mesh filter from
the outside toward the hollow core. Catalytic coating of the
wire mesh, lowering the temperature necessary for trap
regeneration (oxidation of the accumulated particulate
collected by the filter), is inherent in the Johnson-Matthey
design.
Both types of trap are enclosed by a stainless steel
shell, basically the same as that used for the exterior shell
of a catalytic converter.
C. Description of Regeneration Systems
In addition, each type of trap requires a method of
regeneration. Since excess accumulated particulate increases
exhaust backpressure (thereby decreasing fuel economy and
vehicle performance), it must be oxidized or burned off
periodically. The temperature of the diesel exhaust stream is
typically inadequate to initiate or sustain this oxidation.
Therefore, a regeneration system is also required for effective
nar t- i rnl ahe control.
particulate control.
The hardware components required for an effective
regeneration system depend, in part, on whether the trap is
catalyzed or non-catalyzed. The presence of catalytic material
in the trap filter reduces the temperature increase needed for
particulate oxidation, allowing the use of a less complex
regeneration system than is required for non-catalyzed traps.
Each of these is briefly described below; detailed explanations
of the structure and functioning of trap regeneration systems
are available elsewhere.[1,2]
-------
8-3
A typical regeneration system for a non-catalyzed trap is
based on a diesel fuel burner, which injects diesel fuel into
the exhaust stream just before this stream enters the trap.
Burning the added fuel increases the exhaust temperature enough
to ignite the accumulated particulate. The engine exhaust flow
is temporarily routed around the trap, while the burner and
trap are supplied with a controlled air flow to ensure
continued oxidation of the trapped particulate without
excessive heating.
This typical regeneration system has seven primary
hardware components: a burner head, a fuel delivery system, an
ignition system, an auxiliary combustion air system, an exhaust
diversion system, system control sensors, and an electronic
control unit (ECU).
The burner head provides a location for mounting the fuel
spray nozzle, ignition plug, and auxiliary air nozzle. It is
also assumed to include a gas distribution baffle for evenly
distributing the combustion products over the cross-section of
the trap.
The fuel delivery system provides the diesel fuel
necessary for initiating the trap regeneration process. This
system includes a fuel spray nozzle, a fuel feed line, and a
fuel solenoid valve.
The fuel ignition system may be one of two basic types.
One system consists of a long-life spark plug, a step-up
voltage transformer, and signal conditioning electronics for
generating a high-voltage discharge. An alternative to this
system is a glow plug, like those used to cold-start diesel
engines.
The auxiliary air combustion system, which provides a
controlled air supply to the burner and trap to sustain the
oxidation of the accumulated particulate, consists of an air
pump, a check-valve, a diverter valve, and an air delivery
line. The check-valve prevents exhaust backflow into the air
pump, while the diverter valve provides an alternate path in
the event that combustion air must be diverted from the
filter. The air delivery line connects the air pump to the
burner head.
The exhaust diversion system temporarily reroutes the
engine exhaust stream around the trap during the regeneration
process. It consists of a vacuum motor driven by the ECU and
an engine-driven vacuum pump, which generates the vacuum
required for operation of various control elements. An
alternative to this system, not requiring a vacuum pump, is a
solenoid valve operator.
-------
8-4
System control sensor requirements include two temperature
sensors, and a control sensor for determining the need for trap
regeneration. Temperature sensors are required for detecting
overheating in the trap filter during the regeneration, and for
ensuring that the engine has attained normal operating
temperature before the regeneration is initiated. The sensor
determining the need for trap regeneration could be either an
engine revolution or vehicle mileage timer, or an exhaust
backpressure sensor.
The most important regeneration system component, in terms
of system control, is the ECU. The ECU interprets signals
received from the various sensors in order to maintain control
of the regeneration process.
As was noted earlier, the regeneration system for a
catalyzed trap can be less complex, since the increase in the
exhaust stream temperature required is much smaller. Of the
seven primary hardware components required for the
non-catalyzed trap regeneration system described above, only
the system control sensors and the ECU are needed in basically
the same form for the catalyzed trap regeneration system.
Some type of auxiliary air combustion system is still
required; [1] however, since the exhaust flow through the
catalyzed trap is maintained during the regeneration process, a
reed valve system may be adequate. The burner head, exhaust
diversion system, fuel delivery system, and ignition system
described above are not required.
However, an alternate system for providing a moderate rise
in the temperature of the exhaust stream is still required.
One such method, which has been successfully tested on a
Volkswagen Rabbit in conjunction with the Johnson-Matthey
catalyzed wire-mesh trap, is known as delayed in-cylinder fuel
injection. A small amount of fuel is injected into the
cylinder during the exhaust stroke, when the cylinder is too
cool to ignite the fuel. The injected fuel is carried in the
exhaust stream to the catalyzed trap, where it is ignited.
Since the existing fuel system is used to inject the fuel, the
only hardware necessary is a mechanism for transferring a
portion of the fuel being metered from a "normal" injector to
the "delay" injector.
Though actually not part of the trap or of the
regeneration system, one other vehicle modification affecting
the exhaust system should be discussed here. The exhaust pipe,
leading from the engine to the trap-oxidizer, will have to be
fabricated of stainless steel. If fabrication of this pipe
-------
8-5
using normal steel were continued, periodic replacement would
be required, greatly increasing the chances of the
trap-oxidizer being removed from the vehicle. This
modification is required for all of the trap designs and
regeneration systems discussed.
II. Light-Duty Diesels
A. Introduction
This section examines the impact of particulate control
for LDDVs and LDDTs. Since the methodology and many of the
basic assumptions used in this analysis are the same for both
light duty and heavy duty, this section contains considerably
more detail than does the next section on heavy-duty diesels.
The next subsection estimates the costs, in terms of the
retail price equivalent (RPE), of each of the basic trap
designs and regeneration systems described in the
introduction. These costs are largely a function of the size
(volume) of the trap-oxidizer. This discussion is followed by
subsections treating the economic impact on diesel
manufacturers, the overall cost to the consumer, and the annual
and 5-year aggregate costs of these controls.
After the costs of the hardware (trap-oxidizer and
regeneration system) are estimated, the subsequent analysis
examines the economic impact under two regulatory scenarios
(base and relaxed) , and under two sets of future diesel sales
projections ("best estimate" and "worst case"). The regulatory
scenarios are described in detail in Chapter 1. The best
estimate sales projections [3] are exactly what the designation
implies, while the "worst-case" sales projections are based on
the maximum increases in diesel sales that appear to be
reasonable. (The term "worst case" refers to the impact of
increased diesel sales on total particulate emissions, and the
resulting environmental effects.)
The cost of the two basic trap-oxidizers, catalyzed and
non-catalyzed, were previously estimated in the Regulatory
Analysis that accompanied the original light-duty particulate
control regulations. [4] The model used to estimate the
manufacturing costs of each trap design, which was developed by
Lindgren, [5] is again used in this analysis, with cost
estimates provided by the trap manufacturers incorporated into
the model where available. The Lindgren model for estimating
the RPE of manufacturing costs [5] is based on the application
of adjustment factors to the estimated sum of direct material
and labor, and fixed overhead costs. These factors are
expressed as 1.0 plus the sum of the adjustment terms, as shown
below:
-------
8-6
RPE = [ (DM+DL+OH) (1+CA+SP)+TE+LBE] (1+CA+CP+DP) +RD+TE (1)
Where:
DM = Direct material cost.
DL = Direct labor cost.
OH = Fixed and variable overhead.
CA = Corporate allocation term of adjustment factor.
SP = Supplier profit term of adjustment factor.
TE = Tooling expense.
LBE = Land and building expense.
CP = Corporate profit term of adjustment factor.
DP = Dealer overhead and profit term of adjustment factor.
RD = Research and development cost.
Some of the values used in equation 1 were taken directly
from Lindgren's work,[5] while others have been adjusted based
on more recent analyses. [4] Additional adjustment factors for
inflation and production volume (i.e., economy of scale) are
also incorporated in this analysis. These are described in
more detail below.
Regeneration system costs have also been estimated in the
past. [4] In this analysis, these earlier estimates are
essentially supplanted by more recent work performed by Mueller
Associates[2] under an EPA contract.
B. Trap-Oxidizer System Costs
1. Introduction and Assumptions
The adjustment factors for inflation and production volume
are independent of the trap design or regeneration system
used. Therefore, these are discussed first, before estimating
the specific manufacturing costs for each case.
Some of the manufacturing cost data that went into the
development of Lindgren's model and into the previous EPA
analyses dates from as early as 1978. Therefore, an adjustment
factor for inflation must be determined. For application to
particulate control hardware, the increase in LDV (new car)
prices from 1978 through 1982 appears to be a more appropriate
estimate of inflation than the rise in the Consumer Price Index
over the same time span. New car prices were 33 percent higher
in 1982 than in 1978, with annual increases of 6.2, 7.4, 7.5,
6.8, and 1.6 percent in 1978, 1979, 1980, 1981, and 1982,
respectively.[6] These inflation rates are used in this
analysis.
-------
8-7
Production volumes of different traps by various trap
manufacturers are uncertain. The assumption in this analysis
is that two manufacturers will supply trap-oxidizers, and that
each of them will supply approximately half of total production.
It is also necessary to distinguish between different
sizes of LDDVs and LDDTs, since the size (volume) of the trap
is dependent on the engine size (displacement). In this
analysis, LDDVs are divided into small, medium, or large on the
basis of engine displacement. Small LDDVs are those equipped
with 1.6- to 1.8-liter (L) engines, medium LDDVs are equipped
with 2.0L to 2.8L engines, and large LDDVs are those with 3.0L
and larger engines. It is assumed that the projected LDDV
sales will be divided approximately equally among these three
size classes. The LDDTs are considered to be either small
(engines under 4.3L) or full-size (4.3L and larger engines).
Full-size LDDTs are assumed to sell at a 4:1 ratio to small
LDDTs.
Based on the best estimate LDDV sales projections [3] and
the projected rates of trap usage (see Chapter 1) , a standard
average production level of 200,000 traps annually, for each of_
the three LDDV size classes, is a reasonable estimate.
Projected average annual sales of small LDDTs are
approximately half those projected for each LDDV size class
(see Chapter 1) . Thus the standard average trap production
level for small LDDTs is estimated at 100,000 annually, or half
of the standard production level for each LDDV size class.
Full-size LDDT sales are projected to be roughly twice those
for each LDDV size class, therefore, the standard average
production level of traps for full-size LDDTs is estimated to
be 400,000.
In order to develop adjustment factors based on the
standard average production levels of traps for each of the
five size classes of light-duty diesels, the "learning curve"
must be known. For trap-oxidizer production, the learning
curve is assumed to be 12 percent. [4] The learning curve
concept is applied to the production levels by first assuming
that some standard average production level serves as a
baseline, for which the production level adjustment factor is
equal to one. (i.e., no adjustment). Under a 12 percent
learning curve, doubling the baseline production level leads to
a 12 percent decrease in per-unit manufacturing costs, or an
adjustment factor of 0.88. Conversely, halving the baseline
production level leads to a 13.6 percent increase in per-unit
manufacturing costs, expressed as an adjustment factor of 1.136
(1.0/0.88).
-------
8-8
Application of the learning curve to the production of
trap-oxidizers is done by assuming that the baseline standard
average production level is 200,000 traps annually, the level
estimated for each of the three LDDV size classes. Therefore,
no adjustment factor for the level of production is applied to
the 200,000-trap annual production level assumed for each LDDV
size class. The adjustment factor for full-size LDDTs is 0.88,
representing a 12 percent decrease in per-unit manufacturing
costs resulting from a doubling of LDDV trap production. For
small LDDTs the production level adjustment factor is 1.136,
reflecting the production level of small LDDT traps being half
that of each LDDV size class.
The production volumes of traps for each size-based class
of light-duty diesels are all based on the best estimate sales
projections. Under the "worst-case" projections, LDDV
production would double and LDDT production would increase 50
percent over the best estimate projections. The effect would
be to lower per-unit trap manufacturing costs by 12 percent for
LDDVs and by 6.2 percent for LDDTs.
With the information given above, the discussion can now
be focused on estimating the manufacturing costs of each trap
design and of the regeneration systems.
2. Non-Catalyzed (Corning) Trap
A formula for determining the manufacturing costs of
non-catalyzed traps has been developed and used in previous EPA
analyses.[1,7] This formula was derived by relating the
various trap components to similar or identical components of a
monolithic catalyst for gasoline-fueled engines, for which
confirmed manufacturing costs are already known. After
applying the adjustment factors, including those for inflation
and production volume (based on the best estimate sales
projections) determined in the preceding section, the formulae
are:
For LDDVs:
RPE = $23 + 0.318(V) (2A)
For small LDDTs:
RPE = $26 + 0.356(V) (2B)
For full-size LDDTs:
RPE = $20 + 0.280(V) (2C)
Where:
V = the volume of the trap, in cubic inches.
-------
8-9
As an example of applying these equations, consider the
case of a non-catalyzed trap that was recently tested
successfully, by Southwest Research Institute, on a
Mercedes-Benz 300D. This trap had a volume of 302 cubic inches
(5.66 inch diameter x 12 inch length); substituting 302 for V
in equation 2A (for LDDVs) leads to an estimated manufacturing
cost of about $119.
As mentioned earlier, traps of various sizes (volumes)
will be fitted to different sizes of engines. Trap size can
logically be expected to be a function of volumetric exhaust
flow of the engine.[1] While data on the typical volumetric
exhaust flows of various engines are not readily available,
fuel consumption (the inverse of fuel economy) is an adequate
surrogate measure.[1] The ratios of the fuel consumptions, or
the inverse ratios of the fuel economies, over the FTP (EPA
urban) driving cycle can be used to extrapolate trap volume
requirements for other engine sizes, given a known reference
point: The Mercedes-Benz 300D mentioned above has an EPA city
fuel-economy rating of 26 miles per gallon (mpg).
Projected fuel economies for each of the five size classes
under consideration, in 1990, are given in the table below:
Size Class Engine Projected FE
Small LDDVs 1.6 to 1.8L 51.2 mpg
Medium LDDVs 2.0 to 2.8L 43.9 mpg
Large LDDVs 3.0L and up 37.8 mpg
Small LDDTs under 4.3L 44.8 mpg
Full-size LDDTs 4.3L and up 33.6 mpg
These estimates were derived from fuel economy estimates for
gasoline engines in 1990, [3] with a 25 percent improvement
assumed in diesel engine fuel economy over the corresponding
gasoline engines.
Using the Mercedes 300D (26 mpg fuel economy, 302 cubic
inches trap volume) as the reference point, and applying the
fuel consumption ratios as discussed above, the resulting trap
volume requirements are:
.Size Class Trap Volume
Small LDDVs 153 cubic inches
Medium LDDVs 179 cubic inches
Large LDDVs 208 cubic inches
Small LDDTs 175 cubic inches
Full-size LDDTs 234 cubic inches
-------
8-10
These volumes can be substituted for V in the equations
2A-2C, yielding estimated manufacturing costs of $72, $80, and
$89 for small, medium, and large LDDVs, and $88 and $87 for
small and full-size LDDTs, respectively.
As shown in the table of trap volumes above, the small
LDDT trap is projected to require a volume only 4 cubic inches
less than that of the medium LDDV trap. The medium LDDV trap
is also estimated to cost less to manufacture than the small
LDDT trap, $80 versus $88. Thus it is more economical to
produce one trap, of the size required for medium LDDVs, for
both medium LDDV and small LDDT applications. Combining the
standard average production levels of 200,000 annually for
medium LDDVs and 100,000 annually for small LDDTs into a new
standard average production level of 300,000 traps, the assumed
12 percent learning curve lowers the per-unit cost to $74.
The manufacturing costs presented above are summarized in
Table 8-1. Confidential estimates of manufacturing costs
supplied by Corning, while not firm, indicate that the
estimates shown in Table 8-1 are reasonably accurate.
Under the "worst-case" sales projections, the cost
estimates given above are reduced by 12 percent for LDDVs and
by 7.2 percent for LDDTs. These estimates are also shown in
Table 8-1.
3. Catalyzed (Johnson-Matthey) Trap
In the Regulatory Analysis for the 1985 light-duty diesel
particulate regulations,[4] the cost of a catalyzed trap was
also estimated by relating the components of the trap to
similar or identical components of a monolithic, ceramic
catalytic converter, with washcoat and noble metals included.
A formula was then developed for estimating the manufacturing
cost based on the trap volume.
Johnson-Matthey has since publicly stated that the
manufacturing cost of their catalytic trap substrate, ready for
canning, was $100 (in 1982 dollars) for a trap intended for use
with a 2.0L engine. If the volume of this trap is assumed to
be equal to that of a trap recently tested successfully on a
Volkswagen Rabbit with a 1.6L engine (345 cubic inches), then
the RPE of the manufacturing cost can be determined using the
Johnson-Matthey cost information. The $100-estimated
manufacturing cost must first be inflated to 1983 dollars and
then substituted for the non-catalyzed substrate manufacturing
cost in equations 2A-2C. The effects on the total cost of
canning, corporate overhead and profit, and dealer mark-up are
assumed to be unchanged from the non-catalyzed trap.
-------
8-11
Table 8-1
Light-Duty Trap Costs (1983 dollars)
Best
Vehicle Class
Small LDDVs
Medium LDDVs
Large LDDVs
Small LDDTs
Full-Size LDDTs
Estimate Sales
Projections
Non-Catalyzed Catalyzed
Trap Trap
$72
$74
$89
$74
$87
"Worst-Case" Sales
Vehicle Class
Small LDDVs
Medium LDDVs
Large LDDVs
Small LDDTs
Full-Size LDDTs
$188
$199
$246
$199
$236
Projections
Non-Catalyzed Catalyzed
Trap Trap
$63
$66
$78
$66
$81
$165
$178
,$216
$178
$219
-------
8-12
It is assumed that the fixed costs (i.e., tooling and
machinery, fixed overhead) are the same for both trap types,
meaning that all variable costs can be expressed as a function
of trap volume. Finally, by combining the production of traps
for medium LDDVs and small LDDTs as was discussed in the
preceding section, the following equations result:
For small and large LDDVs:
RPE = $23 + 0.582(V)
For medium LDDVs and small LDDTs:
RPE = $22 + 0.536(V)
For full-size LDDTs:
RPE = $20 + 0.501(V)
(3A)
(3B)
(3C)
Where:
V = the volume of the trap, in cubic inches.
The 345 cubic inch catalyzed trap mentioned above is
estimated to cost about $224, based on equation 3A. Use of the
methodology developed in the original Regulatory Analysis, [4]
with adjustments made for inflation, production volume, and
more recent precious metal costs, yields an estimated cost of
$212 for a 345 cubic inch catalyzed trap. Thus incorporating
the Johnson-Matthey estimate into the equations 2A-2C changes
the estimated overall trap cost by less than 6 percent.
As in the non-catalyzed case, it must be assumed that
traps of different sizes (volumes) will be produced for use
with different engines. The 1990 estimated fuel economies for
light-duty diesels given above are used here, with the
reference point changed to the Volkswagen Rabbit (42 mpg fuel
economy, 345 cubic inch trap volume) . Applying the ratios of
fuel consumption as a surrogate measure of volumetric exhaust
flow, as was done in the non-catalyzed case, yields the
following trap volume requirements:
Size Class
Small LDDVs
Medium LDDVs and
small LDDTs
Large LDDVs
Full-size LDDTs
Trap Volume
283 cubic inches
330 cubic inches
383 cubic inches
431 cubic inches
-------
8-13
Substituting these volume requirements for V in equations
3A-3C gives the RPE of the manufacturing cost. The estimated
costs based on the equations are: $188 for small LDDV traps,
$199 for medium LDDV and small LDDT traps, $246 for large LDDV
traps, and $236 for full-size LDDT traps. These cost estimates
for catalyzed traps are also summarized in Table 8-1.
Equations 3A-3C and the cost estimates above are based on
the best estimate sales projections. The impact of the
"worst-case" sales projections on these estimates is the same
as on the non-catalyzed cost estimates, with the LDDV costs
reduced by 12 percent and the LDDT costs reduced by 6.2 percent
from the figures above. These estimates are also shown in
Table 8-1.
4. Regeneration System Costs
The main components of trap regeneration systems, for both
catalyzed and non-catalyzed trap-oxidizers, were described in
the introduction. The positive regeneration system for
non-catalyzed traps, which actively initiates the burn-off of
the accumulated particulate by injecting ignited diesel fuel
into the exhaust stream, is dealt with first. The costs
estimated for both regeneration systems are largely based on
the analysis performed by Mueller Associates for EPA.[2]
The components of both types of regeneration system are
listed, with the estimated RPE of the manufacturing costs, in
Table 8-2. These estimates are based on the production levels
corresponding to the best estimate sales projections. Since
almost all of the regeneration system components listed in
Table 8-2 are also manufactured for purposes other than
particulate control, the production levels are higher than
those of trap components. On this increased base production
level, the impact of the "worst-case" sales projections is much
smaller. Thus, any changes in these estimates due to increases
in trap-equipped diesel sales are also much smaller, and are
not shown in Table 8-2.
The hardware components for each type of regeneration
system are listed in Table 8-2, and discussed below, in the
same order as they were described in the introduction.
The burner head is assumed to be fabricated of stamped and
welded Type 409 stainless steel, and has an estimated cost of
$7. [2] The fuel delivery system, for supplying the fuel to be
ignited to initiate the regeneration process, has an estimated
cost of $9. [2]
-------
8-14
Table 8-2
Light-Duty Regeneration
System Costs (1983 dollars)
Retail Price
Hardware Item Equivalent
Non-Catalyzed Trap:
Burner Head $7
Fuel Delivery System $9
Fuel Ignition System* $5-26
Auxiliary Combustion Air System $30
Exhaust Diversion System* $11-14
System Control:
Sensors $12
ECU** $10
Subtotal $84-108
Stainless Steel Exhaust Pipe* $16-27
Total System Cost $100-135
Catalyzed Trap:
Delayed In-Cylinder Fuel Injection Mechanism $15
Auxiliary Combustion Air System (Reed Valve) $6
System Control:
Sensors $12
ECU** $10
Subtotal $43
Stainless Steel Exhaust Pipe* $16-27
Total System Cost $59-70
* Explanation of cost ranges appear in the text.
** Derivation of the ECU cost appears in the text.
-------
8-15
Two basic fuel ignition systems were described in the
introduction. The more costly system (long-life spark plug,
step-up voltage transformer, and signal conditioning
electronics) is estimated to cost $26. [2] While the use of a
glow plug is less expensive, with an estimated cost of $5, [2]
it is also less reliable for ignition when the temperature of
the exhaust stream is relatively low. Both of these options
are included in Table 8-2.
The regeneration process requires a controlled supply of
air to the burner and trap to sustain particulate oxidation.
The total cost of the auxiliary air combustion system (pump,
delivery line, and valves) is estimated to be $30.[2]
Exhaust must temporarily be rerouted around the
non-catalyzed trap during regeneration. The exhaust diversion
system for accomplishing this is estimated to cost between $11
and $14. [2] The lower cost is for a system utilizing a vacuum
motor (an engine-driven vacuum pump is assumed to already be
present on the vehicle), while a system using a solenoid valve
operator is represented by the higher cost (no vacuum pump need
be present on the vehicle).
System control requires the use of several sensors. The
estimated costs are $9, for a sensor to detect overtemperature
in the filter during the regeneration, and $1, for a sensor to
ensure that the engine has attained the proper operating
temperature before regeneration is initiated.[2] The sensor
for determining the need for trap regeneration could be either
an engine revolution or vehicle mileage timer, or, an exhaust
backpressure sensor. The cost of the former is negligible, [2]
while the latter is estimated to cost no more than $2. [8]
Since the backpressure sensor is more desirable, however, the
latter estimate is included in Table 8-2.
The critical system control component is the electronic
control unit (ECU). For gasoline-fueled engines, the current
total cost of an ECU is approximately $75. [2] Several factors
make this cost inappropriate for direct use in Table 8-2.
First, the current ECU is typically much more sophisticated
than is needed for regeneration system control. Second, it is
highly probable that ECUs on diesels after 1987 will serve
several purposes in addition to their emission control
functions. (For example, Isuzu's 1983 diesel vehicles contain
an ECU which functions to improve fuel economy and vehicle
performance, as well as to control dashboard lighting and other
miscellaneous devices or "gadgets.") Third, and most
importantly, continuing advances in microprocessor technology
can be expected to further reduce the cost of ECUs in
constant-dollar terms, while simultaneously widening the scope
of potential automotive applications.
-------
8-16
No data are available on manufacturers' plans for the
installation of ECUs to serve functions other than emission
control. A conservative estimate is that half of all LDDVs and
LDDTs will be equipped with ECUs, for reasons other than
emission control, during the period 1987-95. The remaining
half of LDDV/LDDT production would incorporate ECUs in order to
comply with emission control requirements; however, once
incorporated into the vehicle design they would certainly serve
additional valuable functions.
The ECU in the LDDV or LDDT of the future will have four
primary functions: improving fuel economy, improving vehicle
performance, device and "gadget" control, and emission
control. Allocating one-quarter of the total $75 cost to the
emission control aspects of the ECU gives an estimated cost of
approximately $19 due to particulate control.
Assuming that half of the ECUs installed for emission
control will be required solely for particulate control reduces
the fleetwide average per-vehicle cost to $10. If ECUs are
installed in more diesel vehicles than projected for purposes
of NOx control, this estimate may be reduced even further.
This regeneration system has a total estimated RPE of
between $85 and $109, depending mostly on the fuel ignition
system chosen. As discussed in the introduction, a stainless
steel exhaust pipe will also be required for trap-equipped
vehicles. When a credit for the deleted standard steel exhaust
pipe is included, the additional cost of this modification is
estimated as $16 (for small and medium LDDVs and small LDDTs)
to $27 (for large LDDVs and full-size LDDTs). These costs are
also shown in Table 8-2.
The regeneration system for a catalyzed trap should be
less complex than the system required for non-catalyzed traps,
as was explained in the introduction. While detailed cost
estimates such as those given above are not available for this
simpler system, the savings over the "burner system" can be
estimated using the information in Table 8-2.
No burner head assembly is required. The fuel delivery
system is replaced by a mechanism for transferring a small
amount of fuel from the normally-functioning injector to the.
"delay" injector. This mechanism is expected to cost about
$15.[2] The auxiliary air combustion system described above
can be replaced by a reed valve (estimated cost $6) , [2] since
the continued exhaust flow through the catalyzed trap during
regeneration will provide the required suction.
-------
8-17
The sensors and the ECU, required for regeneration system
control, are basically identical for either system. The
stainless steel exhaust pipe is also required for both
systems. The cost estimates for these components of the
catalyzed trap regeneration system are the same as for the
non-catalyzed case. All of this information is also shown in
Table 8-2.
5. Total Trap-Oxidizer System Costs
The total cost of the trap-oxidizer system is the sum of
the costs estimated for the trap and for the regeneration
system. Summaries of these costs under both the best estimate
and the "worst-case" diesel sales projections are shown in
Table 8-3. Since the widths of the ranges in cost are quite
small, relative to the absolute costs, only the midpoints of
the cost ranges are shown in Table 8-3. These "midpoint"
estimates are used throughout the rest of the analysis.
It is clear from Table 8-3 that, despite the savings
associated with the regeneration system, the total cost of the
catalyzed trap-oxidizer system is still estimated to be
substantially more than that of the non-catalyzed system.
Since it is considerably less expensive, and appears to be the
preferred design of most diesel manufacturers, only the cost of
the non-catalyzed trap-oxidizer system is used in the remainder
of this analysis.
C. Economic Impact on Diesel Manufacturers
In this section, the impact of the base scenario on
manufacturers' light-duty diesel sales, capital investments
and cash flow will be analyzed. Only the costs of the
trap-oxidizer system and the associated fuel economy penalty
are considered here. There are no test facility costs
associated with the base scenario. Certification costs were
shown to be negligible in the Regulatory Analysis to the 1985
particulate standards.[4]
1. Impact on Manufacturer's Sales
The impact of the base scenario on light-duty diesel sales
depends primarily on three factors. First is the vehicle price
increase resulting from the additional cost of installing a
trap-oxidizer. Second is the fraction of vehicles requiring
trap-oxidizers, which was determined for each scenario in
Chapter 1. Third is the 2 percent fuel economy penalty
associated with the use of trap-oxidizer technology.[1]
-------
8-18
Table 8-3
Total Light-Duty Trap-Oxidizer
System Costs (1983 Dollars)
Best Estimate Sales Projections
Vehicle Class
Small LDDVs
Medium LDDVs
Large LDDVs
Small LDDTs
Full-Size LDDTs
Non-Catalyzed
Trap
$185
$187
$213
$187
$211
Catalyzed
Trap
$246
$258
$316
$258
$306
"Worst-Case" Sales Projections
Vehicle Class
Small LDDVs
Medium LDDVs
Large LDDVs
Small LDDTs
Full-Size LDDTs
Non-Catalyzed
Trap
$176
$179
$202
$179
$205
Catalyzed
Trap
$224
$237
$286
$237
$289
-------
8-19
The next step is applying these factors to determine a net
impact on future diesel sales. This has already been done for
a number of potential combinations of trap costs and trap usage
rates, in a study performed by Jack Faucett Associates (JFA)
for EPA[3] using consumer information on diesel vehicle
purchases from Chase Econometrics.[9] JFA estimated the impact
on LDDV and LDDT sales assuming trap-oxidizer costs of $300,
$500, and $800, and trap usage rates of 0 percent, 35 percent,
65 percent, and 90 percent. A fuel-economy penalty of 2
percent for vehicles equipped with traps was also
incorporated. It was assumed that the largest diesel vehicles
would be equipped with traps first, the medium-size diesels
next, and the smallest diesels last, until the overall trap
usage rate was met.
To simplify the application of JFA's results, an average
trap-oxidizer cost will be used for each vehicle class (LDDV
and LDDT). The trap-oxidizer costs for each vehicle size will
be weighted by the relative sales of each vehicle size, as
estimated by JFA. [3] This average cost is $213 for LDDVs, and
$219 for LDDTs.
Using the average trap system costs and the trap
penetration rates, the future sales of light-duty diesel
vehicles and trucks can be projected by interpolation of the
JFA estimates. Table 8-4 shows the projected sales for the
"relaxed scenario," where no traps are required on light-duty
diesels, under both best estimate and "worst-case" diesel sales
projections. These figures represent the maximum number of
vehicles projected to be sold, as these vehicles do not bear
the cost of a trap-oxidizer.
Also shown are the effects of the base scenario on
light-duty diesel sales under both the best estimate and
"worst-case" sales projections. As can be seen, the impact of
the base scenario is greatest in the early years (e.g., 1987)
and diminishes with time. In addition, LDDV sales are affected
more than LDDT sales. The largest impact occurs in 1987, when
30,000 LDDV sales are lost (3.4 percent of total LDDV sales
under the best estimate sales projections). By 1995, this loss
diminishes to 25,000 on a much larger sales base (1.8 percent
of total "best estimate" LDDV sales). Losses of LDDT sales are
roughly one-third to one-half as great, in the range of 10,000
to 12,000 units annually.
An underlying assumption of this analysis is that the
manufacturers will pass the total cost of the trap-oxidizer
system on to the consumer. Manufactuers have been selling
diesel vehicles at a premium to consumers willing to pay extra
for ownership of a relatively new and advantageous product.[3]
-------
8-20
Table 8-4
Light-Duty Diesel Sales Projections*
(in thousands)
Sales and Percent Reduction
1987 1990 1995
Best Estimate Sales Projections
Relaxed Scenario
LDDV Sales 912 1,300 1,380
LDDT Sales 714 1,029 If322
Total: LDDVs and LDDTs 1,626 2,329 2,702
Base Case Scenario
LDDV Sales 881 (3.4%)** 1,260 (3.2%) 1,355 (1.8%)
LDDT Sales 704 (1.4%) 1,018 (1.0%) 1,310 (0.9%)
Total: LDDVs and LDDTs 1,585 (2.5%) 2,278 (2.2%) 2,665 (1.3%)
"Worst-Case" Sales Projections
Relaxed Scenario
LDDV Sales 1,824 2,875 3,600
LDDT Sales 952 1,400 2,340
Total: LDDVs and LDDTs 2,776 4,275 5,940
Base Case Scenario
LDDV Sales 1,793 (1.7%)** 2,835 (1.4%) 3,575 (0.7%)
LDDT Sales 942 (1.0%) 1,389 (0.8%) 2,328 (0.5%)
Total: LDDVs and LDDTs 2,735 (1.5%) 4,224 (1.2%) 5,903 (0.6%)
* California sales included.
** Percent reduction in sales from relaxed scenario.
-------
8-21
Thus diesel manufacturers have been generating higher than
normal profits on diesel sales, relative to profits on
comparable gasoline-fueled vehicles. If this situation
continues, then manufacturers might be able to absorb some of
the costs of a trap-oxidizer system by reducing their
above-normal profit margin. However, it is expected that the
premium in the price being paid for diesels will decrease as
increased competition from other diesel manufacturers brings
profit margins down to normal levels. Manufacturers then would
not be able to absorb the trap-oxidizer cost, and would pass it
through to the consumer.
Even if diesel sales decrease as a result of manufacturers
adding trap-oxidizer costs to their vehicle sales prices, it is
unlikely that the automobile industry as a whole would lose a
sale. JFA concluded that any consumer deciding not to buy a
diesel would purchase a gasoline-fueled vehicle instead. This
finding is not surprising when it is considered that the
functions of the two types of vehicles are nearly identical,
and that only the economics of ownership differ. Thus the
automobile industry as a whole should suffer no lost sales due
to the base scenario particulate control standards.
2. Investment Costs and Cash Flow Effects
Two other effects that emission regulations can have on
diesel manufacturers are increasing required capital investment
(i.e., tooling, machinery, research and development (R&D),
etc.) and reducing cashflow. These effects are examined below.
The bulk of the capital investment associated with the
required use of trap-oxidizers is not expected to be borne by
diesel vehicle/engine manufacturers, but rather by the
manufacturers of emission control equipment, such as Corning,
NGK, and Johnson-Matthey. The catalyst manufacturers already
have developed the necessary substrate technology, and the
diesel manufacturers have shown little interest in this area.
Even though the manufacturers of emission control equipment
will have to finance the necessary investments, they all have
indicated their willingness and ability to enter this market.
Thus pre-production investment costs should not be a problem
for any affected entities.
With respect to other investment costs, light-duty
manufacturers are presently incurring some R&D costs associated
with applying trap-oxidizer technology to their vehicles.
However, much of this work has already been completed, and
future R&D should be no less fundable. Thus, R&D and capital
investment requirements should not have a significant adverse
impact on any manufacturers' investment plans.
-------
8-22
Given the above, the only impact on cash flow will result
from the inventory of traps, individually and on partially
manufactured vehicles. The time each trap is held should be
much shorter than that for an entire vehicle, which averages
about 90 days. [4] This turnover period should be short enough
to not significantly affect a manufacturer's cash flow. For
example, assuming an average turnover time of six weeks and
industry-wide sales of 1.5 million LDDVs and LDDTs, the value
of the trap-oxidizers on hand at any given time would only be
$37.5 million. This is less than $4 per vehicle spread across
total light-duty sales.
D. Total Cost to the Consumer
The bulk of the total consumer cost of particulate control
is the increased "sticker price" of an LDDV or LDDT. Assuming
that the full RPE of manufacturing cost is passed through to
the retail purchaser, the entries of Table 8-3 represent both
total trap-oxidizer system costs and the increase in new
LDDV/LDDT purchase prices due to particulate control. The
remainder of the total cost to the consumer results from the
fuel-economy penalty, and from any increases in the maintenance
costs for light-duty diesels resulting from the addition of
trap-oxidizer systems.
Installation of trap-oxidizer systems is expected to
result in an average fuel economy penalty of 2 percent.[1]
Estimating the cost of this penalty over the life of an
LDDV/LDDT requires that the following information be
specified: cost of diesel fuel, discount rate, average vehicle
miles travelled (VMT) for LDDVs and LDDTs, average LDDV and
LDDT fuel economy, and average LDDV and LDDT lifetime. In this
analysis, the assumptions used are: $1.20/gallon for the
average cost of diesel fuel; a 10 percent discount rate;
average annual VMT of 10,000 for LDDVs and 10,800 for LDDTs;
the estimated 1990 fuel economies for each size class that were
used in determining trap volume requirements; and average
lifetimes of 10 years for LDDVs and 11 years for LDDTs.[10,11]
To calculate the cost of the fuel economy penalty, the
estimated 1990 fuel-economy values (II.B.2.) are reduced by 2
percent. Knowledge of the fuel economy and annual VMT allows
the annual fuel consumption to be determined, which then is
multiplied by $1.20/gallon to yield annual fuel costs. The 10
percent discount rate and the lifetime periods are used to
determine the present value of lifetime fuel expenditures in
the year of vehicle purchase.
The process is repeated without including the 2 percent
fuel-economy penalty, and subtraction of the lower total from
-------
8-23
the higher total gives the cost of the fuel-economy penalty to
the consumer. Carrying through these calculations, the net
present value of the fuel-economy penalty in the year of
vehicle purchase is $33 for small LDDVs, $46 for medium LDDVs,
$52 for large LDDVs, $41 for small LDDTs, and $55 for full-size
LDDTS.
Increased maintenance costs will result only from
maintenance of the trap regeneration system, since the trap
itself is expected to be maintenance-free and use of the
trap-oxidizer system will have no adverse impacts on other
vehicular maintenance requirements. Regeneration system
maintenance is likely to be limited to replacement of the one
or both of the temperature sensors used for system control.
This maintenance is estimated to require about one hour labor
and $10 in new parts, and should only be required once during
the lifetime of the vehicle. Assuming a labor charge of $25
per hour, the total cost of this maintenance is $35. This
maintenance should occur approximately halfway through the
lifetime of the vehicle, or about five years after the initial
purchase. Discounted to the year of the vehicle purchase, the
regeneration system maintenance cost is estimated to be $22.
Use of trap-oxidizer systems will reduce the cost to the
consumer for exhaust system maintenance. By using a stainless
steel exhaust pipe (to discourage in-use trap removal), the
need for periodic replacement of the standard steel exhaust
pipe is eliminated. A conservative estimate of one exhaust
pipe replacement, at roughly the midpoint of the vehicle
lifetime (5 years), being eliminated results in consumer
savings of $21 (for small LDDVs and LDDTs, and medium LDDVs) to
$36 (for large LDDVs and full-size LDDTs).[4] As in the
estimated cost of regeneration system maintenance, a 10 percent
discount rate is assumed.
The sum of the increased LDDV/LDDT initial purchase price,
the cost of the fuel-economy penalty, and the cost of
regeneration system maintenance, less the savings on exhaust
system maintenance, represents the total cost to the consumer
of particulate control. These costs are summarized in Table
8-5 for each of the five size classes of light-duty diesels,
and range from $210 to $266 per vehicle. Against the net
present value of the cost of owning and operating an LDDV over
its lifetime,[12] these costs represent increases of 1.4 to 1. V
percent.
E. 5-Year (1987-91) Aggregate Costs
The annual costs of the base case scenario are shown, for
the years 1987 though 1995, in Table 8-6. These annual costs
-------
Table 8-5
Total Cost to Consumers of Owning and Operating a
Light-Duty Diesel Equipped With a Trap-Oxidizer (1983 dollars)*
Trap-Oxidizer System:
Best Estimate Sales Projections
"Worst-Case" Sales Projections
Maintenance Costs
Maintenance Savings
Cost of Fuel Economy Penalty
Total Cost to Consumer:
Best Estimate Sales Projections
"Worst-Case" Sales Projections
Small
LDDVS
$185
$176
$22
($21)
$33
$219
$210
Medium
LDDVs
$187
$179
$22
($21)
$46
$234
$226
Large
LDDVS
$213
$202
$22
($36)
$52
$266
$255
Small
LDDTS
$187
$179
$22
($21)
$41
$229
$221
Full-size
LDDTS
$211
$205
$22
($36)
$55
$252
$246
Total Cost of Owning and
Operating Vehicle[13]
Cost increase Due to
Trap-Oxidizer
$14,168
1.5%
$16,377
1.5%
$19,418
1.4%
Allcosts are discounted to year of vehicle purchase using a 10 percent
discount rate.
-------
8-25'
Table 8-6
Annual and Five-Year Aggregate
Costs to the Nation of the Base Scenario
for LDDVs and LDDTs (millions of 1983 dollars)
Best Estimate
Sales Projections
LDDVS LDDTS
"Worst-Case"
Sales Projections
LDDVs LDDTs
Annual Cost:
1987
1988
1989
1990
1991
1992
1993
1994
1995
52
60
66
74
76
77
78
79
80
16
18
20
22
23
25
26
27
29
Five-Year
Aggregate Cost*:
1987-91
95
115
131
148
156
164
172
180
187
21
23
26
28
31
35
39
42
46
269
81
526
106
1983 dollars, present value in 1987 using a 10 percent
discount rate.
-------
8-26
were calculated as the product of the per-vehicle cost to the
consumer (from Table 8-5) , the projected sales of LDDVs and
LDDTs (from Table 8-4) , and the diesel trap penetration rates
(from Tables 1-4 and 1-7 in Chapter 1) , using both the best
estimate and "worst-case" sales projections.
5-year aggregate costs are presented in terms of 1983
dollars, net present value in the first year the regulations
would be in effect (1987). As in earlier calculations, the
adjustments made assume a 10 percent discount rate.
III. Heavy-Duty Diesels
A. Introduction
This section is divided into subsections corresponding to
those in the preceding discussion on light-duty diesels.
First, the RPE of the manufacturing costs of trap-oxidizers,
for different classes of HDDEs, are estimated. Second, the
impact of these costs on sales and the capital investment
requirements of the HDDE manufacturers are examined. Next, the
increase in the total cost to the consumer of owning and
operating an HDDE due to the particulate control regulations of
the base case is estimated. Finally, the annual costs (for
1988 through 1995) and the 5-year (1988-92) aggregate cost of
the base scenario are estimated. All of the costs presented in
this section are in 1983 dollars. It is assumed that 1988 will
be the first effective year of heavy-duty diesel particulate
control regulations.
Much of the methodology used here was described in the
section on light-duty diesels, therefore frequent references
are made to the preceding material.
B. Trap-Oxidizer System Costs
1. Introduction and Assumptions
Although most heavy-duty emission regulations make
reference to heavy-duty engines, which are generally tested
separately on engine dynamometers in lieu of tests on the
actual . finished vehicles, for internal consistency this
analysis refers to heavy-duty diesel vehicles (HDDVs).
As was shown previously (II.B.2.), the trap volume
required for a given application can be related directly to the
volumetric exhaust gas flow to be treated. The cost of both
trap designs, catalyzed and non-catalyzed, was then shown to be
a function of the required trap volume. A similar approach is
used in this section for heavy-duty diesels.
-------
8-27
The estimates of the RPE of manufacturing costs for
light-duty diesels were based on the application of adjustment
factors to the estimated manufacturing costs. Additional
adjustment factors were included in the model developed by
Lindgren[5] to compensate for inflation and production volume.
With the exception of the adjustments for production volume,
these factors are unchanged for the heavy-duty case.
In order to estimate standard average trap production
levels for HDDVs, the number of different trap sizes required
must be determined. In the Regulatory Analysis for the
proposed heavy-duty diesel particulate control regulations,[13]
the assumption was that four sizes of traps would be required
to span the entire range of HDDVs. The grouping of HDDVs into
size classes at that time, based on gross vehicle weight (GVW)
classes, is shown below:
Group GVW Classes
1 IIB*, III, IV
2 V, VI
3 VII
4 VIII
In this analysis, HDDVs are divided into only three
groups, on the basis of both GVW classes and relative sales.
Classes VII and VIII HDDVs are consolidated in one group, and
Class V is placed in the same group as Classes IIB-IV. These
groups are referred to in the rest of this section as
medium-duty diesels (MDVs), light heavy-duty diesels (LHDVs),
and heavy heavy-duty diesels (HHDVs), in order of increasing
GVW. This is summarized below:
GVW Classes GVW (Ibs.)
MDVs IIB-V 8,501-19,500
LHDVs VI 19,501-26,000
HHDVs VII-VIII 25,001 and over
These groups have the advantage that each contains one of
the three GVW classes that dominate HDDV sales (IIB, VI, VIII),
while GVW classes having relatively low sales are grouped with
them through similarity of application. This division is also
consistent with the diesel manufacturers' typical grouping of
HDDVs.[14,15]
The standard average production level, for traps for each
of the three LDDV size classes, was estimated in the preceding
* Class IIB in this analysis refers to all vehicles in the
traditional GVW Class II (6,001-10,000 Ibs.) that EPA
classifies as heavy-duty (GVW over 8,500 Ibs., or frontal
area over 45 square feet, or curb weight over 6,000 Ibs.)
-------
8-28
section to be 200,000 annually. This figure was based on the
best estimate diesel sales projections (Chapter 1). These
projections also indicate that approximately half as many HDDVs
will be sold, compared to sales of each LDDV size class, in
each of the three groups defined above. The standard average
production level for each HDDV group is then 100,000 annually.
However, due to the large trap volume requirements for
LHDVs and HHDVs, this analysis assumes that two traps (each
with half the total volume required) , will be fitted to those
vehicles. The standard average trap production levels are then
100,000 for MDVs, and 200,000 each for LHDVs and HHDVs.
Assuming the 12 percent learning curve used in the light-duty
analysis, the adjustment factors for production volume are
1.136 for MDVs, and 1.0 (no adjustment) for LHDVs and HHDVs.
The trap volume requirements are calculated in the
following two sections in the same way that the light-duty trap
volume requirements were determined. Trap size is related to
volumetric exhaust flow, which in turn is proportional to fuel
consumption (inverse of fuel economy). This calculation
requires estimates for the average new-vehicle fuel economy of
each class (MDV, LHDV, HHDV) in the late 1980's and early
1990's. Actual projections of 1990 fuel economy for heavy-duty
gasoline-powered vehicles (HDGVs)[14] were raised by 30 percent
to account for the increased efficiency of diesel engines,
giving the projections of 15.5 mpg (MDVs), 8.4 mpg (LHDVs), and
7.0 mpg (HHDVs) used in this analysis. These figures represent
1990 project average fuel economies for new heavy-duty
vehicles. Thus, they are slightly higher than the fuel economy
projections used in Chapter 2, which represent the entire
heavy-duty diesel in-use fleet in 1990.
2. Non-Catalyzed (Corning) Trap
The trap volume requirements of non-catalyzed traps for
heavy-duty applications are based, as in the light-duty case,
on the successful testing of a 302 cubic inch Corning trap on a
Mercedes-Benz 300D with fuel economy of 26 mpg. The trap
volume requirements that result are 506 cubic inches for MDVs,
934 cubic inches for LHDVs, and 1,122 cubic inches for HHDVs.
As noted, above, the magnitude of the trap volume
requirements for LHDVs and HHDVs was high enough to assume that
-------
8-29
two traps will be fitted, per vehicle, with the total volume
equal to the required size. The individual trap volumes are
then 467 cubic inches for LHDVs and 561 cubic inches for HHDVs.
In section II.A.2., formulae were given that yielded the
RPE of manufacturing cost as a function of trap volume. The
equations 2A-2C, when adjusted for production level as
discussed in the introduction to this section, become:
For MDVs:
RPE = $26 + 0.358(V) (4A)
For LHDVs and HHDVs:
RPE = $23 + 0.318(V) (4B)
Where:
V = volume of trap, in cubic inches.
Substitution of the trap volume requirements for V in
equations 4A and 4B gives the RPEs of the heavy-duty traps.
The 506 cubic inch traps for MDVs have an RPE • of about $207.
For LHDVs, each of the 467 cubic inch traps needed has an
estimated RPE of about $172; the total RPE for two such traps
(per-vehicle RPE) is $343. The 561 cubic inch traps for HHDVs
have an estimated RPE of about $201 each, with a per-vehicle
RPE for two such traps of $403. All of these estimates, which
are based on best estimate sales projections, are shown in
Table 8-7.
3. Catalyzed (Johnson-Matthey) Trap
As in the light-duty case, calculation of catalyzed trap
volume requirements is based on the successful testing of a 345
cubic inch trap on a Volkswagen Rabbit (fuel economy of 42
mpg) . The projected fuel economy of various HDDVs was related
to this fuel economy to obtain catalyzed trap sizes. The
results are: 935 cubic inches for MDVs, 1,724 cubic inches for
LHDVs, and 2,070 cubic inches for HHDVs. The assumption that
the volume requirements for LHDVs and HHDVs would be met by
fitting two traps of equal volume is also used here. The
single-trap volumes are then 862 cubic inches for LHDVs and
1,035 cubic inches for HHDVs.
Equations 3A-3C were used in section II.A.3. to calculate
the RPE of the manufacturing costs for light-duty catalyzed
traps. When the adjustment factors for heavy-duty production
levels are applied, the new equations are:
-------
8-30
Table 8-7
Heavy-Duty Trap Costs (1983 dollars)
Best Estimate Sales Projections
Non-Catalyzed Catalyzed
Vehicle Class Trap Trap
MDVs $207 $636
LHDVs $343 $1,051
HHDVs $403 $1,252
"Worst-Case" Sales Projections
Non-Catalyzed Catalyzed
Vehicle Class Trap Trap
MDVs $183 $560
LHDVs $274 $1,007
HHDVs $403 $1,252
-------
8-31
For MDVs:
RPE = $26 + 0.652(V) (5A)
For LHDVs and HHDVs:
RPE = $23 + 0.583(V) (5B)
Where:
V = volume of trap, in cubic inches.
Therefore the MDV trap, with a volume of 935 cubic inches,
has an estimated RPE of $636. Each of the 862 cubic inch traps
for LHDVs has an estimated RPE of about $526, for a per-vehicle
RPE of $1,051. The RPE of each 1,035 cubic inch HHDV trap is
estimated to be about $626, or $1,253 for the two traps
required. These estimates are all shown in Table 8-7.
All of the estimates for heavy-duty traps discussed above
are based on the best estimate sales projections. Under the
"worst-case" sales projections, sales of MDVs and LHDVs would
double, with trap production for these vehicles also doubling.
As was discussed in section II.B.l., the 12 percent learning
curve assumed indicates that trap costs would be decreased 12
percent by a doubling of production. Since HHDV sales already
represent nearly all sales in GVW Classes VII and VIII, they
remain relatively constant under both sales projections. Thus,
the "worst-case" sales projections have an insignificant effect
on estimated HHDV trap costs. All of this information is
summarized in Table 8-7.
4. Regeneration System Costs
Section I.C. described the components of both catalyzed
and non-catalyzed trap regeneration systems. These basic
systems will also be used for HDDV applications, with two
changes that will have an impact on the cost estimates
presented in Table 8-2. The use of two traps on LHDVs and
HHDVs means that the quantities required of some regeneration
system components will be doubled. In addition, the difference
in the sizes of LDDV and HDDV engines will have an effect on
the costs of the required stainless steel exhaust pipe, as
discussed later.
Table 8-8 summarizes the estimated RPE of manufacturing
cost for both catalyzed and non-catalyzed trap regeneration
systems, for each of the three groups of HDDVs. The estimates
shown include the doubled quantity of some of the components
required for LHDV and HHDV applications.
-------
8-32
Table 8-8
Heavy-Duty Regeneration
System Costs (1983 dollars)
Hardware Item
Retail Price Equivalent
MDDV LHDV HHDV
Non-Catalyzed Trap:
Burner Head
Fuel Delivery System
Ignition System*
Auxiliary Combustion Air System
Exhaust Diversion System*
System Control:
Temperture Sensors
ECU'
Subtotal*
Stainless Steel Exhaust Pipe**
Total*
$7
$9
$5-26
$30
$11-14
$14
$18
$10-31
$30
$15-18
$14
$18
$10-31
$30
$15-18
$12
$37
$24
$37
$24
$37
$111-135 $148-172 $148-172
$33 $53 $89
$143-168 $201-225 $237-261
Catalyzed Trap:
Delayed In-Cylinder Fuel
Injection Mechanism
Auxiliary Combustion Air
System (Reed Valve)
System Control:
Sensors
ECU
Subtotal
Stainless Steel Exhaust Pipe**
Total System Cost
$30
$6
$12
$37
$85
$33
$118
$30
$12
$24
$37
$103
$53
$156
$30
$12
$24
$37
$103
$89
$192
**
Explanations of ranges in costs are in section II.B.4.
For exhaust pipes only, the assumed average production
volume is 100,000 units.
-------
8-33
In section II.B.4. it is shown that, of the current $75
cost of an electronic control unit (ECU), [2] about $10 is
attributable to particulate control on a per-vehicle basis.
Since ECUs are not projected to be in general use on heavy-duty
vehicles before 1988, but will be required under the base
scenario, a greater share of the total cost should be
attributed to particulate control. This analysis assumes that
the ECU will be applied solely for emission control purposes,
and that it will be used for both particulate and NOx control.
Thus, it's cost is divided equally between particulate and NOx
control functions, yielding the ECU cost estimate for
particulate control of $37 shown in Table 8-8.
The additional costs of a stainless steel exhaust pipe for
light-duty diesels were estimated (II.B.4.) as about $16 for 4-
and 6-cylinder engines, and about $27 for 8-cylinder engines.
These costs presume a single exhaust manifold with both 4- and
6-cylinder engines, and a crossover system with the 8-cylinder
engine. In the case of HDDVs, the majority of engine exhaust
systems are the single, non-branching type. Fewer systems are
of the dual exhaust type, where two entirely separate exhaust
systems are used. [13] This analysis assumes that all HDDVs in
GVW Classes IIB-V are manufactured with single exhaust systems,
while for Class VI and larger HDDVs, * 75 percent are
manufactured with single exhaust systems and 25 percent with
dual exhaust systems.[13]
In the Draft Regulatory Analysis to the Heavy-Duty Diesel
Particulate NPRM,[13] it was stated that the basic design of
the LDDV 6-cylinder engine exhaust pipe should be the best
analogue of the exhaust pipe design of HDDVs with single
exhaust systems. This is also assumed in these estimates. For
HDDVs with dual exhaust systems, the resulting cost estimates
are doubled (i.e., two stainless steel exhaust pipes are
assumed to be used).
The estimated cost of converting from a standard steel to
a stainless steel exhaust pipe for an LDDV with a 6-cylinder
engine was given (II.B.4.) as $16, which included credit for
the deleted standard steel pipe. The corresponding costs for
HDDVs are calculated by assuming a direct relationship of
material cost to engine displacement. The typical LDDV
6-cylinder engine is assumed to have displacement of 3.7L. The
typical engine displacements for heavy-duty diesels are assumed
to be 6.2L (MDVs) , 8. 2L (LHDVs) , and 13.9L (HHDVs) . For GVW
Classes VI, VII, and VIII vehicles (LHDVs and HHDVs), the
average per-vehicle cost is calculated by assuming that 75
percent of these vehicles will require one pipe and 25 percent
of them will require two, as discussed above.
-------
8-34
The result of these calculations is the estimated
per-vehicle average RPE of manufacturing cost for stainless
steel exhaust pipes: $33 for MDVs, $53 for LHDVs, and $89 for
HHDVs.
These costs are also shown in Table 8-8. Except for the
exhaust pipes and the doubled quantities of some components,
the cost estimates in Table 8-2 for light-duty trap
regeneration systems remain unchanged in Table 8-8.
5. Total Trap-Oxidizer System Costs
The sum of the estimated costs of the trap and
corresponding regeneration system is the estimated total
trap-oxidizer system cost. These sums are shown, for both trap
types and under both sets of sales projections, in Table 8-9.
As was the case in the light-duty analysis, the cost
ranges are small relative to the absolute costs. Thus, only
the midpoints of these ranges are used in Table 8-9 and in the
remaining analysis. The non-catalyzed trap, which is much less
expensive and appears to be the preferred design of heavy-duty
diesel manufacturers, is the basis of the rest of the analysis.
C. Economic Impact on Manufacturers
In this section, the impact of the base particulate
control scenario on manufacturers' heavy-duty diesel sales,
capital investments, and cash flow are estimated. As for
light-duty, only the costs of the trap-oxidizer system are
considered. Certification costs have already been shown to be
negligible.[13]
1. Impact on Manufacturers' Sales
Estimating the impact of the base scenario on HDD sales is
considerably more difficult than was the case for light duty.
There are two main reasons for this. First, little research
has been conducted into the economic elasticities at work in
the heavy-duty diesel market, and relevant data are scarce. In
addition, there are complicating factors such as the division
of heavy-duty diesels into three groups (MDV, LHDV, HHDV), and
th.e relatively insignificant sales of vehicles in some GVW
Classes (III, IV, and V). Thus the analysis and estimates
presented in this subsection must be considered to be, at best,
rough approximations.
The discussion and the estimated impact of the base
scenario on sales in each HDD group presented in this section
are based primarily on a report recently prepared for EPA by
-------
8-35
Table 8-9
Total Heavy-Duty Trap-Oxidizer
System Costs (1983 Dollars)
Best Estimate
Vehicle Class
MDVs
LHDVs
HHDVs
"Worst-Case"
Vehicle Class
MDVs
LHDVs
HHDVs
Sales Projections
Non-Catalyzed
Trap
$363
$556
$652
Sales Projections
Non-Catalyzed
Trap
$339
$487
$652
Catalyzed
Trap
$754
$1,207
$1,444
Catalyzed
Trap
$678
$1,163
$1,444
-------
8-36
Jack Faucett Associates (JFA).[16] JFA conducted a thorough
literature search and surveyed a number of knowledgeable
individuals, including members of the heavy-duty vehicle and
engine industries, in order to develop the economic elasticity
estimates used here.
Two kinds of price elasticity, own-price and cross-price,
must be considered. Own-price elasticity refers to the change
in the demand for vehicles of a given category resulting from a
change in the purchase price of vehicles in that same
category. Cross-price elasticity takes into account the shifts
that may occur, from diesel to gasoline-fueled engines or
conversely, as a result of changes in the purchase price of
vehicles of one or both engine types within a given category.
In the heavy-duty market, distinct own-price elasticities
exist for each engine type (diesel or gasoline fueled), within
each GVW class (IIB through VIII). JFA supplied estimates of
own-price elasticity for HDDs in Classes IIB, VI, VII, and
VIII; no estimates were given for Classes III, IV, and V due to
low sales.[16] These estimates are applied to the three groups
under consideration here by assuming that own-price elasticity
for MDVs is approximately equal to that of Class IIB alone, due
to the very low sales of vehicles in the other GVW classes.
HHDV own-price elasticity is approximated by the sales-weighted
average of the elasticities of Classes VII and VIII. The
estimates for Class VI are also the estimates for LHDVs, by
definition of LHDV.
The best estimate and "worst-case" sales projections, for
each of the three HDD groups, for 1990 and 1995 are shown in
Table 8-10. Only the sales projections under the relaxed
regulatory scenario are given. Since there is considerable
uncertainty associated with the elasticity estimates used, the
impact on sales of the base regulatory scenario are given in
Table 8-10 as percent reductions from relaxed scenario sales.
Cross-price elasticity is a directional concept, depending
on whether "from diesel to gasoline fueled" or "from gasoline
fueled to diesel" is being considered. In this analysis only
the former is of interest: Given an increase in the purchase
price of HDDs in a given category, the own-price elasticity
estimates how many sales are lost in that category, and the
cross-price elasticity estimates how many of those "lost" sales
are compensated for by increased sales of gasoline-fueled
engines in the same category.
The uncertainties in the cross-price elasticity estimates
used are fairly substantial. Although not shown in Table 8-10,
the results of using the estimated cross-price elasticities are
discussed below.
-------
8-37
Table 8-10
Heavy-Duty Diesel Sales Projections
(in thousands)
Best Estimate Sales Projections
Reduction Due
Vehicle Class 1990 1995** to Base Scenario***
MDVs 124 170 4.8%
LHDVs 94 126 2.2%
HHDVs 159 186 1.0%
"Worst-Case" Sales Projections
Reduction Due
Vehicle Class 1990 1995** to Base Scenario***
MDVs 248 340 4.8%
LHDVS 188 252 2.2%
HHDVs 159 186 1.0%
* California sales included.
** These sales figures are extrapolated from EPA sales
projections for 1985 and 1990.
*** Percent reduction in relaxed scenario sales, applicable to
both 1990 and 1995 projections.
-------
8-38
Of the 4.8 percent reduction in MDV sales projected to
occur under the base control scenario, over a third are
estimated to be made up by increased sales of gasoline-fueled
engines in Classes IIB-V. Thus, the net reduction in sales of
all engines in Classes IIB-V is estimated to be approximately 3
percent. Similarly, the net reduction in LHDV sales is
estimated to be approximately 2 percent. For HHDVs, a drop in
sales of about 1 percent is projected to occur under the base
scenario, and only about one in 50 of those "lost" sales is
projected to be offset by new gasoline-fueled engine sales in
Classes VII and VIII.
It should also be noted that the own-price and cross-price
elasticities estimated by JFA were based only on changes in the
initial purchase price. The effects of increases in operating
and maintenance (O&M) costs are more difficult to incorporate
into the model. In this analysis, the increase in O&M costs
(net present value in year of vehicle purchase, 10 percent
discount rate) was considered to be part of the initial
purchase price increase. Although this is not appropriate,
strictly speaking, it is an adequate approximation when the
uncertainties inherent in the elasticity estimates are taken
into account.
2. Capital Investment and Cash Flow Effects
Implementing a trap-based particulate standard for
heavy-duty diesels should have only minor effects on the
capital expenditures of HDDV manufacturers. The reasons are
basically the same as discussed for light-duty in section
II.B., and are briefly recapped below.
It is quite unlikely that any heavy-duty manufacturer will
choose to make the necessary investments for the production of
trap-oxidizers, as the sophisticated technology required has
already been developed by other firms. In addition, the
production volumes of most individual manufacturers will be far
too small to justify establishment of in-house trap-production
capability. Thus for heavy-duty as well as light-duty, the
bulk of the investments required for trap-oxidizer production
will be financed by emission control equipment manufacturers.
Future R&D investments by the manufacturers are difficult to
estimate, but should not be so high as to adversely affect
other investment plans.
The cash flow impact of these regulations is limited to
the inventory of traps, individually and on partially
manufactured HDDVs. This investment is recovered upon sale of
the trap-equipped HDDV, and the sales turnover period of HDDVs
is short (generally less than four months) . The short
-------
8-39
inventory period, and the relatively small amount of cash
represented, should not significantly affect the cash flow of
any manufacturer.
D. Total Cost to the Consumer
The total cost to the consumer is the sum of the costs of
the trap-oxidizer system, shown in Table 8-9, and the costs of
the 2 percent fuel-economy penalty[4] and increased maintenance
costs, less any maintenance savings. As in the light-duty
analysis, it is assumed that manufacturers pass all of their
costs increases through to the retail purchaser.
The costs of the 2 percent fuel-economy penalty are
estimated by the same methods used for light-duty diesels in
section II.D. The information used in this calculation is:
$1.20 per gallon average diesel fuel cost; 10 percent discount
rate; new-vehicle fuel-economy estimates of 15.5 mpg (MDVs),
8.4 mpg (LHDVs) , and 7.0 mpg (HHDVs) ; annual average VMT of
12,000 (MDVs), 20,000 (LHDVs), and 47,500 (HHDVs); and lifetime
average VMT of 120,000 (MDVs), 200,000 (LHDVs), and 475,000
(HHDVs). When this information is used as described earlier,
the net present value of the lifetime fuel-economy penalty, in
the year of vehicle purchase, is $126 (MDVs), $386 (LHDVs), and
$917 (HHDVs). This is summarized in Table 8-11.
The trap should be maintenance-free, but the regeneration
system will require maintenance once during the lifetime of the
HDDV, after approximately five years of operation. For
light-duty diesels, the discounted cost of regeneration system
maintenance is estimated at $22 (Table 8-5). This cost should
be applicable without adjustment to MDVs, which will be
equipped with a single trap. For LHDVs and HHDVs, with two
traps per vehicle, this estimate is simply doubled to $44.
Table 8-11 also shows these estimates.
A maintenance savings will result from the use of
stainless steel exhaust pipes, which eliminate the need for
periodic replacement of standard steel exhaust pipes. On
average, the total per-vehicle savings would range from $39
(MDVs) to $97 (HHDVs) over the vehicle lifetime, using a 10
percent discount rate and an appropriate schedule for HDDV
standard steel exhaust pipe replacement. [13]
The components of total consumer cost discussed, as well
as the totals, are shown in Table 8-11. The total consumer
costs are given for both best estimate and "worst-case" sales
projections. Also in Table 8-11 is the estimated overall cost
of owning and operating an HHDV over its lifetime, in terms of
-------
8-40
Table 8-11
Total Cost to Consumers of Owning
and Operating a Heavy-Duty Diesel Equipped
with a Trap-Oxidizer (1983 dollars)*
Trap-Oxidizer System:
Best Estimate Sales Projections
"Worst-Case" Sales Projections
Fuel Economy Penalty
Maintenance Costs
Maintenance Savings
Total:
Best Estimate Sales Projections
"Worst-Case" Sales Projections
Total Cost of Owning and
MDDV
$363
339
$126
$22
($39)
$472
$448
-
LHDV
$556
$487
$386
$44
($61)
$925
$856
-
HHDV
$652
$652
$917
$44
($97)
$1,516
$1,516
$274,911
Operating a HHDV[16]
Cost Increase Due to
Trap-Oxidizer
0.6%
All costs are discounted to the year of vehicle purchase
using a 10 percent rate.
-------
8-41
net present value in year of purchase (1983 dollars).[17] As
can be seen, the impact of particulate control on this overall
cost is small, about 0.6 percent.
E. Annual and 5-Year Aggregate Costs
The annual costs of the base regulatory scenario are
shown, for the years 1988 to 1995, in Table 8-12. These costs
were calculated by multiplying the net present value of the
total cost to the consumer, per vehicle, by annual sales. The
5-year aggregate costs are the net present value in 1988 (the
first year the HDDV particulate standard is assumed to be
effective) of the annual costs for 1988 through 1992.
The costs summarized in Table 8-12 are shown for two
possible situations: trap-oxidizers are applied to all HDDVs,
and to only 70 percent of HDDVs. As was discussed in Chapter
1, the lower trap usage rate would be adequate if emissions
averaging is made available to HDDV manufacturers and 85
percent efficiency ceramic traps are used on all trap-equipped
vehicles.
-------
8-42
Table 8-12
Annual Costs and Five-Year Aggregate Costs
to the Nation of the Base Scenario for
Heavy-Duty Diesels (millions of 1983 dollars)
Annual Cost:
Best Estimate Sales
Projections
Trap
70%
st:
88 239
89 251
90 262
91 295
92 328
93 361
94 395
95 430
Usage
100%
341
358
375
422
469
516
564
614
"Worst-Case" Sales
Projections
Trap Usage
70% 100%
325
344
362
402
443
480
521
564
464
491
517
574
633
686
744
805
Five-Year
Aggregate Cost*:
1988-92
1,129
1,614
1,541
2,201
1983 dollars, present value in 1988 using 10 percent
discount rate.
-------
8-43
References
1. "Trap-Oxidizer Feasibility Study," U.S. EPA, OANR,
OMSAPC, ECTD, SDSB, March 1982.
2. "Estimated Costs of Diesel Engine Vehicle Exhaust
Particulate Filter Regeneration Hardware," Mueller Associates,
December 7, 1982.
3. "The Impact of Light-Duty Diesel Particulate
Standards on the Level of Diesel Penetration in the Light-Duty
Vehicle and Light-Duty Truck Markets," Jack Faucett Associates,
EPA Contract No. 68-01-6375, November 30, 1982.
4. "Regulatory Analysis of the Light-Duty Diesel
Particulate Regulations for 1982 and Later Model Year
Light-Duty Diesel Vehicles," U.S. EPA, OANR, OMSAPC, ECTD,
SDSB, October 1979.
5. "Cost Estimations for Emission Control Related
Components Systems and Cost Methodology Description," L.
Lindgren, EPA-460/3-78-006, March 1978.
6. Oral Communication with the Bureau of Labor
Statistics.
7. "Summary and Analysis of Comments on the Notice of
Proposed Rulemaking for the Control of Light-Duty Diesel
Particulate Emissions from 1981 and Later Model Year Vehicles,"
U.S. EPA, OANR, OMSAPC, ECTD, SDSB, October 1979.
8. "Cost Estimations for Emission Control Related
Components/Systems and Cost Methodology Description--Heavy-Duty
Truck," L. Lindgren, EPA-460/3-80-001, February 1980.
9. "The Future of the Diesel Engine: Opportunity and
Risk for the 1980's," Chase Econometrics, June 1982 (not
available to the public until March 1983).
10. Determination of Useful-Life Values for Light-Duty
Trucks and Heavy-Duty Engines, EPA memo From R. Johnson,
Standards Development and Support Branch To Public Docket No.
A-81-11, Index No. IV-B-3, December 13, 1982.
11. "Average Lifetime Periods for Light-Duty Trucks and
Heavy-Duty Vehicles," U.S. EPA, OANR, OMSAPC, ECTD, SDSB,
November 1979.
12. "National Transportation Statistics," Research and
Special Programs Administration, Department of Transportation,
August 1979.
-------
8-44
References (cont'd)
13. "Draft Regulatory Analysis, Heavy-Duty Diesel
Participate Regulations," U.S. EPA, OANR, OMSAPC, ECTD, SDSB,
December 23, 1980.
14. "The Highway Fuel Consumption Model, Eighth
Quarterly Report," prepared by Energy and Environmental
Analysis, Inc., U.S. Department of Energy Contract No.
'DE-AC01-79PA-70032, Task No. 13, July 1, 1982.
15. "Data Resources U.S. Long-Term Review," TRENDLONG
0682, Data Resources Incorporated, Summer 1982.
16. "Estimation of Economic Elasticities in the
Heavy-Duty Vehicle Market," Jack Faucett Associates, EPA
Contract No. 68-01-6375, February 23, 1983.
17. "Operating Costs: Up 20 Percent," Heavy-Duty
Trucking, July 1981.
-------
CHAPTER 9
COST EFFECTIVENESS
I. Introduction
Cost effectiveness is a measure of the economic efficiency
of an action toward achieving a specified goal. It is
primarily useful in comparing alternative means of achieving
that goal. In the context of this study, the goal is to reduce
particulate emissions, or perhaps more importantly, to reduce
ambient levels of particulate where people are exposed. In
this case, cost effectiveness is expressed in terms of the
dollar cost per ton of particulate emission controlled or the
annual dollar cost per microgram per cubic meter of ambient
particulate reduced.
The primary purpose of this chapter is to estimate the
cost effectiveness of the base control scenario over the
relaxed scenario for each diesel vehicle subgroup for both: 1)
comparison among the diesel vehicle subgroups, and 2)
comparison relative to non-mobile source strategies. (The base
and relaxed scenarios are described fully in Chapter 1.)
To determine cost effectiveness, two pieces of information
are necessary: the costs and benefits of the strategies to be
examined. The measure of cost will be the annualized net
present value of all purchase, operating, and maintenance
costs. The measure of benefits will be the annual emission
reduction or ambient concentration of either total, inhalable
or fine particulate.* The three classes of suspended
particulate as examined in order to focus the analysis on the
most important particulate matter with respect to public health
and welfare. As determined in Chapter 5, fine and inhalable
particulate have the primary effect on human health. As
determined in Chapter 4, only fine particulate affects
visibility. As outlined in Chapter 6, all particulate can
participate in soiling.
The remainder of this analysis is divided into three major
sections. The first section estimates the cost-effectiveness
(S/metric ton) for the base control scenario (relative, to the
relaxed scenario) and concludes with a comparison of these
figures for the various diesel subgroups on a nationwide and
Total particulate is all suspended particulate matter
regardless of diameter, inhalable particulate is
considered to be all particulate matter less than 10
micrometers in diameter, and fine particulate is
considered to be all particulate matter less than 2.5
micrometers in diameter.
-------
9-2
urban basis. The second section of the analysis will estimate
cost-effectiveness values for several stationary sources. The
third section will conclude the analysis by: 1) applying a
discount factor to the cost-effectiveness values for both
mobile and stationary sources to account for their relative air
quality impacts, and 2) comparing the cost-effectiveness of
diesel particulate control to those of stationary sources.
II. Cost Effectiveness of Controlling Particulate Emissions
from Diesel Vehicles
A. Methodology
In this section, the cost effectiveness of proceeding from
the relaxed to the base scenario for five diesel vehicle
subgroups will be estimated and compared. These subgroups
include light-duty diesel vehicles (LDDVs), light-duty diesel
trucks (LDDTs), and three subgroups of heavy-duty diesel
vehicles (HDDVs): medium-duty vehicles (MDVs), light
heavy-duty vehicles (LHDVs), and heavy heavy-duty vehicles
(HHDVs).
Most previous EPA cost-effectiveness analyses for mobile
source emissions have determined cost effectiveness using total
lifetime costs discounted to the year of vehicle purchase and
undiscounted lifetime benefits. However, this approach is
somewhat simplistic, since it disregards the fact that the
emission reductions cannot be obtained at the time of vehicle
purchase, when the cost of control is determined. Because of
this, the cost-effectiveness value calculated is entirely
dependent on the point in time costs are determined, which is
somewhat arbitrary.
It would be more appropriate if costs could be allocated
to each period of time in which benefits were produced and in
proportion to the size of these benefits. The result would be
a cost effectiveness which is applicable at any point in that
life as well as over the entire life of the vehicle.
This can be done here for mobile sources through the use
of two simplifying assumptions which will not affect the
accuracy of the cost-effectiveness comparisons. First, it will
be assumed that the number of miles a diesel is driven annually
is constant throughout its useful life. This simplifies the
determination of the miles producing emission reductions each
year. Second, the per-mile emission reduction occurring at the
vehicle's half life will be assumed to apply throughout its
life. This assumption allows direct use of the emission
results of Chapters 1 and 2, since the analysis there also
assumed that emissions were constant with mileage except for
the effect of trap failure. This only results in a slight
-------
9-3
underestimation of emission reductions early in life, with a
compensating overestimation late in life. The overall effect
on cost effectiveness is negligible.
With the use of these two assumptions, the annual emission
reduction throughout a vehicle's life becomes constant and the
cost of control can simply be allocated equally (using discount
theory) to each year of the vehicle's life. This latter
annualized cost is simply an annuity equivalent to the total
cost of control discounted to the year of vehicle purchase,
which was determined in Chapter 7. Costs will be addressed
first and then emission reductions, followed by calculation of
the cost-effectiveness values.
B. Costs of Control
Essentially all of the cost information necessary for the
cost-effectiveness calculations has been developed in Chapter
8. Tables 8-5 and 8-11 of that chapter contain detailed cost
information on the purchase and operating cost impacts for
LDDVs, LDDTs, and HDDVs. These costs are given in 1983
dollars, discounted at 10 percent to the year of vehicle
purchase.
This cost-effectiveness analysis does not require the
level of disaggregation given in Table 8-5 for LDDVs and LDDTs
(e.g., small, medium, and large LDDVs as opposed to simply
LDDVs). Therefore, the costs presented will be combined to
obtain total lifetime consumer costs for LDDVs and LDDTs. As
outlined in Chapters 1 and 8, the largest vehicles are likely
to be equipped with trap-oxidizers first, since they are the
highest emitters. Since the trap usage rates under the base
scenario (22 percent for LDDVs and 9 percent for LDDTs) are
below the projected sales fractions of large LDDVs (26 percent)
and full-size LDDTs (66 percent ),[!] only the largest size
vehicles in each class are likely to have traps. Thus, the
lifetime costs for these largest vehicles will be used here.
Table 9-1 shows the discounted total lifetime consumer
costs for each of the five diesel vehicle groups. (HHDV costs
can be take directly from Chapter 8.) Only those costs for the
best estimate sales scenarios are shown. Costs for worst case
sales would be 0-4 percent lower, because of economies of
scale. (Each vehicle class has a different factor since the
relationship between best estimate and worst case sales is
different for each vehicle class.)
These discounted total costs can be annualized (at
mid-year) over the appropriate average lifetime for each of the
diesel vehicle classes using present value theory. The
expected vehicle lifetimes and the resultant annualized costs
are shown in Table 9-1.
-------
9-4
Table 9-1
Base Scenario Costs (1983 dollars)[1]
LDDV LDDT MDV LHDV HHDV
Lifetime Costs for $266 $252 $472 $925 $1,516
Base Scenario
Vehicle 10 yrs 11 yrs 8 yrs 11.5 yrs 10.5 yrs
Lifetime
Annualized $41.23 $36.97 $84.26 $132.39 $228.46
Cost For A
Trap-Equipped
Vehicle
Percent of 22.3 7.6 100 100 100
Vehicles With
Trap-Oxidizers
Fleet Average $9.19 $2.81 $84.26 $132.39 $228.46
Annualized
Cost Per
Vehicle
[1] Discounted at 10 percent to year of vehicle purchase, best
estimate sales.
-------
9-5
For LDDVs and LDDTs, trap-oxidizers will be used only on
the portion of each manufacturer's sales necessary to bring the
manufacturer's sales-weighted particulate levels down to the
required standard. Since the particulate reduction benefits
will be measured on a fleetwide basis, but costs shown in Table
9-1 only apply to a portion of the fleet, these costs must be
spread over the entire fleet. This can be accomplished by
multiplying the annualized costs of Table 9-1 by the percent of
vehicles requiring traps (taken from Tables 1-4 and 1-7 of
Chapter 1). Since the base scenario does not assume the
availability of an averaging concept for HDDVs, this affects
only LDDVs and LDDTs. (Without averaging, all HDDVs will use
trap-oxidizers and no adjustment needs to be made.) These
fleet-average annualized costs for LDDVs and LDDTs are also
shown in Table 9-1. With averaging, about 70 percent of all
HDDVs would require traps and the fleetwide costs shown in
Table 9-1 would be reduced by approximately 30 percent.
C. Diesel Particulate Emission Reductions
Calculation of the annual diesel particulate emisson
reduction accompanying the base scenario requires information
on annual vehicle miles travelled (VMT) and the emission rates
under the two control scenarios. Table 9-2 shows the average
annual mileage for each of the five diesel vehicle subgroups,
which were derived from each subgroup's average lifetime
mileage and average life (also shown).
Vehicle particulate emission rates (g/mi) tend to increase
gradually with mileage, in a manner in which can be
characterized as linear over the life of the vehicle. Thus,
for either the relaxed or base scenario, one can conceptualize
a stream of annual particulate emissions, increasing by a
constant amount each year. If the emissions in each year for
the base scenario were subtracted from the emissions in each
year for the relaxed scenario, a stream of emission reductions
would be created. Costs could then be allocated to this stream
of benefits to provide a constant and applicable cost
effectiveness throughout the vehicle's life.
As already mentioned in the previous section, a close
approximation to this can be obtained by ignoring the small
change in emissions with time and determining the emission
reduction at the vehicle's half life. The half life for LDDVs
and HDDVs is the fifth year; for LDDTs it is the sixth year.
Vehicle emission rates at the half life for the five
diesel vehicle subgroups were calculated in Chapter 2, and are
shown in Table 9-2. It is worth noting that, for the base
scenario, where trap-oxidizers are used to gain the emission
reductions, the emission rates include the effect of
trap-oxidizer failures.
-------
9-6
Table 9-2
Average Annual Vehicle Miles of Travel
LDDV
Average Lifetime 100
Mileage
Lifetime (years)
Average Annual 10
Mileage
Vehicle Emission Rates at
Relaxed Scenario
Base Scenario
Difference
Annual Emission
,000
10
,000
Half Life
.270
.203
.067
670
LDDT
120,000
11
10,900
(g/mile)
.280
.261
.019
210
MDV
110,000
8
13,750
.997
.454
.543
7,470
LHDV
268,000
11.5
23,300
1.419
.647
.772
17,990
HHDV
529,000
10.5
50,100
2.163
1.015
1.148
57,510
Reduction (grams)
(metric tons)
6.70 x 10-4 2.1 x 10-4 7.47 x 10-3 0.0180
0.0575
-------
9-7
The annual benefits for each diesel vehicle subgroup can
now be calculated by simply finding the difference in the
emission rates from the relaxed and base scenarios and
multiplying by the average annual mileage. These are shown in
Table 9-2 in both grams and metric tons of diesel particulate
controlled.
Diesel particulate matter is very small in size, with mass
mean diameters varying from 0.05 to 0.2 micrometers. As such,
essentially all diesel particulate falls into the fine
particulate cateqory.[2,3,4] Therefore, the benefits for total
particulate given in Table 9-2 also represent the benefits for
inhalable and fine particulate.
D. Cost-Effectiveness Values for Diesel Vehicles
The cost effectiveness of the base scenario is computed by
dividing the fleet average annualized costs from Table 9-1 by
the annual emission reductions from Table 9-2. The resulting
cost-effectiveness values for the five diesel classes are given
in Table 9-3 in the form of 1983 dollars per metric ton. The
cost effectiveness of diesel particulate control is essentially
equivalent for LDDVs and LDDTs (at $13,000-14,000 per metric
ton) , but appears to be better for MDVs and especially LHDVs
and HHDVs.
It is important to note that the figures in Table 9-3
presume the availability of averaging for LDDVs and LDDTs, but
not for MDVs, LHDVs, or HHDVs. In Chapter 8, it was determined
that HDDV compliance costs would drop approximately 30 percent
if an averaging approach was used. These revised figures are
also shown in Table 9-3. As can be seen, this change makes the
control of HDDVs even more attractive relative to that of LDDVs
or LDDTs.
However, perhaps even more important is the fact that
these cost-effectiveness values consider all emission
reductions, regardless of whether the reduction occurs in an
urban or rural area. Since the great majority of Americans
exposed to violations of the NAAQS for particulate matter live
in urban areas, the control of diesel particulate in these
areas should receive the greatest emphasis. One way to do this
is to only consider those emissions reductions occurring in
urban areas in determining cost effectiveness.
As estimated in Chapter 2, the five diesel vehicle
subgroups accumulate different fractions of their annual VMT in
urban areas: LDDVs, 59.4 percent; LDDTs, MDVs, and LHDVs, 48.8
percent; and HHDVs, 26.9 percent. Urban cost-effectiveness
values taking these fractions into account are also shown in
Table 9-3 with averaging for all classes. A comparison of
these values shows all five figures to be much more similar
-------
9-8
Table 9-3
Cost-Effectiveness Values
Total, Inhalable, and Fine Diesel Particulate[l]
Fleet Average
Annual!zed
Cost ($)
Annual Emission
Reduction
(metric tons)
Urban Cost
Effectiveness
with Averaging
for HDDVs
(I/metric ton)
LDDV
LDDT
MDV
LHDV
HHDV
$9.19
$2.81
$84.26 $132.39 $1,228.46
6.70 x ID'4 2.10 x 10-4 0.00747 0.0180 0.0575
Cost Effectiveness 13,700
($/metric ton)
Cost Effectiveness 13,700
with Averaging
for HDDVs
($/metric ton)
13,400
13,400
11,280
7,870
7,350
5,150
3,970
2,780
23,100
27,400
16,200 10,500 10,300
[1] Cost-effectiveness values are the same for total, inhalable and fine particulate.
-------
9-9
than before; however, the control of HDDVs still appears to be
more cost effective than that of LDDVs and LDDTs. These urban
cost-effectiveness values have been developed only for
comparison among the five diesel subgroups. The nationwide
cost-effectiveness values of Table 9-3 will be used in
comparisons with stationary source controls since urban
cost-effectiveness values are not available for stationary
sources.
III. Cost Effectiveness of Controlling Particulate Emissions
from Selected Stationary Sources
A. Introduction
One means of gauging the appropriateness of controlling
diesel particulate emissions is to compare the cost
effectiveness of diesel particulate control against the cost
effectiveness of controlling particulate emissions from
stationary sources. This section of the analysis will be
devoted to developing cost-effectiveness values for stationary
sources. The following section will then develop a methodology
for converting the cost-effectiveness values derived both here
and in the previous section into values which are comparable on
an air qualtiy basis.
A total of eight stationary sources have been selected for
study, based on the availability of control cost information
and emission reductions on a total, inhalable, and fine
particulate basis. These eight sources are listed below:
Source Particulate Control System
Borax Fusing Furnace Venturi Scrubber
Wet Cement Kiln Electrostatic Precipitator
Medium-Sized Industrial Boiler Baghouse
Electric Utility Coal-Fired Electrostatic Precipitator
Generator
Kraft Recovery Furnace Electrostatic Precipitator
Kraft Smelt Tank Venturi Scrubber
Rotary Lime Kiln Electrostatic Precipitator
and Baghouse
Electric Arc Furnace (steel) Baqhouse
Two sets of data and, therefore, two different approaches
will be used in this analysis. Costs and emission reduction
benefits for the first two sources listed above will be
developed here from data contained in a recently published EPA
report on control techniques for stationary source particulate
emissions (herein after referred to as the Control Techniques
document).[5] Cost-effectiveness values for the last six
sources listed above have already been developed in a previous
EPA analysis.[6] These will be used directly here, with some
-------
9-10
adjustments to the costs due to inflation, and, where data
permits, some adjustment to the amount of the inhalable
particulate benefits due to a change in the assumed maximum
diameter for inhalable particulate from 15 to 10 micrometers.
The particle size distributions, source and emission
control systems characteristics, and costs used in this
analysis are based on the best available data and are
representative of the sources being considered. However, it is
important to note that all of the values used would likely vary
from source to source within each source category, so this data
and the analysis which follows cannot be routinely applied to
every individual source. Stationary source emission control
systems are not standardized, but are designed to meet the
needs of each user. However, even with these qualifiers, the
cost-effectiveness values developed here will serve as a valid
basis of comparison with the cost-effectiveness of diesel
particulate control.
B. Cost Effectiveness of Controlling a Borax Fusing
Furnace and a Wet Cement Kiln
1. Costs of Control
Given the necessary information on source and emission
control system characteristics, Volume 1 of the Control
Techniques document mentioned above contains a number of
correlations which can be used to estimate the annualized costs
of particulate emission control systems. These annualized
costs include both capital, direct and indirect operating
costs, and have been developed from data presented in a more
detailed EPA report.[7]
The annualized costs given in the Control Techniques
document cover 8,700 hours per year of operation, or
essentially continuous use. This is probably unrealistic since
a normal downtime for scheduled and unscheduled maintenance of
approximately 10 percent would be expected. Using 8,700 hours
per year without downtime will tend to improve cost
effectiveness, since fixed costs remain during downtime but the
emission reduction is completely lost. However, since no
accurate estimates of downtime experienced by the various
stationary sources are available, no adjustment will be made
here. (Assuming continuous operation happens to also be
consistent with the manner in which the cost-effectiveness
values were calculated in the draft HDD particulate regulatory
analysis, which are addressed in Section C.)
The Control Techniques document presents costs in January
1980 dollars. Updating them to 1983 dollars using the producer
price index for all industrial commodities [8] leads to an
annualized cost increase of about 32 percent.
-------
9-11
Table 9-4 presents the annualized costs for the borax
fusing furnace, the wet cement kiln, and the values of the
particulate control system parameters used to estimate these
costs from the previously mentioned figures. In some cases,
the values for these parameters were taken from the Control
Techniques document. In other cases, the values were based on
emissions data in EPA's Office of Air Quality Planning and
Standards.[9]
2. Emission Reduction Benefits
As was done for diesels, emission reductions for
stationary sources will be developed on a total, inhalable, and
fine particulate basis. Table 9-5 presents size-specific
emissions data for the uncontrolled and controlled cases for
each source. The first column shows particulate concentration
in terms of milligrams per dry nominal cubic meter (mg/DNCM) of
exhaust gas. The annual emission levels were determined by
multiplying the mass concentrations by the exhaust gas flow
rates expressed in dry nominal cubic meters. These exhaust gas
flow rates were estimated to be 32 DNCM/sec for the borax
fusing furnace and 52 DNCM/sec for the wet cement kiln using
the actual exhaust gas flow rates from Table 9-4 and the
appropriate adjustment factors for temperature, pressure, and
moisture content.
Now, given the exhaust gas flow rate in DNCM/sec, size
specific mass concentration in mg/DNCM before and after
control, and an annual operation period of 8,700 hours per
year, the annual metric tons of particulate emissions and
reductions by particle size can be calculated. Table 9-5 shows
these annual emission rates on a particle size basis before and
after control for both the borax fusing furnace and wet cement
kiln, assuming a constant reduction efficiency with time.
Subtracting emission rates before and after control gives the
emission reduction.
Given the annualized cost values in Table 9-4 and the
annual emission reduction benefits in Table. 9-5,
cost-effectiveness values on a total, inhalable, and fine
particulate basis can be determined. These are shown in Table
9-6.
C. Update of Previously Developed Cost-Effectiveness
Values
In previous analyses, EPA developed cost-effectiveness
values for a number of different stationary sources and
particle sizes.[6] These values require two adjustments before
being used in this analysis. First, costs must be updated from
1980 to 1983 dollars. This can be accomplished using the 32
percent change in the producer price index for all industrial
commodities which was also used above.
-------
9-12
Table 9-4
Parameter Values and Annualized Control Costs
for Selected Stationary Source Particulate Controls
(1983 dollars)
Control Control System Exhaust Gas Annualized
Source System Parameters Rate (Am-Vsec) Cost
Borax Furnace Scrubber P = 11 kPa 38 $1,170,000
Wet Cement Kiln ESP SCA =120 m2/(m3/sec) 130 $1,320,000
-------
Table 9-5
Emissions Data for Borax Fusing
Furnace and Wet Cement Kiln
Total
Borax Fusing Furnace
Uncontrolled
Controlled
Reduction
Wet Cement Kiln
Uncontrolled
Controlled
Mass
Concentration
(mg/DNCM)
784
24 .3
2.02 x 107
67.4
Annual
Emissions
(metric tons)
786
24
762
3.29 x 107
110
Particulate
Size Basis
Inhalable
Mass
Concentration
(mg/DNCM)
596
20.6
14,800
109
Annual
Emissions
(metric tons)
598
20
578
24,000
109
Fine
Mass
Concentration
(mg/DNCM)
531
19
6,000
63.1
Annual
Emissions
(metric tons)
532
20
512
U)
9,770 ^
103 "
Reduction
3.29 x 107
23,991
9,667
-------
9-14
Table 9-6
Cost Effectiveness
Diesel Vehicles and Stationary Sources[1/2]
(1983 Dollars per metric ton
Particulate Size Basis
Source Total Inhalable Fine
Wet Cement Kiln [8] 55 136
Kraft Smelt Tank 250 299 455
Electric Arc Furnace[3] 924 1,440 1,452
Electric Utility[4] 1,254 1,805 4,092
Industrial Boiler[5] 1,320 1,848 5,544
Rotary Lime Kiln (ESP)[6] 1,584 1,980 3,168
Borax Fusing Furance 1,532 2,021 2,281
Rotary Lime Kiln (Baghouse)[6] 1,716 2,112 3,300
Kraft Recovery Furnace[7] 1,678 2,145 3,055
[1] Ranked according to Inhalable Particulate Cost
Effectiveness.
[2] For simplification, the midpoint of the ranges were used
where applicable.
[3] Direct evacuation with 90 percent efficient canopy hood
versus direct evacuation with open roof.
[4] High efficiency ESP (0.03 lb/106 BTU) versus lower
efficiency ESP (0.1 lb/106 BTU).
[5] Baghouse (0.03 lb/106 BTU) versus cyclone (0.3 lb/106
BTU) .
[6] High efficiency ESP (0.6 Ib/ton limestone) versus lower
efficiency ESP (0.6 Ib/ton limestone) for 500 TPD plant;
baghouse (0.3 Ib/ton) versus lower efficiency ESP for 125
TPD plant.
[7] High efficiency ESP (99.5 percent) versus lower efficiency
ESP (99.0 percent).
[8] Less than $1 per metric ton.
-------
9-15
Second, the inhalable particulate emission reductions
estimated previously also require some adjustment due to a
change in the assumed cutoff diameter from 15 micrometers in
the 1980 analysis to 10 micrometers in the present analysis.
This reduction in emission benefits will in turn lead to an
increase in the relative cost effectiveness on an inhalable
particulate basis.
After reviewing the sources for the original estimates and
other data developed since that time, entirely new estimates
for the mass percent of inhalable particulates have been
developed for the electric utility and the electric arc
furnace. The inhalable fraction of electric utility
particulate was decreased from the 90-100 percent range to 66
percent based on discussions with OAQPS staff.[9] Electric arc
furnace inhalable particulate fraction data was adjusted from
90 to 66 percent based on data in the Control Techniques
document. In the other cases, no data were available to make
any adjustments, so it was assumed that all of the particulate
controlled at 15 micrometers or less were also all less than 10
micrometers. This may overestimate the amount of inhalable
particulate controlled and, thus, improve inhalable particulate
cost-effectiveness. However, given the absence of data to the
contrary, this is the best estimate that can be made at this
time.
After adjustments for inflation and the change in
inhalable particle diameter, Table 9-6 gives the final
estimates of the cost effectiveness on a total, inhalable, and
fine particulate basis for the six stationary sources
previously analyzed and the two sources addressed in Section
B. They are listed in order of their inhalable particulate
cost effectiveness, from best to worst. Also shown is some
information on the control strategy on which the costs and
emission reduction benefits are based for the previously
analyzed sources.
IV. Discounted Cost Effectiveness for Mobile and Stationary
Particulate Sources
A. Introduction
As described in the Introduction, it is not strictly
appropriate for the purposes of this study to compare the cost
effectiveness of particulate control from various sources on
the basis of simple dollar per metric ton values. As the goal
of diesel particulate control is to reduce population exposure
to suspended particulate, the measure of effectiveness should
be based on changes in population exposure to suspended
particulate, like those estimated in Chapter 3. However, since
such exposure estimates are not available for specific
stationary source emission control scenarios, this basis cannot
be used for comparison purposes here.
-------
9-16
One step short of assessing population exposure impacts is
the assessment of ambient (at ground level) particulate
concentration impacts. This is a definite improvement over
simple dollar per ton meaures since the dispersion
characteristics of the source, which have a definite impact on
population exposure, are included. Comparing cost
effectiveness on this basis is also somewhat appropriate, since
the determination of satisfactory air quality is made on this
level via the NAAQS. There also happens to be an approximate
means of assessing the relative impact of various sources of
emissions on ground level particulate concentration available.
Given that this is the case, the effectiveness of various
particulate source controls will be compared in this study on
the basis of their relative impact on ground level particulate
concentration.
However, performing the comparison at this level ignores
the location of these ground level concentrations, particularly
with respect to the number of people exposed and the local need
for control (i.e., is the area in or out of compliance with the
NAAQS). Unfortunately, this is a significant limitation since
stationary sources can often be controlled on an individual
basis (i.e., where the air quality problems are), while mobile
sources cannot. This effect results in a relative inefficiency
of the mobile source approach which cannot be factored in at
this time. Thus, the conclusion of this cost-effectiveness
comparison cannot be conclusive (i.e., that diesel particulate
control is cost effective relative to stationary source
particulate control) . At best, it can only be said that there
is no evidence that diesel particualte control is not cost
effective with respect to stationary source control and that
control of diesel particulate should not be avoided due to
cost-effectiveness concerns.
The comparison of cost effectiveness on an air qualtiy
basis will be conducted in three steps. First, it will be
necessary to determine an expression which relates the effect
of various source characteristics on the ground level
particulate concentration resulting from a given emission
rate. Second, the pertinent source characteristics for the
various sources under consideration here and the resultant air
quality discount factors will also have to be determined.
Third, once these factors have been determined, it will be
possible to calculate discounted cost-effectiveness values for
all sources which can then be compared with those for diesels.
A. Methodology for Evaluating the Ground Level Impact
of Stationary Source Particulate Emissions
There are many characteristics unique to each source which
can affect its relative contribution to ground level
particulate concentrations. The meteorological conditions of
-------
9-17
the area, particle size and density, release height, and others
can all affect dispersion. Given that 1) local meteorological
conditions cannot be taken into account in a study of this
breadth, and 2) this study is primarily concerned with
particulate less than 10 microns in diameter (i.e., similar
particle-related dispersion), the primary remaining factor
affecting dispersion is release height.
In a recently released EPA document, an expression has
been developed which provides a reasonable approximation of the
dependence of the maximum ground level particulate
concentration on effective release height.[10] This
relationship is provided below:
W = 10/H for H greater than 10 meters;
W = 1 for H less than or equal to 10 meters.
Where:
W = discount factor, maximum ground level particulate
concentration relative to a ground level source,
H = effective release height, in meters (m).
This relationship is being used by OAQPS in their
reconsideration of the NAAQS for particulate matter to relate
the impact of various emission source controls on ambient
particulate levels and compliance with the NAAQS. The general
concept is also analogous to the use of source discount factors
in rollback air quality modelling.
As would be expected, this equation showns an inverse
relationship between maximum ground level contribution and
effective release height; i.e., as release height increases,
the maximum contribution from this source decreases.
B. Effective Release Heights and Discount Factors
The effective release height for any emission source is
equal to the sum of the physical stack heiaht and the vertical
height which the plume rises before significant horizontal
dispersion occurs. While stack height is easily measured and
fixed over time, plume rise varies according to source
characteristics and meteorological conditions (e.g., stack gas
temperature, exhaust gas flow rate, atmospheric stability, air
temperature, wind velocity).
It is intuitively clear that the effective release height
for diesel vehicles is less than 10 meters, and when evaluated
in the equation above, yields the conclusion that diesel
vehicles can be considered a ground level source (discount
-------
9-18
factor equal to 1.0). However, for stationary sources this may
not be the case, and effective release height calculations are
necessary.
A number of different models to calculate plume rise under
various atmospheric stability conditions have been developed
over the past 35 years. One approach which has gained
widespread acceptance was developed by Briggs and will be used
here to estimate the plume rise for the eight stationary
sources under consideration.[11] As a further simplification,
the Briggs formulae for a stable/near neutral atmosphere will
be used in preference to those for an unstable atmosphere. It
should be noted that this will tend to improve cost
effectiveness (low cost-effectiveness values) of stationary
source particulate controls, since particulate dispersion is
significantly increased during increased atmospheric
instability relative to that for neutral to stable atmospheres
and the resulting ground-level impacts would be lowered.
The Briggs formulae (shown in Figure 9-1) require
information on both source and atmospheric characteristics.
Source characteristic values needed include the exhaust gas
exit temperature and exhaust gas volumetric flow rate. These
are shown in Table 9-7 along with their sources. Atmospheric
conditions needed include the ambient air temperature, wind
velocity, and atmospheric vertical temperature gradient at the
stack exit. The choice to use an atmosphere with stable to
near neutral characteristics will dictate values for these
conditions. The values used here are -2°C/305 m for the
ambient air temperature lapse rate, 288°K for the ambient air
temperature at the stack exit, and 5 m/sec for the wind speed
at the stack exit. These are fairly typical values for a
midwestern U.S. city under stable to near neutral conditions,
based on the TCAO U.S. standard atmosphere. The resultant
plume rise heights are also shown in Table 9-7.
The effective release height is the sum of the stack
height and the plume rise. Typical stack heights for the
sources/control systems under consideration are given in Table
9-7. When these terms are added for the sources under
consideration here, the effective release heights shown in
Table 9-7 result. Using these effective release heights, and
the relationship given in the equation above, Table 9-7 gives
the values of the weighting factor for the sources/control
systems under construction here. Note that for diesel vehicles
the weighting factor is 1.0 since the effective release height
is less than 10 meters.
C. Air Quality Discounted Cost-Effectiveness Values
All that remains to be done to estimate cost effeciveness
on an air quality basis is to divide the cost effectiveness
-------
1. h = 2.3
9-19
Figure 1
Plume Rise Calculation Equations
v 1/3
Us
2. _, _ q Q (TS - Ta)
F ' Ti
3. g dT 3C'
Ta dz + 305m
h = plume rise (meters).
F = bouyancy flux.
U = wind speed at stack exit (m/s).
s = stability parameter.
g = acceleration of gravity (9.8 m/s2).
Q = exhaust gas volumetric flow rate (m3/s).
Ts = exhaust gas exit temperature (°K).
Ta = ambient air temperature at stack exit elevation (°K)
^L = ambient air temperature lapse rate.
dz
-------
Table 9-7
Source Characteristic Parameters, Plume Rise,
Effective Release Height, and Weighting Factor
Source/Reference
Borax Furnace[5,9]
Cement Kiln[5,9]
Electric Utility[5]
Industrial Boiler[5]
Electric Arc Furnace[12]
Rotary Lime Kiln[13]
Kraft Furnace[14]
Kraft smelt Tank[14]
Control
System
Scrubber
ESP
ESP
Baghouse
Baghouse
ESP
Baghouse
ESP
Scrubber
Flow
Rate
Q(Am3/s)
38
30
533
163
62
3,000
800
76
7,000
Stack
Temp(°K)
353
433
400
470
346
474
405
430
351
Plume
Rise(m)
83
101
242
191
95
509
281
136
470
Stack
Height(m)
12
46
175
55
19
30
25
75
53
Effective
Release
Height(m)
95
147
417
246
114
539
306
211
523
Discount
Factor (W)
.105
.068
.024
.041
.088
.019
.033
.047
.019
vo
I
o
-------
9-21
values of Table 9-6 by the discount factors of Table 9-7.
These discounted cost effectiveness values are shown in Table
9-8.
The figures in Table 9-8 show that after consideration of
relative air quality effects, the base scenario is quite cost
effective relative to stationary source controls regardless of
the size of particulate examined. While the control of wet
cement kilns is more cost effective than diesel particulate
control across the board, only one other source is
significantly more cost effective on a TSP basis (industrial
boilers) . No other sources are more cost effective on a fine
or inhalable particulate basis. As mentioned earlier, the
control of both fine and inhalable particulate are most
important with respect to protecting the public health, the
control of fine particulate is most important with respect to
visibility, and the control of total particulate is most
important with respect to soiling.
To further place these figures in perspective, Table 9-9
shows estimates of annual emissions nationwide for most of the
source categories listed in Table 9-8. However, the two tables
do not match up exactly one-to-one. The emission estimates
apply to entire industrial categories, while in a few cases
(e.g., lime kilns and electric arc furnaces) the sources listed
in Table 9-1 represent only a fraction of the industrial
category emissions. Nonetheless, these emission estimates will
be sufficient for our purposes here.
The nationwide emission estimates of Table 9-9 can be used
to compare the potential for emission reduction from the
stationary sources to that available for diesels. As can be
seen, the base scenario will reduce nationwide emissions by
roughly 120,000 metric tons per year in 1995. Only three of
the stationary sources being considered here could potentially
provide the same emission reduction: electric utilities, the
cement industry, and industrial boilers. Given that the cement
industry is predominantly located in rural areas,[16] only the
remaining two sources can produce the same emission reduction
where it is most needed. In addition, the impact of these
sources on ground-level ambient concentrations relative to that
of diesels must also be kept in mind.
VI. Summary
The cost effectiveness of the base scenario relative to
the relaxed scenario has been estimated for five classes of
diesels. For the purposes of comparing control between the
diesel vehicle classes, cost-effectiveness values were
determined on both a nationwide and urban basis, as well as for
the control of total, inhalable and fine particulate. The cost
effectiveness of controlling stationary source particulate
emissions was also estimated. In order to compare these varied
-------
9-22
Table 9-8
Summary
Air Quality Discounted Cost Effectiveness
Diesel Vehicles and Stationary Sources
($ per metric ton)[lf2,3,4]
Source
Wet Cement Kiln
HHDV[5]
LHDV[5]
MDV[5]
LDDV[5]
LDDT[5]
Kraft Smelt Tank
Electric Arc Furnace
Borax Fusing Furance
Industrial Boiler
Kraft Recovery Furnace
Lime Kiln (Baghouse)
Electric Utility
Lime Kiln (ESP)
Particulate Size Basis
Total
1
3,780
7,350
10,200
13,700
14,400
13,200
10,500
14,600
32,200
35,700
52,000
52,250
83,400
Inhalable
810
3,780
7,350
10,200
13,700
14,400
15,700
16,400
19,250
45,100
45,600
64,000
75,200
104,000
Fine
2,000
3,780
7,350
10,200
13,700
14,400
23,900
16,500
21,700
135,000
65,000
100,000
170,500
167,000
inhalable particulate cost
TT]1983 dollars.
[2] Ranked according to
effectiveness.
[3] For simplification, the midpoint of the ranges shown in
Table 9-12 were used.
[4] Cost Effectiveness (Table 9-12) divided by Discount Factor
(Table 9-14).
[5] Assumes presence of emissions averaging.
-------
9-23
Table 9-9
Annual Nationwide Emission Rates by Source Category
Stationary Source (1981)[15] Metric Tons Per Year
Electric Utilities
Cement Industry
Industrial Boilers
Concrete, Lime, Gypsum Industry
Pulp Mills
Iron and Steel Foundries
Borax Furnaces
On-Highway Diesels
(best estimate sales)
1980
1995 Relaxed Scenario
Base Scenario
1,000,000
460,000
400,000
140,000
110,000
50,000
Unavailable
Metric Tons Per Year
140,000
285,000
166,000
-------
9-24
sources against the goal or improving air quality, emission
control effectiveness was discounted according to the effective
release height of the emission and its effect of dispersion.
While this methodology accounts for source-specific dispersion
effects, it does not account for the importance of the location
of the air quality improvement. This is a significant
drawback, and prevents a fully appropriate comparison from
being made.
The results of the analysis show that on an air quality
basis the control of diesel particulate is cost effective
relative to stationary source controls regardless of whether
fine, inhalable, or total particulate are considered. However,
due to the limitations of the methodology, the best that can be
said at this time is only that there is no evidence that diesel
particulate control is not cost effective with respect to
available stationary source control and that the control of
diesel particulate should not be avoided due to
cost-effectiveness concerns. Between the subgroups of diesel
vehicles, on an urban basis (the most appropriate) and assuming
the presence of averaging, HHDVs are the most cost effective to
control, followed by LHDVs, MDVs, LDDVs, and LDDTs.
-------
9-25
References
1. "The Impact of Light-Duty Diesel Particulate
Standards on the Level of Diesel Penetration in the Light-Duty
Vehicle and Light-Duty Truck Markets," Jack Faucett Associates,
EPA Contract No. 68-01-6375, November 30, 1982.
2. "Particulate Size Variation in Diesel Car Exhaust,"
Groblicki, P., and C. Begeman, SAE 790421.
3. "Characterization of Diesel Exhaust Particulate
Under Different Engine Load Conditions," Presented at 71st
Annual Meeting of APCA, Schreck, R., et.al., June 25-30, 1978.
4. "Characterization of Particulate and Gaseous
Emissions from Two Diesel Automobiles as Functions of Fuel and
Driving Cycle," Hare, C. and T. Baines, SAE'Paper No. 790424.
5. "Control Techniques for Particulate Emissions from
Stationary Sources," Vols. 1 and 2, U.S. EPA, OANR, OAQPS,
EPA-450/3-81-005a and b, September 1982.
6. "Draft Regulatory Analysis, Heavy-Duty Diesel
Particulate Regulations," U.S. EPA, OANR, QMS, ECTD, SDSB,
December 23, 1980.
7. "Capital and Operating Costs of Selected Air
Pollution Control Systems," U.S. EPA, OANR, OAQPS,
EPA-450/5-80-002, December 1978.
8. Figures gathered by the Bureau of Labor Statistics
and compiled in the February 1983 "Economic Report of the
President."
9. Extracted from selected Fine Particulate Emissions
Inventory System data.
10. "Draft of Regulatory Impact Analysis for Proposed
Revision of National Ambient Air Quality Standard for
Particulate Matter,"
11. Air Pollution, McGraw Hill Book Company, Perkins,
H., 1974.
12. "Background Information for Standards of
Performance: Electric Arc Furnaces in the Steel Industry, Vol.
1: Proposed Standards," U.S. EPA, OAWM, OAQPS,
EPA-450/2-74-017a, October 1974.
13. "Standards Support and Environmenal Impact
Statement, Vol. 1: Proposed Standards of Performance for Lime
Manufacturing Plants," U.S. EPA, OAWM, OAQPS,
EPA-450/2-77-007a, April 1977.
-------
9-26
References (cont'd)
14. "Standards Support and Environmental Impact
Statement: Vol. 1: Proposed Standards of Performance for
Kraft Pulp Mills," U.S. EPA, OAWM, OAQPS, September 1976.
15. "National Air Pollutant Emission Estimates,
1970-1981," U.S. EPA, OAN.R, OAQPS, EPA-450/4-82-012, September
1982.
16. Information Concerning Particulate Emissions from
Non-Mobile Sources, Memo from R. Neligan to C. Gray, U.S. EPA,
July 11, 1979.
17. "Air Quality Assessment of Particulate Emissions
from Diesel Powered Vehicles," PEDCo Environmental Inc. for
EPA, Contract No. 68-02-2512, March 1978.
18. "The Impact of Future Diesel Emission on the Air
Quality of Large Cities," PEDCo Environmental Inc. for EPA,
Contract No. 68-02-2595, February 1979.
19. "An Investigation of Future Ambient Diesel
Particulate Levels Occurring in Large-Scale Urban Areas,"
Technical Report, U.S. EPA, OANR, OMS, ECTD, SDSB,
EPA-AA-SDSB-79-30, November 1979.
20.. "Regulatory Analysis of the Light-Duty Diesel
Particulate Regulations for 1982 and Later Model Year
Light-Duty Diesel Vehicles," U.S. EPA, OANR, OMS, February 20,
1980.
-------
CHAPTER 10
SENSITIVITY
I. Introduction
This chapter contains a variety of analyses intended to
address the sensitivity of the previous technical analyses to
key assumptions that were made. The first analysis addresses
the assumed levels of the LDV and LDT NOx standards. While the
current NOx standard were presumed to continue indefinitely for
ease of analysis, this is actually not likely to be the case.
As additional NOx control tends to increase engine-out
particulate levels, more stringent NOx standards would increase
emissions under the relaxed scenario and increase the number of
traps required under the base scenario. The cost effectiveness
of trap application would also be affected.
The second analysis has a two-fold purpose. One, it
addresses the assumption that the analysis of the base
scenario, which only requires a minority of LDDVs and LDDTs to
be equipped with traps, adequately addresses the economic
viability (cost and cost effectiveness) of trap-oxidizer usage
in general. Two, it expands the previous benefits analyses by
estimating emissions (and, thus, other environmental effects)
under the stringent particulate control scenario.
The third analysis addresses the possibility of using HDD
emissions under the relaxed scenario as an estimate of
uncontrolled emissions, which is usually desirable to present
in a regulatory analysis.
The first two analyses will be presented together, as they
overlap technically to a significant degree. The third
analysis will follow. This analysis will only address certain
basic features of each scenario, such as fleet emissions, trap
usage and cost effectiveness. More advanced aspects, such as
exposure, cancer risk, and economic impact, are not presented.
This was done because all of the benefits described in this
study are proportional to fleet-wide emissions in a given
calendar year, except for visibility effects, which are nearly
proportional in the range being examined. Thus, the
sensitivity of urban emissions in either of the two sensitivity
analyses indicates the same sensitivity in any other benefit
category. A quantitative estimate of any or all benefits under
one of the new scenarios being analyzed here can be determined
simply by applying the ratio of fleetwide urban emissions to
the estimate of benefits under one of the scenarios analyzed in
the previous nine chapters. The same is true for economic
impact, which is essentially proportional to the fraction of
vehicles with traps.
-------
10-2
II. Light-Duty NOx Standards and the Stringent Control Scenario
The previous nine chapters assumed that the NOx standard
for LDVs and LDTs would remain at 1.5 and 2.3 g/mi,
respectively, throughout the time period covered by this
study. In this section, three additional sets of LDV/LDT NOx
standards are investigated: 1) 1.0/1.2 g/mi, 2) 1.5/1.7
g/mi, and 3) 2.0/2.3 g/mi. I
In addition, the previous nine chapters only addressed two
control scenarios, the relaxed and the base scenarios. Here, a
third scenario, the stringent scenario, will be examined. It
consists of full, trap-based standards of 0.08 g/mi for LDDVs,
0.10 g/mi for LDDTs, and 0.10 g/BHP-hr for HDDs. The LDDV
standard of 0.08 g/mi is that promulgated by California for the
1989 model year. The LDDT and HDD standards follow from this
level in that they require the same percentage reduction from
the base scenario.
Four key aspects of these scenarios will be addressed.
The first aspect addressed will be manufacturers' corporate
average particulate standard levels associated with the relaxed
scenario under the three sets of NOx standards. The second and
third aspects are directly related, the fraction of vehicles
requiring traps under the base and stringent scenario, and
urban particulate emissions in 1995 under the relaxed, base and
stringent scenarios under the various NOx standards. The
fourth aspect will be the cost effectiveness of the base and
stringent scenarios under the various NOx standards.
A. Manufacturers' Corporate Average Standard Level
The methodology used to estimate each manufacturer's
current (relaxed scenario) corporate average standard level
under NOx standards of 1.5 and 2.3 g/mi for LDVs and LDTs,
respectively, was presented in Chapter 1. There, each engine
configuration's low mileage particulate emission level was
first adjusted for the NOx emission level under consideration.
This was accomplished through the use of estimated
NOx/particulate tradeoff curves. The slope of the curve for
small LDDV engines (1.6-1.8 liters displacement) was -0.033 for
NOx values less than or equal to 1.35 g/mi and zero for NOx
values greater than 1.35 g/mi. For medium LDDV engines (2.0 to
2.8 liters displacement), the slope of the curve was -0.20 for
NOx values less than or equal to 1.35 g/mi and -0.10 for NOx
values greater than 1.35 g/mi. For large LDDV engines, the
slopes were -0.40 and -0.10 for NOx values less than and
greater than 1.35 g/mi, respectively. The slopes of the
NOx/particulate tradeoff curves were the same for LDDTs.
-------
10-3
However, small LDDTs have displacements from 1.6 to 2.3 liters
and full-size LDDTs have displacements of 6.2 liters. There
were no "medium" LDDTs.
Once each engine configuration's low-mileage particulate
emission level was estimated, its particulate standard level
was determined by multiplying the particulate emission level by
its deterioration factor and the safety factor. Each
manufacturer's engine configurations were then sales-weighted
to give that manufacturer's corporate average standard level.
Tables 10-1 and 10-2 show each manufacturer's corporate
average particulate standard levels for LDDVs and LDDTs under
the various NOx standards. For LDDVs, going to a 1.0 g/mi NOx
standard from a 1.5 g/mi NOx standard increases particulate
emissions more than twice as much as going from a 2.0 g/mi NOx
standard to a 1.5 g/mi NOx standard. The effect of moving to a
1.2 g/mi from a 1.7 g/mi NOx standard for LDDTs is small for
small LDDTs but is dramatic for full-size LDDTs. The impact of
moving from a 2.3 g/mi to a 1.7 g/mi NOx standard is negligible
for small LDDTs but is measurable (18 percent increase) for
full-size LDDTS. These impacts will reappear below when the
effects of various NOx standards on urban emissions under the
relaxed scenario are considered later in this section.
B. Percent of Trap-Equipped Vehicles
The methodology for calculating the percentage of each
model year's LDDVs and LDDTs to be equipped with trap-oxidizing
systems was also presented in Chapter 1. Basically, the number
of vehicle grams per mile (veh-g/mi) of diesel particulate
allowed to each manufacturer under the base scenario (i.e.,
particulate "averaging" standards of 0.20 and 0.26 g/mi for
LDDVs and LDDTs respectively) was determined from
manufacturer's projected 1985 sales. Then, the number of
veh-g/mi of diesel particulate that would actually be emitted
by each engine configuration under NOx standards of 1.5 and 2.3
g/mi for LDVs and LDTs, respectively, without traps were
calculated. Finally, traps were applied to reduce each
manufacturer's diesel particulate veh-g/mi to the allowable
level which gave the percentage of each manufacturer's
production that would need to be equipped with traps.
Tables 10-3 and 10-4 show the percentage of each
manufacturer's LDDV and LDDT production (and that of the
overall fleet) that would need to be equipped with traps under
the base scenario and three sets of NOx standards, and under
the strinqent scenario and two sets of NOx standards. Under a
1.0 g/mi LDV NOx standard, the percentage of the LDDV fleet
which would require traps under the base scenario would more
-------
10-4
Table 10-1
Relaxed Scenario
Corporate Average Particulate
Standard Levels for LDDVs
(grams per mile)
1.0 g/mi 1.5 g.mi 2.0 g/mi
Manufacturer NOx Standard NOx Standard NOx
Standard
General Motors .50 .29 .25
Volkswagen .21 .20 .20
Nissan .29 .26 .25
Mercedes-Benz .60 .42 .34
Isuzu .22 .20 .20
Audi .26 .20 .18
Peugeot .36 .26 .21
Volvo .41 .29 .24
Sales-Weighted .42 .27 .24
Industry Wide
Average
Percentage of 48% 22% 14%
Industry Requiring
Traps Under Base
Scenario
Percentage of 95% 82% 72%
Industry Requiring
Traps Under Stringent
Scenario
-------
10-5
Table 10-2
Relaxed Scenario
Corporate Average Particulate
Standard Levels for LDDTs
(grams per mile)
1.2 g/mi 1.7 g/mi 2.3 g/mi
Manufacturer NOx Standard Nox Standard Nox Standard
Small LDDTS:
Ford .30 .29 .29
Isuzu .33 .25 .25
Nissan .37 .35 .35
Mitsubishi .43 .39 .39
Toyota .20 .19 .19
Volkswagen .32 .31 .31
Toyo Kogyo .30 .29 .29
Full-Size LDDTs:
General Motors .56 .34 .28
Sales-Weighted, .52 .33 .28
Industry-wide
Average
Percentage of 56% 24% 8%
Industry Requiring
Traps Under Base
Scenario
Percentage of 95% 83% 77%
Industry Requiring
Traps Under Stringent
Base
-------
10-6
Table 10-3
Percentage of LDDVs Requiring Traps Under
Various NOx and Particulate Standards
Stringent
Manufacturer
General Motors
Volkswagen
Nissan
Mercedes-Benz
Isuzu
Audi
Peugeot
Volvo
Sales-Weighted
1.0
NOx
100
83
89
100
82
84
96
100
95
1.5
NOx
81
78
88
96
77
77
87
93
82
2.0
NOx
73
67
75
79
67
62
64
71
72
1.0
NOx
61
6
33
80
9
28
55
58
48
Base
1.5
NOx
27
0
26
55
0
2
30
34
22
2.0
NOx
15
0
23
45
0
0
5
16
14
Industry-wide
Percentage
-------
10-7
Table 10-4
Percentage of LDDTs Requiring Traps Under
Various NOx and Particulate Standards
Stringent
Manufacturer
General Motors
Volkswagen
Nissan
Isuzu
Ford
Mitsubishi
Toyota
Toyo Kogyo
Sales-Weighted
1.2
NOx
98
79
89
85
82
92
55
81
95
1.7
NOx
85
78
87
74
80
89
55
80
83
2.3
NOx
77
78
87
74
80
89
55
80
77
1.2
NOx
63
14
37
26
16
47
0
15
56
Base
1.7
NOx
26
15
32
0
12
40
0
11
24
2.3
NOx
7
15
32
0
12
40
0
11
8
Industry-wide
Percentage
-------
10-8
than double from that required under a 1.5 g/mi NOx standard.
A 2.0 g/mi NOx standard would reduce the requirement for traps
by almost half. The stringent scenario require nearly all
LDDVs to be trap-equipped under a 1.0 g/mi NOx standard (95
percent) and 80 percent under a 1.5 g/mi NOx standard. The
trap fractions for LDDTs follow very closely those for LDDVs.
The one exception is under the base scenario and a 2.3 g/mi NOx
standard, when only 7.6 percent of LDDTs require traps,
compared to 13.9 percent for LDDVs.
As explained in Chapter 1, 100 percent of HDDs are
equipped with traps under the base scenario without averaging.
With averaging the percentage of traps would drop to about 70
percent. Under the stringent scenario, all HDDs are equipped
with traps with averaging.
C. 1995 Urban Diesel Particulate Emissions Under
Various NOx Standards
Having calculated industry-wide particulate standard
levels and percentages of traps required under each scenario,
the 1995 particulate emission factors for LDDVs and LDDTs can
be calculated. As explained in Chapter 2, the 1995 particulate
emission factors for LDDVs or LDDTs of a specific model year
are calculated using the age distribution of the in-use fleet,
the percentage of that model year's fleet equipped with traps,
the average non-trap emission level of those vehicles which are
equipped with traps, the particulate standard, and the annual
trap-failure rate (i.e., 1.5 percent per year).
The particulate emission factors for 1961-86 model year
LDDVs and LDDTs remain the same as those in Chapter 2, due to
the fact that all regulatory changes are assumed to occur in
1987. For the relaxed scenario, the emission factors for model
year 1987-95 are the sales-weighted industry-wide averages
shown in Tables 10-1 and 10-2. For the base and stringent
scenarios, the 1987-95 emission factors are essentialy the same
for all NOx standards since the presence of a standard
requiring control sets the emission level regardless of the
starting point. However, these particulate emission factors
are slightly different for each NOx standard, because both the
fleet-wide trap fraction and the average non-trap particulate
emission levels of those vehicles with traps change as the NOx
standard changes. When a trap-oxidizer system fails, the
particulate emission level that the vehicle reverts to is
different under each NOx standard. For vehicles with properly
operating traps, the emission factors are the same.
These 1995 particulate emission factors were combined with
the vehicle miles traveled (VMT) breakdown by model year and
-------
10-9
the diesel sales fractions (for both best estimate and worst
case sales) to yield weighted fleet-wide particulate emission
factors for each set of NOx standards. Again, this methodology
is fully described in Chapter 2. To obtain 1995 urban diesel
particulate emissions, the weighted particulate emission
factors were multiplied by the total 1995 VMT for each vehicle
class and by the urban fraction of VMT for each vehicle class
(i.e., 0.594 and 0.488 for LDVs and LDTs, respectively).
Table 10-5 presents the 1995 urban diesel particulate
emissions for best estimate and worst case diesel sales under
the three control scenarios and the three sets of NOx
standards. Table 10-6 shows the relative contribution of each
vehicle type to the totals of Table 10-5.
The results shown in Table 10-5 indicate that 1995 urban
diesel particulate emissions under the relaxed scenario do not
change substantially until the most stringent 1.0-1.2 g/mi NOx
standards are considered. Then, LDDV emissions increase by 50
percent as compared to a 1.5 g/mi NOx standard and LDDT
emissions increase by 79 percent as compared to a 2.3 g/mi NOx
standard. Total 1995 urban diesel particulate emissions
increase by 19 percent under the 1.0/1.2 g/mi set of NOx
standards as compared to the 1.5/2.3 g/mi set of Chapter 2.
Under the base scenario, 1995 LDDV urban diesel
particulate emissions increase only 3 percent under a 1.0 g/mi
NOx standard as compared to a 1.5 g/mi NOx standard. For the
other changes in the LDDV and LDDT NOx standards, the situation
is similar, with very little change in emissions occurring.
The change in emissions for the stringent scenario with changes
in NOx standards is only slightly larger (overall range of 8
percent).
The results of Table 10-6 are similar to Table 10-5 in
that the only NOx standards causing strong difference from the
main analysis are the 1.0/1.2 g/mi NOx standards under the
relaxed scenario. The contribution of LDDVs and LDDTs under
best sales estimate to total 1995 urban diesel particulate
emissions increases from 21 to 27 percent, and from 11 to 16
percent, respectively, under the more stringent set of NOx
standards. The other vehicle types (i.e., MDV/LHDV and HHDV)
decrease . their relative contribution with HHDV's share
decreasing the most (from 60 to 50 percent). The results are
similar for the worst case diesel sales situation.
Under the stringent scenario, there is little difference
among NOx standards (see Table 10-7) . Overall, the breakdown
under the stringent scenario is between that under the relaxed
and base scenarios.
-------
Table 10-5
1995 urban Diesel Particulate Emissions under
Various Best Estimate Diesel Sales, NOx Standards (metric tons)
Vehicle
Type
LDDV
LDDT
Total*
LDDV
LDDT
Total*
LDV NOX =
LOT NOX =
Relaxed
Scenario
36,800
21,800
137,000
84,000
33,900
207,300
1.0 g/mi
1.2 g/mi
Base
Scenario
19,700
11,700
72,800
43,800
18,000
108,600
LDV NOx
LOT NOx
Relaxed
Scenar io
24,600
14,100
117,000
Worst
55,300
21,800
166,500
= 1.5 g/mi
= 1.7 g/mi
Base
Scenario
19,100
11,500
72,000
Case Diesel
42,400
17,600
106,800
LDV NOx =
LOT NOx =
Relaxed
Scenar io
24,600
12,200
115,100
Sales
55,300
18,800
163,400
1.5 g/mi
2.3 g/mi
Base
Scenario
19,100
11,500
72,000
42,400
17,600
106,700
LDV NOx =
LOT NOx =
Relaxed
Scenario
22,200
12,200
112,700
49,600
18,800
157,800
2.0 g/mi
2.3 g/mi
Base
Scenar io
19,000
11,500
71,900
42,200 ?
17,600
106,500
Totals include MDV/LHDV and HHDDV emissions of 9,700 and 68,600 (relaxed), and 4,800
and 36,600 (Base) for best estimate sales and 19,100 and 70,200 (relaxed) and 9,300
and 37,400 (Base) for worst case sales. These are not shown since they are the same
regardless of LDV and LOT NOx standards.
-------
Table 10-6
Relative Contribution of 1995 urban Diesel
Particulate Emissions Under Various NOx Standards (percent)
Vehicle
Type
LDDV
LDDT
MDV/LHDV
HHDV
Total*
LDDV
LDDT
MDV/LHDV
HMDV
Total*
LDV NOX =
LOT NOX =
Relaxed
Scenario
27
16
7
50
100
40
16
10
34
100
1.0 g/mi
1.2 g/mi
Base
Scenario
27
16
7
50
100
40
17
9
34
100
LDV NOX •
LOT NOX •
Relaxed
Scenario
21
12
8
59
100
Worst
33
13
12
42
100
= 1.5 g/mi
= 1.7 g/mi
Base
Scenario
27
16
6
51
100
Case Diesel
40
16
9
35
100
LDV NOX =
LOT NOX =
Relaxed
Scenario
21
11
8
60
100
Sales
34
12
11
43
100
1.5 g/mi
2.3 g/mi
Base
Scenario
27
16
6
51
100
40
16
9 .
35
100
LDV NOx =
LOT NOX =
Relaxed
Scenario
20
11
8
61
100
31
12
12
45
100
2.0 g/mi
2.3 g/mi
Base
Scenario
26
16
7
51
100 ?
40
16
9
35
100
-------
Table 10-7
1995 urban Diesel Particulate Emissions under
the Stringent scenario (metric tons)
Vehicle Type
1.0/1.2
g/mi NOx
1.5/1.7
g/mi NOx
2.0/2.3 Relative Con-
g/mi NOx tribution (%)
Best Estimate Diesel Sales
LDDVs
LDDTs
MDDV/LHDDVS
HHDDVS
10,700
5,900
2,700
22,800
10,000
5,600
2,700
22,800
9,500
5,400
2,700
22,800
24 •
13
7
56
?otal
LDDVS
LDDTS
MDDV/LHDDVS
HHDDVs
Total
42,100
22,500
8,900
5,100
23,400
59,900
41,100 40,400
Worst Case Diesel Sales
20,900
8,500
5,100
23,400
19,700
8,100
5,100
23,400
57,900 56,300
100
36
14
9
41
100
Reduction From
Base (%)
48
53
44
!§.
43
51
54
46
37
46
o
I
M
to
-------
10-13
The decrease in urban emissions obtained under the
stringent scenario versus the relaxed scenario is between 56
and 72 percent for all vehicle types with the total decrease
being 65 percent. Compared to the base scenario, the stringent
scenario reduces total emissions by 43-46 percent, with the
change in each vehicle class being 37-54 percent.
Table 10-8 compares urban emissions under the three
control scenarios coupled with NOx standards of 1.0/1.2 g/mi to
previous diesel particulate studies. As can be seen,
projections for the base scenario are virtually the same as
that projected in 1979-80 for the same standards (controlled
scenario). However, even under the stringent NOx standards,
emissions under the current relaxed scenario are well below the
uncontrolled levels projected in previous analyses.
D. Cost Effectiveness
The cost effectiveness for LDDs and HDDs under the base
scenario was already determined in Chapter 9. Tables 9-1, 9-2,
and 9-3 of that chapter show the development of those
cost-effectiveness values..
In this study, cost effectiveness is the annualized cost
per vehicle divided by the annual emission reduction per
vehicle, both relative to the relaxed scenario and on a
fleet-average basis. The fleet-average annualized cost is a
straight-forward annualization of the fleet-average lifetime
costs using a 10 percent discount rate. The fleet-average
lifetime costs is a function of the lifetime costs of
trap-equipped vehicles of various sizes, the trap-equipped
fraction of each vehicle size category, and the relative sales
of each vehicle size category. The lifetime trap-oxidizer
system costs for different size vehicles and the relative sales
of these vehicle sizes were .described in Chapter 8. The
trap-equippped fractions of the LDDV and LDDT fleets were
estimated in Section IIB of this chapter.*
The determination of annual emission reductions was
explained in Chapter 2. Basically, the annual emission
reduction per vehicle is approximately the reduction in the
vehicle's emission rate at half-life (compared to the relaxed
It is assumed in this analysis that for LDDVs, large
vehicles are first equipped with traps, followed by medium
vehicles, and then small vehicles until the trap-equipped
fraction is met. Similarly, for LDDTs, full-sized LDDTs
are first equipped with traps, and then small LDDTs, until
the trap-equipped fraction is met.
-------
10-14
Table 10-8
Comparison of Current Urban Emission Estimates
Under Various NOx Standards* to Urban Emission
Estimates of Previous Studies
1995 Urban Emissions Under LDV
and LOT NOx Standards of 1.0
Scenario and 1.2 g/mi (metric tons)
Best Estimate Diesel Sales
1979-80 Uncontrolled 239,000
Relaxed 137,000
Base 72,900
1979-80 Controlled 71,000
Stringent 42,300
Worst Case Diesel Sales
1979-80 Uncontrolled 287,000
Relaxed 207,300
Base 108,600
1979-80 Controlled 85,000
Stringent 59,900
The NOx standard scenarios of LDV = 1.5/LDT = 1.7, and LDV
= 2.0/LDT = 2.3 g/mi are not shown because all "relative
reductions" are less than 4 percentage points different
than the 1.5/2.3 g/mi case.
-------
10-15
scenario) multiplied by the lifetime-average annual VMT. The
effect of trap failures is included in the vehicular emission
rate.
Table 10-9 compares the cost effectiveness of the various
LDD particulate control scenarios under different NOx
standards. Table 10-10 compares the cost effectiveness of the
various HDD particulate control scenarios. These tables
include the fleet-average annualized cost per vehicle, the
annual emission reductions per vehicle, and the urban cost
effectiveness (as described in Chapter 9) . Table 10-9 also
shows the trap-equipped fraction for LDDs (assumed to be 100
percent for HDDs).
Table 10-9 shows that under a given set of standards,
cost-effectiveness of LDDT control ranges between $1000-3000
per metric ton less than that for LDDVs, meaning that LDDT
control is slightly more cost effective. More importantly, the
table also shows that a given particulate scenario becomes less
cost effectiven with higher NOx standards. Control is
noticeably less cost effective when the NOx standards change
from 1.0/1.2 g/mi to 1.5/1.7 g/mi. This is due to the fact
that stringent NOx controls raise engine-out particulate levels
and increase the degree of control provided by adding a trap.
Under a given NOx standard, the stringent scenario is
moderately less cost effective than the base scenario. The
difference, which ranges between 7 and 27 percent, is to be
expected, since the additional traps being applied under the
stringent scenario are being applied to vehicles with lower
engin-out particulate levels, thus providing less control.
Trap costs, on the other hand, are relatively constant.
For heavy-duty diesels (Table 10-10), cost effectiveness
improves from the lighter to the heavier vehicles. While the
emission reductions for the various HDD classes are the same on
a percentage basis, they are greater for the heavier vehicles
on an absolute basis (due to greater absolute emission rates
and greater annual VMT). These effects more than compensate
for the increase in trap cost with vehicle size and the lower
urban VMT fraction of Class VII-VIII HDDs.
Without averaging, the base scenario for HDDs is less cost
effective than the stringent scenario. Without averaging, the
base scenario, like the stringent scenario, requires all HDDs
to be equipped with traps. It was assumed that traps under the
base scenario would only be as efficient as needed, but would
cost the same as traps under the stringent scenario. Thus, the
costs of both scenarios are the same, but the emission
reduction under the base scenario is less. Thus, the higher
cost-effectiveness value of the base scenario.
-------
Table 10-9
LDDV and LDDT Cost-Effectiveness Values Under Various
Particulate Control Scenarios and NOx Standards ($/metric ton)
Base Scenario
N0x=l .O/
1.2 g/mi
N0x=1.5/
1.7 g/mi
Percent Vehicles Equipped with Traps
LDDVs
LDDTS
48%
56%
22%
24%
Fleet Average Annualized Cost Per Vehicle*
LDDVs
LDDTs
$18.85
$20.77
$9.19
$8.83
NOX=2.0/
2.3 g/mi
14%
8%
$5.73
$2.81
Annual Emission Reduction Per Vehicle (metric tons)**
LDDVs
LDDTS
2.07 x 10-3
2.65 x lO-3
6.70 x 10-4 3.90 x 10~4
7.00 x 10-4 2.10 x 10~4
Cost Effectiveness ($/metric ton)
LDDVS
LDDTS
$9,100
$7,800
$13,700
$12,600
Urban Cost Effectiveness ($/metric ton)
LDDVs
LDDTS
$15,400
$16,100
$23,100
$25,800
$14,700
$13,400
$24,700
$27,400
Stringent Scenario
N0x=1.0/
1.2 g/mi
95%
95%
$35.27
$34.56
3.15 x 10-3
4.10 x 10-3
$11,200
$8,400
$18,900
$17,300
N0x=1.5/
1.7 g/mi
82%
83%
$30.79
$31.24
1.77 x 10-3
2.20 x ID"3
$17,400
$14,200
$29,300
$29,100
N0x=2.0/
2.3 g/mi
72%
77%
$27.85
$28.34
1.54 x 10-3
1.77 x 10-3
$18,100
$16,000
$30,800
$32,800
**
Based on estimated sales fractions of 29, 37, and 32 percent for large, medium, and
small LDDVs, respectively; trap-oxidizer systems fitted to these vehicles have an
average lifetime cost of $219, $234, and $266, respectively. Small and full-sized LDDTs
are estimated at 34 and 66 percent of sales respectively, with trap-oxidizers system
lifetime costs of $229 and $252, respectively.
Based on estimated annualized travel of 10,000 miles and 10,900 miles for LDDVs and
LDDTs, respectively; reductions are compared to relaxed scenario.
-------
10-17
Table 10-10
HDD Cost-Effectiveness Values
Under Various Control Scenarios ($/metric ton)
Base Scenario Stringent Scenario
Fleet-Average Annualized Cost Per Vehicle*
MDVs $ 84.26 $ 84.26
LHDVs $132.39 $132.39
HHDVs $228.46 $228.46
Annual Emission Reduction Per Vehicle (metric tons)**
MDVs 0.00747 0.0107
LHDVs 0.0180 0.0257
HHDVs 0.0575 . 0.0822
Cost Effectiveness ($/metric ton)
MDVs $11,280 $7,870
LHDVs $ 7,350 $5,150
HHDVs $ 3,970 $2,780
Cost Effectiveness, With Averaging ($/metric ton)***
MDVs $7,870 $7,870
LHDVs $5,150 $5,150
HHDVs $2,780 $2,780
Urban Cost Effectiveness, With Averaging ($/metric ton)
MDVs $16,200 $16,200
LHDVs $10,550 $10,550
HDVs $10,340 $10,340
* Assumes all HDDVs are equipped with traps, unless
averaging is used. Trap-oxidizer systems for MDVs,. LHDVs,
and HHDVs have an average lifetime cost of $472, $425, and
$1,516, respectively.
** Based on estimated annualized travel of 13,750, 23,300,
and 50,100 miles for MDVs, LHDVs, and HHDVs, respectively;
reductions are compared to relaxed scenario.
*** Averaging affects the base scenario only; average
fleetwide costs are estimated to decrease by 30 percent
(i.e., 70 percent of HDDs equipped with traps).
-------
10-18
With averaging, the cost effectiveness of the base and
stringent scenarios becomes the same. This is to be expected.
Trap costs and efficiency under the two scenarios are assumed
to be the same. The only difference between the two scenarios
is that only 70 percent of all HDDs are equipped with traps
under the base scenario, while all HDDs are trap-equipped under
the stringent scenario. However, this difference affects both
costs and emission reductions. Thus, cost effectiveness
remains constant.
In general, particulate control for HDDs is more cost
effective than that for LDDs when compared under the same
scenario.
VI. Comparison of Uncontrolled and Controlled HDD Emissions
Urban diesel particulate emissions from HDDs were
estimated for the relaxed control scenario in Chapter 2.
However, the corresponding values for a completely uncontrolled
HDD fleet were never derived in that chapter because such a
strategy is not considered to be a viable option.
Nevertheless, it is of interest to know what future urban HDDV
emissions would be if the fleet were totally uncontrolled so
that the benefits of the relaxed control scenario can be placed
in perspective.
As indicated in Chapter 1, uncontrolled HDDVs are
estimated to emit particulate at a rate of 0.7 g/BHP-hr
throughout their lifetime. Using the methodology of Chapter 2,
the resulting vehicular emission factors are shown in Table
10-11. Table 10-12 presents the 1995 urban particulate
emissions for the various HDDV scenarios using best estimate
sales projections. Relative to the uncontrolled scenario, the
relaxed scenario would reduce particulate emissions by about 12
percent, the base scenario by about 54 percent, and the
stringent scenario by about 72 percent.
-------
10-19
Table 10-11
Uncontrolled HDDV Emission Factors By Model Year (g/mi)
MDV
Model Year
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1961-80
Class IIB
0.946
0.944
0.941
0.937
0.934
0.939
0.935
0.952
0.955
0.958
0.977
0.994
1.007
1.026
1.021
1.014
LHDV
HHDV
Classes III-V Class VI Classes VII-VIII
1.307
1.300
1.294
1.284
1.277
1.271
1.275
1.268
1.272
1.263
1.267
1.258
1.259
1.263
1.265
1.256
1.677
1.672
1.668
1.660
1.656
1.651
1.661
1.648 .
1.662
1.650
1.638
1.644
1.675
1.682
1.667
1.651
2.554
2.551
2.536
2.533
2.523
2.516
2.520
2.531
2.543
2.549
2.557
2.564
2.573
2.583
2.593
2.597
-------
10-20
Table 10-12
HDDV Urban Emissions in 1995--Best Sales Estimates
Uncontrolled Relaxed
MDV/LHDV 11,100 9,700
HHDV 78,300 68,600
TOTAL HDD 89,400 78,300
------- |