BGDLATORY IMPACT ANALYSIS FOR THE
NATIONAL AMBIENT AIR QUALITY STANDARDS
FOR NITROGEN
ENERGY AND ENVIRONMENTAL ANALYSIS, INC.
1111 North 19th Street
Arlington, Virginia 22209
(703)528 1900
-------
REGULATORY IMPACT ANALYSIS FOR THE
NATIONAL AMBIENT AIR QUALITY STANDARDS
FOR NITROGEN DIOXIDE
Draft Report
Prepared For:
STRATEGIES AND AIR STANDARDS DIVISION
OFFICE OF AIR QUALITY PLANNING AND STANDARDS
U.S. ENVIRONMENTAL PROTECTION AGENCY ,
Research Triangle Park, N.C.
. By:
James H. Wilson, Jr.
and
Dale L. Keyes
ENERGY AND ENVIRONMENTAL ANALYSIS, INC.
1111 North 19th Street
Arlington, Virginia 22209
February 1981
-------
ACKNOWLEDGEMENTS
This report reflects the effort of many investigators over a three-year
period of time, during which the data bases, methodological approaches,
analytical assumptions, and the standards to be evaluated have all
undergone considerable change. The authors wish to thank the following
investigators for their participation in earlier phases of the study:
Robert Coleman, Barry Kumar, Robert Reid, and Paul Siebert. Contributors
to this document include Linda Carroll (editing), Vivian Daub (research
assistance), Bruce Henning (economic impact analysis), James Lent (general
model development and execution), Raymond O'Keefe (special dispersion
modeling analyses), Edward Pechan (Chicago case study), Mark Scruggs
(dispersion modeling advice), and Barry Wald (special dispersion modeling
analyses).
-------
TABLE OF CONTENTS
Page
ACKNOWLEDGEMENTS i
EXECUTIVE SUMMARY 1
1. INTRODUCTION 1-1
1.1 Purpose of the Regulatory Impact Analysis 1-1
1.2 Study Methodology 1-2
1.3 Report Organization ,r
2. CHARACTERISTICS OF NO FORMATION AND NO.,
EMISSION CONTROL. . 2-1
2.1 Sources of NO Emissions 2-1
x
2.2 Chemistry of N02 Formatiar 2-3
2.3 Methods of NO Emission Control 2-6
x
2.3.1 Combustion Modifications 2-7
2.3.2 Flue or Exhaust Gas Treatment 2-8
References for Section 2 2_n
3. AMBIENT N02 CONCENTRATIONS. 3-1
3.1 Temporal Patterns of Recorded Concentrations 3-1
3.1.1 Point Source Oriented Monitors 3-1
3.1.2 Mobile and Area Sources 3-8
3.2 Spatial Variation in NO- Concentrations 3-16
3.2.1 Regional Scale Variation in N0_
in Los Angeles 3-16
3.2.2 Microscale Gradients in N0_ Downwind
of a Roadway 3-22
3.2.3 Implications for the Regulatory
Analysis 3-25
3.3 Deriving Peak Hourly N02 3-25
3.3.1 Relationships between Continous and
Bubbler Data 3-25
11
-------
TABLE OF CONTENTS (Cont'd)
3.3.2 Relationships between Annual Averages
and Peak Hourly Values at Continuous
Monitors 3-31
References for Section 3 3-32
4. NATIONWIDE STUDY METHODOLOGIES 4-1
4.1 Point Sources 4-3
4.1.1 Emissions Data 4-3
4.1.2 Air Quality Modeling Approach 4-17
4.1.3 Control Options and Costs 4-40
4.1.4 Least Cost Model 4-44
4.1.5 Uncertainty Analysis 4-56
4.2 Mobile and Other Area Sources 4-60
4.2.1 Mobile and Other Area Source
Emissions Data 4-60
4.2.2 Air Modeling Approach 4-60
4.2.3 Control Options and Costs 4-66
4.2.4 Least Cost Modeling 4-74
References for Section 4 4-75
5. NATIONWIDE ANALYSIS RESULTS 5-1
5.1 Point Source Nationwide Analysis Results 5-1
5.1.1 One-Hour Standard Analysis 5-1
5.1.2 24-Hour Standard Analysis 5-6
5.1.3 Annual Standard Analysis 5-6
5.2 Mobile and Area Source Analysis Results 5-9
5.2.1 One-Hour Standard Analysis 5-9
5.2.2 24-Hour Standard Analysis 5-14
5.2.3 Annual Standard Analysis 5-17
iii
-------
TABLE OF CONTENTS (Cont'd)
5.3 Combined Results of Point Source and Mobile
and Area Source Analysis 5-17
5.4 Point Source Uncertainty and Plume Bouyancy
Sensitivity Analyses 5-23
5.4.1 Uncertainty Analysis 5-23
5.4.2 Plume Bouyancy Analysis 5-33
5.5 Mobile and Area Source Sensitivy Analysis 5-35
5.6 Caveats 5-41
References for Section 5 5-44
6. CHICAGO CASE STUDY 6-1
6.1 Selection of a Case Study 6-2
6.2 Ambient NO- Concentrations in Chicago 6-2
6.3 Chicago Emissions Inventories 6-3
6.3.1 Point Sources 6-3
6.3.2 Area and Mobile Sources 6-8
6.4 Modeling Specifications 6-11
6.4.1 Meteorological Assumptions 6-11
6.4.2 Receptor Network 6-12
6.4.3 NO -to-NO- Conversion 6-13
A £»
6.4.4 Calibration of RAM Results 6-13
6.4.5 Least-Cost Modeling 6-13
6.5 Treatment of Growth 6-14
6.6 Results 6-16
6.6.1 Estimated Air Quality Levels 6-16
6.6.2 Estimated Costs and Attainment Status 6-16
6.7 Conclusions 6-21
References for Section 6. . 6-22
lv
-------
TABLE OF CONTENTS (Cont'd)
NEW SOURCE CONTROLS 7-1
7.1 Cost of Meeting New Source Performance Standards . . . 7-1
7.1.1 Utility Boilers: Coal-Fired 7-1
7.1.2 Utility Boilers: Oil- and Gas-Fired 7-2
7.1.3 Industrial Boilers: Coal-Fired and Oil- and
Gas-Fired 7-3
7.1.4 Stationary Gas Turbines 7-4
7.1.5 Reciprocating Internal Combustion Engines . . . 7-6
7.1.6 Summary of NSPS Costs 7-7
7.1.7 Summary of NSPS Costs 7-10
7.2 Federal Motor Vehicle Control Program Costs 7-10
7.2.1 Light-Duty Vehicles 7-12
7.2.2 Light-Duty Trucks 7-21
7.2.3 Heavy-Duty Vehicles 7-27
7.2.4 Summary 7-33
References for Section 7 ., 00
/"JO
ECONOMIC IMPACT ANALYSIS OF THE ANNUAL STANDARD FOR N02 . . 8-1
8.1 Introduction 8-1
8.2 Stationary Source Impacts 8-1
8.2.1 Introduction 8-1
8.2.2 Economic Impact to the Industrial Sector:
Results 8-2
8.2.3 Economic Impact to the Industrial Sector:
Results 8-5
8.2.4 Economic Impact on the Commercial Sector . . . 8-6
8.2.5 Economic Impact on the Residential Sector . . .8-9
v
-------
TABLE OF CONTENTS (Cont'd)
Page
8.3 Mobile Source Control Impacts 8-10
8.3.1 Approach 8-10
8.3.2 Results 8-15
8.4 Urban and Community Impact Analysis for the
Annual Standard 8-20
References for Section 8 8-22
VI
-------
EXECUTIVE SUMMARY
This report presents a regulatory analysis of both the current NO-
ambient standard and a range of short-term NO- standard alternatives.
The current N02 standard is a 0.053 ppm annual average. The short-term
standards considered include the following one-hour and 24-hour average
levels:
Averaging Time Standard Levels
24-hours 0.06, 0.08, 0.10, 0.12, 0.14 ppm
One-hour 0.10, 0.15, 0.20, 0.25, 0.35, 0.50 ppm
This regulatory analysis provides nationwide estimates of the costs to
reduce NO emissions such that each alternative standard could be achieved
x
or approached as closely as possible. The regulatory analysis further
provides an accounting of the economic effects of the estimated control
costs on industries, individuals, and governments and on different
regions of the country.
The methodologies chosen for estimating nationwide control costs are
based on current knowledge of the characteristics of NO-, especially
those situations that are likely to cause high observed NO,, concen-
trations. Based on an identification of different types of NO- episodes,
two situations were modeled. One uses linear rollback with only emissions
from mobile and stationary area sources included in the emission inventory
This accounts for suppressed mixing episodes where an inversion limits
vertical mixing and where point sources with stack heights above the
inversion layer do not affect ground level concentrations. The ambient
concentrations of NO- during these episodes are likely to be recorded by
the existing network of NO- monitors. The second methodology accounts'
for situations where emissions from point sources can cause high NO-
-------
levels at the point of plume touchdown. Ground level NCL concentrations
for each significant source in the NEDS point source file of NO emitters
A
ar estimated with a dispersion model, considering source interaction
within plants. A modified ozone limiting approach is used to translate
NO into N00. In both the mobile and area source and the point source
*» ^
modeling methodologies, a least cost algorithm is used to estimate the
control costs needed to meet each alternative NO- standard. Growth is
explicitly considered only in the mobile and area source analysis.
Table 1 presents a summary of the nationwide results for both point and
mobile/area sources. For the annual standard, a large portion of the
costs is estimated to be incurred by small industrial sources, especially
those using reciprocating 1C engines for natural gas transmission or
other applications. In addition, almost $15 million of the total annual-
ized cost is attributed to performing NO related motor vehicle inspec-
A
tions and maintenance. All areas and/or plants should be able to attain
this standard given current control technologies.
For all the 24-hour and one-hour NO- standards considered, utilities
incur the major cost burden. For example, to reach or approach as
closely as possible the 0.25 ppm one-hour standard, utilities are esti-
mated to bear 65 percent of the total annualized costs for all sources.
Other industries with significant control costs at the more stringent
short-term standards include iron and steel, chemical manufacturing,
petroleum refining, and oil and gas extraction. Mobile source control
programs (inspection and maintenance and transportation control measures)
account for approximately 10 percent of the total annualized costs for
each short-term standard considered.
For each of the alternative short-term standards, some plants may not
reach attainment. For these plants, even the maximum controls will not
-------
TABLE 1
TOTAL ANNUALIZED COSTS OF CONTROL TO MEET
ALTERNATIVE N02 AMBIENT STANDARDS
do6 $)
Standard (ppm) Annualized Cost
Annual
0.053 54
One-hour
0.10 10,540
0.15 5,710
0.20 1,570
0.25 750
0.35 100
0.50 30
24-hour
0.06 3,690
0.08 1,530
0.10 540
0.12 270
0.14 160
-------
reduce NO emissions enough to bring their NO- contributions down to the
X £
standard level. The number of nonattaining plants ranges from four (for
a 0.50 ppm one-hour standard) to more than 400 (for a 0.10 ppm one-hour
standard). Likewise, the number of counties unable to attain the standard
increases as the standard becomes more stringent.
By 1990, all counties are predicted to attain the 0.35 ppm one-hour NO-
standard while one county may not be able to attain the 0.25 ppm one-hour
level. As many as 120 counties may not be able to reduce NO,, concentra-
tions below 0.10 ppm.
The uncertainty in the point source control estimates for meeting alter-
native one-hour NO- standards was investigated through a series of
sample plant simulations. The results of these simulations showed that
the most probable annualized cost estimate for the one-hour standards is
10 to 20 percent higher than the nationwide "point" estimate of these
costs. The use of sulfur dioxide scrubbers on powerplants also was
shown to increase maximum ground level NO concentrations by up to 70
X
percent and the effect of complex terrain (though not considered explic-
itly in this study) likewise will make attainment more difficult and
more costly for point sources. The need to control new point sources
beyond NSPS level was not analyzed but could elevate costs even further.
A sensitivity analysis of the mobile and area source results showed that
changes in the assumptions about new source growth and the effect of VOC
control on NO- levels can significantly affect the cost and attainment
status estimates. The effect on ambient N0_ levels of light-duty diesel
vehicle waivers to the 1981-1984 NO emission standards is negligible.
A
The nationwide analysis is augmented by a separate and more detailed
modeling study of the control costs to attain alternative short-term NO-
standards in the Chicago AQCR. Comparing the results of the Chicago
case study with those obtained with the nationwide methodologies reveals
that the area and mobile source cost estimates are in agreement, but
-------
that the point source costs obtained with the nationwide approach for
Chicago are significantly greater than those estimated in the case
study. The disparity in these cost estimates can be traced to differences
in the emission inventories in the two studies, but further investigation
is required to determine the specific cause of these differences.
An economic analysis was only performed for the annual average NCL
standard. Based on the nationwide analysis, it was determined that none
of the initial investment costs in the industrial sector exceeds six-
tenths of one percent of the expected capital budgets of the industry.
Therefore, because fluctuations in capital expenditures of one percent
are considered normal, no industry should experience difficulty in
raising the capital to finance the necessary control equipment. None of
the estimated percentage increases in cost of goods sold for an industry
exceeds one-quarter of one percent of the total costs of goods sold.
For the commercial and residential sectors, no significant economic
effects are expected either.
Mobile source control costs to meet a 0.053 ppm annual standard are all
due to motor vehicle inspection and maintenance (I&M). These costs are
only incurred in urbanized areas that need ISM programs in order to meet
the standard. While costs to motorists for vehicle inspection and
repair are low, they tend to put more of a burden on low income house-
holds, which have older cars and are more likely to need repair.
-------
1. INTRODUCTION
1.1 PURPOSE OF THE REGULATORY IMPACT ANALYSIS
Section 106 of the Clean Air Act requires that the EPA Administrator
"complete a thorough review of the criteria published under section 108
and the national ambient air quality standards promulgated under this
section and shall make such revisions in such criteria and standards and
promulgate such new standards as may be appropriate...." In accordance
with this requirement, a review of the National Ambient Air Quality
Standard (NAAQS) for nitrogen dioxide (NO-) is currently being per-
formed. As part of this review procedure, this report analyzes alter-
native standard levels and averaging times and the potential costs of
control to meet those standards. In addition, the economic effects of
controlling to meet these alternative standards are analyzed.
A report of this type is mandated by Executive Order 12044, which calls
for careful analysis of available regulatory alternatives, including the
current standard, for new significant regulations and regulations already
issued. The Environmental Protection Agency presented its plan to
implement the Executive Order in the Federal Register (43 FR 29891).
The Regulatory Impact Analysis (RIA) presented here is designed to
fulfill the requirements of this plan.
The current national primary and secondary ambient air quality standard
3
for N0» is 100 micrograms per cubic meter (p/m ) (0.053 parts per million
[ppm]) annual arithmetic mean (40 CFR 50.1). The costs to control
oxides of nitrogen (NO ) to meet this standard and the subsequent economic
A
effects of these costs are analyzed in this report. In addition, the
following alternative standards are analyzed:
1-1
-------
Averaging Time Standard Levels (ppm)
One-hour 0.10, 0.15, 0.20, 0.25, 0.35, 0.50 ppm
24-hour 0.06, 0.08, 0.10, 0.12, 0.14 ppm
The form of the annual average NO- standard is such that an area's
attainment status can be determined by simply comparing the highest
recorded annual average from the past three years with the standard. If
this value is less than 0.053 ppm, then the area is in attainment. The
assumed form of the standard for the one-hour and 24-hour averaging
times is more complex. The form analyzed here is based on the current
ozone standard. In general, the average number of days per year above
the level of the standard must be less than or equal to one. For a
one-hour averaging time, an equivalent expression is the second highest
value that does not occur on the same day as the highest value (i.e.,
the "second high daily maximum hour").
1.2 STUDY METHODOLOGY
There are two primary objectives of this RIA. One is to produce a
quantitative estimate of the cost of attaining alternative NO- standards.
The second objective is to assess the economic effects of these control
costs. This economic analysis includes an Urban and Community Impact
Analysis (UCIA) that examines the economic effects of an NO- NAAQS on
employment, income, the cost of living, and population subgroups.
The methodology used for this RIA includes both a nationwide cost analysis
and a case study of control costs in the Chicago Air Quality Control
Region (AQCR). The Chicago modeling is compared with the nationwide
cost study estimates for the Chicago AQCR to determine the reasonableness
of the nationwide costs.
1-2
-------
The first step in estimating nationwide control costs to meet an alter-
native standard level is to project how many areas will not attain that
standard. For the purposes of this analysis, attainment status is
projected for 1985 and 1990. Based on an identification of different
types of NO- episodes, two situations are modeled here. One is a sup-
pressed mixing episode where an inversion limits vertical mixing of
emissions from mobile and stationary sources with short stacks. Point
sources with stack heights above the inversion layer do not contribute
to these episodes. Therefore, the modeling approach used to characterize
this type of episode is linear rollback with only the emissions of
mobile and stationary area sources included in the emissions inventory.
For areas projected not to attain the alternative standards in either
1985 or 1990, control measures are assumed to be added until the standard
is attained, using cost minimization as the decision criterion.
The second type of NO- episode modeled in the nationwide analysis is the
point source episode, in which emissions from point sources can cause
high NO- levels at the point of plume touchdown. The methodology used
for this situation is atmospheric dispersion modeling ground level NO
X
concentrations from NO emitters in the NEDS point source file, with
A
plume interaction from sources within plants considered explicitly.
Conversion of NO to NO- is estimated by assuming that the oxidation of
A £»
nitric oxide (NO) is limited by the ambient 0- concentration in the
vicinity of each plant. A least-cost algorithm then is used to estimate
the control costs needed to meet each alternative NO- standard.
Total costs to meet each alternative standard are estimated by summing
the costs to mobile and area sources and point sources.
The Chicago case study considers the simultaneous interaction of all
point, area, and mobile sources in the AQCR. A mutiple source dispersion
model (RAM) is used to estimate one-hour NO- concentrations at receptors
1-3
-------
throughout the AQCR. Controls then are applied in a cost-minimizing
manner until the standard is met or until all control options are exhausted
The. costs of control from this detailed modeling effort then are compared
with costs estimated for the Chicago AQCR in the nationwide analysis to
provide an estimate of their reasonableness.
An uncertainty analysis of the nationwide point source cost estimates
also is performed. Probability distributions for each of 10 parameters
affecting the cost estimates are used for a sample of plants from the
NEDS point source file to simulate how uncertain the cost estimates are.
These probability distributions are based on values reported in the
literature and "intelligent estimates." A separate analysis considers
the effect of flue gas scrubbing for sulfur dioxide on ambient NO
concentrations.
A sensitivity analysis is performed for the mobile and area source
analysis results using variations in growth rates, peak to mean relation-
ships, and control assumptions.
Using the results of the nationwide cost analysis, the economic analysis
for the current annual standard focuses on the distribution of costs
resulting from mobile source inspection and maintenance (I&M) programs
and the industry-by-industry effects of NO control costs. The distribu-
X
tion of I&M costs is examined for the burden to both State and local
governments and to individual motorists. Stationary source costs are
presented in terms of their relationship to the capital availability of
the firm and the effects on the prices of their goods.
The Urban and Community Impact Analysis (UCIA) examines economic effects
on different population groups within an urban area or a community.
Therefore, the UCIA provides an estimate of how increased NO control
A
costs can affect minorities, employment, personal income, and the cost
of living.
1-4
-------
1.3 REPORT ORGANIZATION
This report is organized in eight sections. Following this introduction,
Section 2 discusses the characteristics of NO-, the major NO sources,
Z* X
ambient NO* formation mechanisms, and current NO control principles.
Section 3 presents evidence on current NO- levels and concentration
patterns based on available ambient monitoring data. This section
includes an analysis of the relationship between peak one-hour and
annual mean N0_, spatial variations in NO-, and a comparison of con-
tinuous versus bubbler data for co-located monitors. In addition,
ambient NO- concentrations near point sources are compared with those
measured near mobile and area sources.
Section 4 presents the nationwide study methodologies for both point
sources and mobile and other area sources. This discussion includes the
methodologies used in:
Processing the NEDS emission data
Relating emissions and ambient air quality
Estimating control costs and effectiveness values for NO
control
Selecting control strategies based on cost-minimization
Analyzing uncertainty.
Section 5 presents the results of the nationwide analyses for both point
sources and mobile and other area sources. These results show the costs
incurred and the attainment status for each alternative standard considered
The costs presented are the incremental costs above those for currently
mandated NO controls such as the Federal Motor Vehicle Control Program
X
(FMVCP) and New Source Performance Standards (NSPS's). However, the
effects of growth are considered only for mobile and area sources.
1-5
-------
Section 6 presents the methodology and results of the Chicago case
study. In this section, the costs predicted for the Chicago AQCR using
the nationwide study methodology are compared with the costs predicted
for the Chicago AQCR using the more detailed modeling procedure. In
addition, a least cost and a universal Reasonably Available Control
Technology (RACT) solution to meeting the alternative NO- standards are
compared.
Section 7 provides estimates of the costs of current NO emission stan-
X
dards which will apply regardless of whether or not there is a change in
the N02 NAAQS. These are
standards and the NSPS's.
the NO. NAAQS. These are costs of control to meet the FMVCP emission
Section 8 presents an economic analysis of the current annual average
NO- standard and summarizes the urban and community impact analysis
results for the same standard. The Appendices are included in a
separate document.
1-6
-------
2. CHARACTERISTICS OF NO FORMATION AND NO
EMISSION CONTROL
This section summarizes the sources of NO , the primary chemical mechanism
X
by which NO- is formed in the atmosphere, and the methods employed to
control the emission of NO . A more detailed description of these
topics together with an extensive bibliography can be found in the N0~
Criteria Document (U.S. EPA, 1979a) and in the NO Control Techniques
X
Document (U.S. EPA, 1978).
2.1 SOURCES OF NO EMISSIONS
x
By far the most significant source of NO emissions is the combustion of
fuel. High temperatures and the turbulent mixing conditions that accompany
fuel combustion are conducive to the oxidation of both atmospheric and
fuel-bound nitrogen. The predominant oxidation product is nitric oxide
(NO); small amounts of nitrogen dioxide (N0») and much smaller amounts
of other oxides (N^O,., etc.) also are formed. These are known collectively
as nitrogen oxides (NO ).
X
Table 2-1 summarizes the estimated nationwide NO emissions in 1976. Of
X
the estimated 23 million metric tons, mobile sources accounted for 44
percent and stationary source fuel combustion for 51 percent. The
remaining 5 percent were due to industrial processes (3 percent), solid
waste disposal (less than one percent), and miscellaneous sources, such
as forest fires (one percent). The mobile source category was dominated
by highway vehicles, primarily autos and small trucks. Stationary
source fuel combustion emissions were divided between electric utilities
(56 percent) and industrial fuel combustion sources, such as boilers,
process heaters, furnaces, and kilns (38 percent). The remainder accrues
to the residential, commercial, and institutional category.
2-1
-------
TABLE 2-1
NATIONWIDE EMISSION ESTIMATES, 1976
(10 metric tons/year)
Source Category
Transportation
Highway Vehicles
Non-Highway Vehicles
Stationary Fuel Combustion
Electric Utilities
Industrial
Residential, Commercial, and Institutional
Industrial Processes
Chemicals
Petroleum Refining
Metals
Mineral Products
Oil & Gas Production Marketing
Industrial Organic Solvent Use
Other Processes
Solid Waste
Miscellaneous
Forest Wildfires and Managed Burning
Agricultural Burning
Coal Refuse Burning
Structural Fires
Miscellaneous Organic Solvent Use
TOTAL
NO
x
10.1
7.8
2.3
11.8
6.6
4.5
0.7
0.7
0.3
0.3
0
0.1
0
0
0
0.1
0.3
0.2
0
0.1
0
0
23.0
Percent
44
34
10
51
27
20
4
1
1
0
0
0
0
0
1
0
0
0
0
100
SOURCE: U.S. EPA. 1979b. 1976 National Emissions Report.
EPA-40/4-79-019. National Air Data Branch, Monitoring
and Data Analysis Division, OAQPS. Research Triangle Park,
N.C. August.
2-2
-------
This is not to suggest that the impact of these types of sources is
necessarily a simple function of their aggregate emission level. Indeed,
the dispersion characteristics of plumes emitted from a source as well
as the chemical and meteorologic environment through which the plume
passes will determine the ultimate ground level concentration of NO-
from the mass of NO emitted. For example, relatively low levels of NO
X *»
are emitted by stationary industrial internal combustion (1C) engines
nationwide, but the low stack heights typical of 1C engine applications
may produce high ground level concentrations of NO and, under conditions
of high ambient 0,, concentrations, high levels of N0».
2 . 2 CHEMISTRY OF N0£ FORMATION
As noted above, nitrogen oxidized during fuel combustion is emitted
primarily as NO. Only about 5 percent of the NO in the combustion
products is in the form of NO-. In general, the oxidation of NO to NO-
in the atmosphere occurs by two mechanisms: 1) oxidation of NO to NO-
in the presence of atmospheric oxygen mixed into the plume (thermal
oxidation) and 2) oxidation of NO to NO- in the presence of oxidizing
agents such as ozone, hydroxyl radicals, or organic peroxyl radicals,
again introduced as the plume mixes with the ambient air.
Chemically, the first mechanism is:
2NO
In this reaction, the rate of NO- production is proportional to the
square of nitric oxide concentration; nitrogen dioxide production is
immediate for NO concentrations greater than about 100 ppm. These high
concentrations only occur very near the point of exhaust. As the NO is
diluted to concentrations below 100 ppm, the rate of conversion decreases
to a point where, at one ppm, the direct reaction with oxygen becomes
unimportant. It is believed that the conversion of NO to NO- through
this mechanism is limited to roughly 10 percent of the initial NO
2-3
-------
concentrations. However, for high initial NO concentrations and high
excess air or moderate levels of mixing, Calvert suggests that this
mechanism could result in ratios of NCL to NO of 25 percent (Calvert,
1973).
The second major pathway features oxidizing agents, such as ozone and
hydroxyl radicals, and results in the rapid conversion of NO to N0~. In
the absence of photochemical activity, NO- formation due to ozone titra-
tion is stoichiometric. This appears to describe conditions within an
individual point source plume of NO as it disperses and entrains ambient
A
0- and within line source plumes as they disperse from highways, especially
in late afternoon (when 0~ is high). This is depicted chemically as:
NO + 0 * N0 + 0
However, where multiple plumes of NO and volatile organic compounds
A
(VOC) interact in the presence of ultra-violet (UV) light, an equilibrium
set of reactions between N02, 0~, and VOC is established:
N0? Formation
NO + 0,, » NO,, + 0,,
NO + H02 -> N02 + HO
NO + R02 -> N02 + RO
NO + RC03 -> N0£ + RC02
NO,, Transformation
_2 .
N02 -» NO + 0
M
OH -> HN0
M
3
RCO- + NO -» RCO NO
j £ 3
2-4
-------
Major Coupling Reaction
0- + 0 -» 0-
where
M = a molecule that absorbs vibrational energy
hy = ultra-violet energy
During days when UV radiation levels are high, the late morning and
afternoon levels of NO- are set by these competing reactions.
Meteorological conditions also play an important role. Weather condi-
tions obviously determine the amount of sunlight available to drive the
photochemical reactions. In addition, mixing depth and wind speed will
influence the degree of atmospheric dispersion. Thus, the type and mix
of NO sources, the presence of VOC sources, and the amount of solar
radiation all combine to create distinct types of situations that are
conducive to high short-term NO- levels. These situations are further
stratified by meteorologic conditions.
SAI, Inc. has identified four types of NO- episodes in the Los Angeles
Basin, based on the NO- formation mechanism in combination with the
meteorological setting (Hayes et al., 1980):
Suppressed Mixing Ground-based atmospheric inversions
inhibit dispersion of NO emitted from sources with short
stacks. Plumes from tall stacks typically disperse above the
inversion and do not contribute to ground level concentrations
except, infrequently, during inversion break-up events (fumiga-
tion) . Fog or low cloud ceiling conditions associated with
ground-based inversions reduce solar radiation and inhibit
photochemical production of NO-. High levels of N0« are thus
the result of extremely high NO levels and the initial 10
percent or so conversion of NO to NO-.
Photochemical Generation During days when skies are clear
but dispersion is inhibited by an inversion aloft, high con-
centrations of VOC and NO from numerous sources combine
A
2-5
-------
photochemically to produce high concentrations of NO-. The
highest NO- levels typically occur in the morning before the
NO-NO--VOC-0- steady state is established in early afternoon.
The extent to which emissions from tall stacks contribute to
these levels depends on the height of the plume in relation to
the inversion lid.
Mobile Source Titration -- When 03 levels are high during the
late afternoon period, 0- will titrate the fresh NO emitted by
mobile sources during rush hour.
Stationary Source Titration Individual plumes from elevated
stationary sources also may produce high ground level concen-
trations of NO- through the titration of NO by entrained 0^,
as noted previously. Actual photochemical generation of
oxidized organics which subsequently react to form NO- is not
believed to be significant within individual plumes. Ground
level concentrations of NO- may, on occasion, be substantial
if the elevated plume impacts the ground shortly after release,
as is likely to happen under A stability. However, there is
probably a trade-off between rapid ground impact and complete
plume mixing, such that the maximum reaction of plume NO with
ambient 0» may not have occurred by the time the plume strikes
the grouna. This is discussed further in Section 4.1.2.
These generic types of NO- formation may not be applicable everywhere,
but they do illustrate the variety of problems that may be encountered
and the need to develop multi-faceted control strategies for individual
regions.
2.3 METHODS OF NO EMISSION CONTROL
There are two approaches to NO control: modify the process by which
A
NO is produced during combustion, or capture it after combustion but
X
before release from the stack. General principles are outlined here;
identification of specific control techniques appears in Sections 4.1.3
and 4.2.3.
2-6
-------
2.3.1 Combustion Modifications
Because high temperatures and turbulent mixing greatly enhance the re-
action of oxygen with gaseous N9 or fuel-bound nitrogen, several tech-
niques are aimed at reducing flame temperatures, modifying the mixing of
air and fuel through burner configuration changes, or reducing the
amount of available oxygen.
The flame temperature has the most dramatic impact on the NO formation
A
rate and is therefore the most important combustion modification para-
meter. The flame temperature can be reduced by 1) shifting the air-fuel
ratio away from the stoichiometric value in the primary flame zone, 2)
diluting the incoming air with recirculated flue gas, 3) reducing the
temperature of the incoming air, or 4) injecting a coolant such as water
or steam. Each of these modifications has been found to reduce NO from
X
combustion systems; their applicability depends on the combustion para-
meters of each furnace. NO reductions up to 60 percent can be achieved
A
with the application of a combination of these modifications to existing
systems.
Excess air is an operationally controllable parameter in most combustion
processes; however, the reduction of oxygen in the combustor air results
in excess carbon monoxide (CO) when burning gas, or excess smoke when
burning oil. NO emissions can be reduced by lowering excess air to the
A
level just above that of the onset of excess CO or smoke. Reducing
excess air from 25 to 15 percent normally results in lowering NO by an
A
equivalent percentage with an additional benefit of reducing fuel require-
ments . Equipment for modifying the existing combustion system for low
excess air (LEA) operation consists of improved oxygen measuring systems
and controls. Mobile source control techniques include reducing the
air:fuel ratio, although this approach is tempered by the potential for
undesirable increases in unburned hydrocarbons and CO. Exhaust gas
recirculation is also employed, primarily as a way to lower combustion
temperatures.
2-7
-------
2-3.2 Flue or Exhaust Gas Treatment
Gas treatment post-combustion involves the removal of NO primarily by
X
chemical means. The most popular approach for stationary sources (and
the only approach for mobile sources) employs noble metal catalysts to
chemically reduce NO . Both NO - selective and nonselective systems
A X
with removal efficiencies up to 90 percent are available for stationary
sources and have been employed routinely in Japan and on a limited basis
in the United States. Combined SO /NO systems are under development
X X
for applications to oil- and coal-fired boilers. Mobile source techniques
incorporate the reduction catalyst in a three-way reduction/oxidation
system for all three pollutants.
Flue gas control systems are also available for process sources such as
nitric acid plants. Some are based on a catalytic reduction principle,
while others use alternative methods such as molecular sieve adsorption
or wet scrubbing.
2-8
-------
REFERENCES FOR SECTION 2
Calvert, I.G. 1973. Interactions of Air Polltants - Proceedings of the
Conference on Health Effects of Air Pollution. National Academy of
Sciences, Washington, B.C. October.
Hayes, S.R.; Yocke, M.A.; Hogo, H.; Johnson, J.A. 1980. Evaluation of
Requirements for the Control of Ozone and Nitrogen Dioxide in the
South Coast Air Basin (Draft). Systems Applications, Inc. San
Rafael, CA. October. (Prepared for the National Commission on Air
Quality, Washington, B.C.)
U.S. Environmental Protection Agency. 1978. Control Techniques
for Nitrogen Oxides Emissions From Stationary Sources - Second
Edition (EPA-450/1-78-001). Emission Standards and Engineering
Division, OAQPS. Research Triangle Park, NC. January.
U.S EPA. 1979a. Air Quality Criteria for Oxides of Nitrogen (Draft).
Environmental Criteria and Assessment Office, Office of Research
and Development. Research Triangle Park, NC.
U.S. EPA. 1979b. 1976 National Emissions Report (EPA-40/4-79-019).
National Air Data Branch, Monitoring and Data Analysis Division,
OAQPS. Research Triangle Park, NC. August.
2-9
-------
3. AMBIENT N02 CONCENTRATIONS
This section reviews selected ambient monitoring data around point
sources and in the vicinity of mobile and area sources. Diurnal, seasonal,
and spatial variations in the NO and N0_ concentrations at these sites
are discussed. Second, a method for estimating peak one-hour concentra-
tions is developed from the relationship between the N02 levels recorded
by continuous monitors versus bubblers and the relationship between high
one-hour NO- levels and annual averages.
3.1 TEMPORAL PATTERNS OF RECORDED CONCENTRATIONS
Although the regulatory analysis considers both annual and shorter-term
standards, hourly N0« levels are emphasized here. Recorded annual
levels are almost universally low, with only two urban areas now re-
cording violations of the current NAAQS. Hourly values, on the other
hand, show considerable spatial and temporal variability.
3.1.1 Point Source Oriented Monitors
The observations reported in this section are based on maximum hourly
NO and N09 concentrations measured at continuous monitoring stations
A. jL
around selected American Electric Power Service Corporation (AEP) plants.
Continuous NO data at each of these sites have been collected using the
chemiluminescent measuring technique. The monitoring stations were
sited to measure the maximum ground-level impact of nonreactive emis-
sions from the sources based on diffusion modeling.
3-1
-------
3.1.1.1 Diurnal Variations
Figures 3-1 through 3-3 show the diurnal variation in the NO, NO-, and
NO concentrations observed at two monitors in the vicinity of an AEP
4
plaftt for different periods of the year. The values in these curves are
averages for each hour over a period of three months.
It is difficult to generalize the diurnal variations in the ground-level
concentrations of a pollutant on any given day. But with concentrations
for an hour of the day averaged over an extended period, anomalies are
smoothed and the emerging trends can then be generalized. A pronounced
NO bulge in the mid-morning, significantly higher than that in the
X
evening, can be seen in almost all these curves. Another interesting
observation is the occurrence of N0« bulges, less pronounced than those
for NO , but again in the mid-morning and evening hours. Evening peaks
A
in this case are slightly higher than the morning levels. Overall, NO
X
appears to be considerably more variable than NO-.
The highest hourly N09 and NO concentrations at one of these sites have
-------
FIGURE 3-1
AVERAGE DIURNAL VARIATIONS IN THE AMBIENT NO, N02 AND NOX
CONCENTRATIONS FROM STATIONARY SOURCES
CREEK
3C o.o«
a.
CL
o.oi
cc
>-
2
Ul
u
g
o
X
p
4.84
O.C1
o.aj
a.ci
-------
FIGURE 3-2
AVERAGE DIURNAL VARIATIONS IN THE AMBIENT NO, NO2. AND NOX
CONCENTRATIONS FROM STATIONARY SOURCES
TAWSERS CR3IK
= o.«* PHTERSsJSS C!3
S: '
s
Max
8
e.ct.
S e 7 9 4 IU 11 13 te 14 1C 1« 17 1C 19 *9 it a »
f«X^ 0? DAV CSCS^S2 127S - FE8SUARY 1S77
K3JS1.V AVSRAg; W32 K£3 A6G ^JS CGJCtKTRATIC^ CPPK3
SOURCE; American Electric Power Service Corporation
3-4
-------
FIGURE 3-3
AVERAGE DIURNAL VARIATIONS IN THE AMBIENT NO, N02 AND NOX
CONCENTRATIONS FROM STATIONARY SOURCES
a.
TANNERS CREEK
«.-. PETERSBURG (13
o.«*
t 0.04
z
LU
U
g ..«
9.09
a
(N
g
0.0
01 a t 4 t « 7 * » 10 it 13 I* 14 1* 1« 17 t* 1* 20 21 22 21
HOUR Or DAY MARCH 1377 - MAY 1977
HOURLY AVERAGE NQ2 NO AND NOX CONCENTRATION (PPM)
E ««
a. -
8 *
^ 0.«4-
§
y
5 ..w.
x
§ ...
TANNERS CREEK
; 7 ELIZA8ETHTOWNC53
e.o
.
/,
i ( j
^
/
/
f
/
f
/
~~"
/
«J
-J
-J
OX
\
J02
43
\
V
~\
s.
V
\
s.
X
*-,
S,
^
*»^
^M*
MOM
M^MM
^^M
^^W
RM*^">"
m*
^-^i
^-"
^^IMM
X
^
"
V.
~-~,
_
!!"
+DL.n? OF DAY K/WCH 1S77 - KAY 1977
HOURLY AVEKAS2 KD2 KO AND NOR COKCENTRATIOf4 CPPM)
SOURCE: American Electric Power Service Corporation
3-5
-------
TABLE 3-1
MAXIMUM NO AND NO CONCENTRATIONS AT
AEP - TANNER^ CREEK PLANT
MONITORING SITE: PETERSBURG (1)
u>
I
5 Highest
1-hr NO
Concentrations
5 Highest
1-hr NOX
Concentrations
Date Hour Ending (ppm) Date Hour Ending
June 5, 1977 16 0.087 March 4, 1977 14
April 15, 1977 13 0.074 May 10, 1977 11
July 23, 1977 14 0.066 October 30, 1976 9
May 25, 1977 12 0.062 December 5, 1975 11
October 3, 1976 14 0.062 February 19, 1977 6
(ppm)
0.23
0.142
0.126
0.119
0.114
Source: Data from American Electric Power Service Corporation.
-------
trapping. But plume mixing is limited and the NCL concentrations in the
morning are mainly governed by the initial oxidation of NO, that is, the
thermal oxidation of NO. As can be seen from the diurnal curves, the
ratio for this time of the day is at its lowest.
The observed peak NO. excursions in the afternoon, upon analysis of the
corresponding meteorological data, were found to be associated with
unstable atmospheric conditions that result in occasional high ground-
level NO concentrations in the afternoon. The high N00 concentrations
X ^
are a result of both relatively high NO levels and a greater fraction
of NO in the plume converted to NO- . The latter is driven by good
atmospheric mixing, which creates good conditions for oxidation by 0~.
The highest NO-/NO ratios are found in the late afternoon or evening,
^ X
with the upper limit to this ratio determined by the level of 0.
From this analysis, the following conclusions emerge:
NO and NO- maxima occur during different periods of the day;
while the NO maxima show a trend of mid-morning excursions,
the NO- peaks occur most frequently in the afternoon and
evening .
Diurnal variations in NO concentrations are much more pro-
nounced than those in NO- levels .
The NO-/NO ratio varies from a low in mid-morning to a high
in the early evening.
The low levels of even the peak N0~ concentrations are also noteworthy.
This suggest that either elevated sources do not contribute substan-
tially to ground-level concentrations of NO-, or that siting monitors in
order to detect peak NO- is extremely difficult. Based on the results
described in Section 5, the latter explanation is favored.
3-7
-------
3.1.1.2 Seasonal Variation
Based on a review of the highest monthly NO and NO- data, the following
X ^
observations can be made:
N0x concentrations show significant seasonal trends with higher
peak values occurring in the colder months
No consistent seasonal variation in NO- concentrations is
observed.
3.1.1.3 Background Levels of N02
The AEP Rockport station was a pre-construction monitoring site and thus
provides data on background levels. The diurnal variation with hourly
values averaged over a quarter and the monthly peak one-hour NO and NO-
A £
values are shown in Figure 3-4.
Peak N0~ concentrations at this site averaged around 0.04 ppm and have
been observed to be as high as 0.07 ppm. Annual average NO- concen-
trations were on the order of 0.01 ppm. However, there are several
small powerplants west and southwest of this site within a distance of
25 to 50 kilometers, making these values suspect as true measures of
natural background NO-.
3.1.2 Mobile and Area Sources
The spatial and temporal patterns of ambient NO- levels resulting from
mobile and area source emissions differ somewhat from those due solely
to point source emissions. Our knowledge of these patterns is based on
the continuous record of ambient NO- at those monitors which, due to
their location, are unlikely to reflect point source contributions. In
urban areas, these include most continuous monitors.
Washington, D.C., a mobile and area source-dominated urban area (over 70
percent of total NO emissions come from mobile and area sources),
A
3-8
-------
FIGURE 3-4
AVERAGE DIURNAL VARIATIONS IN THE AMBIENT NO, NO2 AND NOX
CONCENTRATIONS FROM STATIONARY SOURCES
TANKERS CREEK
ii.
a
W
o **
t-
£ C.S*
tu
1
1
a
8 *-«
f
F
^
j
^r
rn
!
,-*
^
-H
X
**»
=±».
-^*
«
=
.^_
M
^ip*»"
^__
^MMI
^
y
»«
I
« t Z 'S » t> a 7 U « Ib it i£ ia 14 IU IO U W 16 £J U C-i i.
«3UR OF DAY «TUHE 1S77 - AUSUST 1S77
C2URLY AVERAG2 N02 NO AN-3 NOX CONCENTRATION CPPH3
0.
CL
I
TANKERS CRSEK
E e~
ui
u
a e.cs
u
X
a
"E i a! 3 5 S 5 7 S S iii Ti il K! t^ t5 is t'/ is
HOUR OF DAY JUKE 1S77 - AUSU3T 1S77
HSJKLY AVERAGE N02 NO AND NOX CX3NCEHTRATION CPPM3
SOURCE; American Electric Power Service Corporation
3-9
-------
provides an exceptionally good opportunity to profile ambient air pollu-
tion arising from area source emissions. These stations reflect a
variety of area source settings -- office complexes, intersections with
high traffic volumes, suburban commercial centers yet they display a
remarkable uniformity in recorded concentrations. This is due to a
relatively even spread of area source emissions and/or to a spatial
smoothing effect due to the relatively slow rate of NO- formation from
NO. The latter interpretation is investigated in Section 3.1.3.
Figures 3-5 through 3-7 provide evidence for the similarity of ambient
NCL levels at these monitors, and suggest possible explanations for the
patterns observed. Figures 3-5 and 3-6 show diurnal variations during
typical summer days at two of the stations. Values shown are concentra-
tions for each hour averaged over an entire month. These are days when
ozone levels are likely to be high. NO and NO- levels both fall after
midnight due primarily to low NO emissions and to atmospheric dilution.
A
If 0» (or other oxidant) levels are initially also low, NO will build up
rapidly in the morning in concert with traffic levels, while the rise in
NO- lags behind the NO buildup. The resultant NO- peak in early to
mid-morning then subsides as photo-chemical dissociation becomes a
balancing factor. In both Figures 3-5 and 3-6, afternoon NO levels are
shown to be fairly low (due to high 0»), while N09 has reached a steady
J ^
state level. With the reduced UV insolation in late afternoon and
evening and the increased NO emissions from rush-hour traffic, NO-
X ^
begins a rapid rise followed by or simultaneously with an NO increase as
0- is depleted.
Wintertime diurnal variations are similar, with NO levels showing the
same daily variation corresponding to traffic fluctuations. Figure 3-7
3-10
-------
FIGURE 3-5
AVERAGE HOURLY NO AND NO CONCENTRATIONS FOR JULY 1977
(Engleside Station, Washington, D.C. Area)
co
i
NO, NO,
(ppm)
0.035-
0.030-
0.025-
0.020-
0.015-
o.oio-
0.005-
NO
NO.
r
4
6
8
10
~T
14
12
Hours
.URGE: Data from local pollution control agencies
16
18
20
22
24
-------
FIGURE 3-6
AVERAGE HOURLY NO AND NO2 CONCENTRATIONS FOR JANUARY 1977
(Engleside Station, Washington, D.C. Area)
CO
i->
ro
NO, NO,
(ppm)
1.0000 -
0.0875 -
0.0750 -
0.0625 -
0.0500 -
0.0375 -
0.0250 -
0.0125 -
-iiir
2 468
-r
-T-
i ' 1 ' T ' 1
10 12 14 16 18 20 22 24
Hours
SOURCE: Data from local pollution control agencies
-------
FIGURE 3-7
AVERAGE HOURLY NO AND NO CONCENTRATIONS FOR JUNE 1977
(Lewinsville Station, Washington, D.C. Area)
NO, NO
(ppm)
.05--
Q.04-
CO
0.03"
0.02-
O.Ol"
NO
NO
\
\
\
-t-
2
6
10
12
14
16
ia
22
24
SOURCE: Data from local pollution agencies
-------
shows that the peaks in NO- are somewhat muted, though still obvious.
With reduced UV radiation, lower temperature, and low CL levels in the
winter, NCL concentrations are limited by slower conversion rates and
atmospheric dilution.
The EPA (OAQPS, MDAD) has recently reviewed patterns of hourly N02
levels in several urban areas (U.S. EPA, 1978) and has labeled the
morning peak "photo-chemical synthesis" and the afternoon peak "ozone
titration," although the processes that lead to each are both obviously
related to photochemistry. In addition, they stress the importance of
"carry-over" NO high levels of NO. formed one day which are not
depleted when new NO- production begins the next. The highest hourly
N02 levels observed are likely to be a result of NCL "build-up" over
several days.
Recent studies on the relative impact of mobile and stationary sources
on high NO- concentrations show that measured high NCL values are gener-
ally believed to be caused by area, especially mobile, sources. One
study, which analyzed NO-, CO, and SO- data from the Welfare Island
monitoring site in New York City (Iverach, 1978), assumed that ambient
SQj can be used as a surrogate for stationary source emissions and that
ambient CO can be used as a surrogate for mobile source emissions.
Therefore, the concentrations, of CO and SO- during periods of high NO-
should show which source type predominates. The conclusion of the
analysis is that for the area around the Welfare Island site, NO emitted
from mobile and other sources is more likely to result in high measured
NO- than NO emitted from point sources.
^ X
Other studies also tend to suggest that mobile and area sources of NO are
A
the primary contributors to the levels occurring at ambient monitoring
sites. First, using Los Angeles Basin data (Chang, Norbeck, and Weinstock,
3-14
-------
1980), it has been shown that high one-hour NO- levels resulted mainly
from vehicular sources. An examination of the NO /CO and SO-/CO ratios
X £
shows that a significant impact of elevated NO sources on high NO-
X »
rarely occurs. In fact, a stationary (both ground level and elevated)
source contribution of greater than 10 percent was rare at the LA Basin
sites examined. The only exception to this would be a case where stationary
sources with low stacks were near a monitor.
Chang, Norbeck, and Weinstock also concluded that rollback calculations
which assume that high hourly NO- concentrations are proportional to the
total tonnage of NO emissions from all sources will greatly underestimate
X
the improvement in NO- air quality to be expected from a reduction in
vehicle NO emissions in the LA Basin. Therefore, source contribution
factors must be used to discount the effect of emissions from stationary
sources on observed N0? levels.
A recent study by SRI International (Martinez and Nitz, 1979) examined
high (greater than 0.20 ppra) hourly N0~ levels at 48 monitoring sites in
California during 1975-1977. After reviewing approximately 1800 site-
days when NO- exceeded 0.20 ppm were reviewed, it was found that mobile
source emitters predominated at all stations. Point source effects
linked to high hourly NO- were infrequent.
In addition, Martinez and Nitz found that high hourly NO. levels in
Southern California coincide with increased emissions from stationary
area sources of NO . The diurnal variations of the high NO- levels at
X £
the California monitors were associated with the traffic cycle. There-
fore, some combination of mobile and stationary area sources contributes
to peak NO- levels.
As noted above, these studies tend to support the contention that high
measured NO- concentrations are attributable to mobile or other area
3-15
-------
source impacts and, thus, point source emissions can be neglected.
Therefore, although ambient concentrations of N02 are created by a mix
of emissions from point and area sources, only area sources will be
modeled using linear rollback. This is a reasonable approach, particu-
larly if, as appears true from the available evidence, most monitors in
urban areas are sited to reflect contributions primarily from area
sources of NO emissions.
A
3.2 SPATIAL VARIATION IN NO,, CONCENTRATIONS
The degree of spatial variation in NO- levels observed in urban areas
carries important implications for the mobile and area source analysis.
As discussed in Section 4, the mobile and other area source analysis
assumes that N09 concentrations are relatively constant within a county
and that linear rollback can be used to estimate the effects of NO
A
reductions on ambient NO- levels. This assumption is tested by examining
NO- monitoring data from two urban areas.
3.2.1 Regional Scale Variation in NO^ in Los Angeles
Few areas have enough NO- monitors to allow spatial variation patterns
to be determined; the Los Angeles Basin (through regular monitoring) and
the St. Louis area (from the RAPS study) are exceptions. Figures 3-8
and 3-9 are maps of air quality isopleths drawn according to standard
cartographic practices from ambient data at 20 continuous NO- monitors
in Los Angeles. Figure 3-8 demonstrates that annual average NO- values
vary in a relatively uniform way from the city center outward, with most
of the AQCR recording violations of the current NAAQS. However, the
second-high hourly values in Figure 3-9 show considerably more variation
with geographic location. Though high levels are recorded in most parts
of the AQCR, a steep gradient is apparent from the central city outward.
Moreover, nodes and corridors of high concentration are evident east and
southeast of the center, suggesting that ambient levels of N02 are not
3-16
-------
FIGURE 3-8
1977 N02 CONCENTRATIONS IN THE LOS ANGELES AIR QUALITY CONTROL REGION
u>
I
AQCR tommy (
SANTA aARDARA COUNTY I
A-AubiiB IM
B-tauu Kl
C-Buibii* 134
D-bttiillg 43
E-CuttDta U*
MONITORING SITE AND READINGS
F-UHatu 1042 K-UtAoieluCly 105
6 Lamm 12Z \_. Lymnrf IM
H Un( ttat* 131 M NeOull U
I -Us An|ele> IX M-Faafcii IU
J Us Anitlct 1411 0 Puaua 134
P -'Redlauls 60
Q-Riibiduui IS1
R-SwBunaiiliiui 55
S-tolaltirtaa 70'
T-Kkilliir I3C
1977 ARITHMETIC MEAN fog/m3)
I Measuied by Salt2»in Colaiwliic fclhod
I
possible obseivalions
0
LWM^MM^MBMM!
WLFS
-------
FIGURE 3-9
1977 NQ2 CONCENTRATIONS1 IN THE LOS ANGELES AIR OUALITY CONTROL REGION
CO
I
oo
A-Anaheiai 501
B-fcuisi 5ZS
C-Buibaak >U
0 Canaiillo MS
E-CoslaHtsa 432
MONITORING SITE AND READINGS
F La Haora MS K Los Angles Of i20
G Lama '52 L Lymnood SI3
H Loni Beat* CM M Neiihall 320
I Los An|el« Mi N Pasatoa 714
J Los Anples I01S 0 Pomona m
f Redlands 432
Q.fiubidoui 37S
R Sao Bnoaidioo 32Q
S Sanla Baiban 2(3
T-Whilliti HI
1977 HOURLY SECOND HIGH (jig/m3)
Keismed b) Salliman Coluineliic Uelhod
MILES
-------
necessarily uniform throughout an AQCR, but that on a county level they
remain relatively constant.
During 1976, 15 individual stations monitored hourly NO- concentrations
in St. Louis as part of the RAPS study. The site locations are shown in
Figure 3-10, with Site 101 in the city center. If the other sites are
grouped with respect to their distance from Site 101, conclusions can be
made about the spatial variation in NCL concentrations in the vicinity
of St. Louis:
Site Distances from Number of Annual Average
the City Center (km) Sites N02 (ppm)
0 1 0.028
0-4 3 0.027
4-10 5 0.027
10-20 7 0.015
20-30 2 0.012
These annual averages are the 1976 annual means at all of the sites in
each group. While the annual averages within 10 km of the city center
are relatively constant at 0.027-0.028 ppm, there is a considerable drop
in the annual average at sites more than 10 km from Site 101. Therefore,
it seems reasonable to assume that a county rather than an AQCR-wide
emission inventory is appropriate for characterizing the sources contri-
buting to the design value concentrations in St. Louis. This is consis-
tent with the findings in Los Angeles. Table 3-2 shows the maximum
hourly NCL concentrations observed at 18 RAMS stations in St. Louis
during 1976 and the distances of those sites from the city center. No
consistent pattern in spatial variation can be observed from these
values. Hourly NCL concentrations greater than 0.20 ppm occur even at
stations 10-20 km from the city center, suggesting that some of these
sites are influenced by local sources.
3-19
-------
FIGURE 3-10
RAMS MONITORING SITES IN ST. LOUIS
RAPS CENTRAL
RAMS STATION
3-20
-------
TABLE 3-2
MAXIMUM HOURLY N02 CONCENTRATIONS OBSERVED AT
18 RAMS STATIONS IN ST. LOUIS DURING 1976
Site Distances from Hourly N02
Site Number the City Center (km) Concentration (ppm)
101
104
105
107
102
106
110
111
112
108
109
114
115
116
117
118
119
120
0
0-4
0-4
0-4
4-10
4-10
4-10
4-10
4-10
10-20
10-20
10-20
10-20
10-20
10-20
10-20
10-20
10-20
9. Air Quality Criteria
0.2556
0.1559
0.1864
0.1907
0.1990
0.2449
0.2155
0.2230
0.1689
0.3383
0.1537
0.1624
0.1520
0.2430
0.3594
0.0791
0.1917
0.1916
for Oxides of Nitrogen
SOURCE: U.S. EPA. :
(Draft). ORD, Research Triangle Park, NC.
3-21
-------
3-2.2 Microscale Gradients in NO^ Downwind of a Roadway
To probe potential spatial variations in more detail, the results of a
recent investigation of roadside air quality undertaken as part of the
Los Angeles catalyst study (LACS) were reviewed (U.S. EPA, 1977). During
this study, continuous monitors measured hourly NO , NO-, NO, and 0,,,
X ^ J
along with wind speed and direction and traffic counts from July 20-
August 30, 1978, at sites upwind and downwind of the San Diego Freeway.
The San Diego Freeway is a major urban expressway with average daily
traffic approaching 200,000 vehicles, which should lead to high downwind
concentrations of NO-. Figure 3-11 shows the configuration of the
highway and monitors. These monitors were located in an open area that
is unlikely to be impacted by NO sources other than the roadways shown.
X
Data on NO and NO- levels for four afternoon hours during one represen-
tative day are shown in Figure 3-12. A significant decrease in NO- with
distance from the highway is apparent; at about 400 meters the con-
centrations approximate 20 percent of those at 5 meters. This variation
is substantial, especially given the derived nature of NO-. To further
examine the cause for the degree of change, N09-to-NO ratios were
^ X
computed at these sites for two different levels of 0.- for those days
that had the most complete data sets and during which the wind direction
was roughly perpendicular to the freeway. The results indicate that the
NO_-to-NO ratios reach a peak and then decrease after several hundred
4b A
meters, suggesting that chemical dissociation of NO- as well as atmospheric
dilution contributes to this rapid decrease in ambient levels. However,
the decrease in NO-:NO ratios is somewhat unusual given the levels of
^ X
observed 0_.
These findings must be considered preliminary at this time. The recorded
NO- levels are considerably below peak levels in Los Angeles, due perhaps
to the perpendicular wind field characteristic of the test site and the
low background levels at the LACS site. Oblique, low speed winds are
3-22
-------
FIGURE 3-11
LOS ANGELES CATALYST STUDY
MONITORING SITES
SAN DIEGO FREEWAY
8 Lanes \
o
3
CO
I
CO
o
5
SEPULEDA BLVD.
/ 4 Lanes
O
6
PREVAILING WINDS
O
9
O
10
O
11
o
12
DISTANCE FROM NEAR EDGE
OF FREEWAY
Receptor
3
5
6
9
10
11
12
Distance (m)
30
8
30
121
195
286
385
NORTH
Meters
ru
0 10 20 30 40 SO
-------
FIGURE 3-12
CONCENTRATIONS OF N02 DOWNWIND FROM THE SAN DIEGO FREEWAY
HOURS: 1200-1500. August 15,1978
NCh 0.015 ppm
0.100 ppm
.01
DOWNWIND DISTANCE (METERS)
3-24
-------
expected to inhibit dilution and cause high concentrations over larger
areas.
3-2.3 Implications for the Regulatory Analysis
The above findings at both the regional and microscale levels suggest
that treating NCL as a homogeneous problem for entire counties may not
be wholly appropriate. On the other hand, many of the area source
control strategies considered in Section 4 are difficult if not imposs-
ible to implement on less than a countywide basis (e.g., mobile source
I&M programs). Moreover, many area sources are ubiquitous in urban
areas (e.g., autos) making them likely contributors to high concentra-
tions wherever they occur. Finally, major point sources may account for
some of the "nodes" of high NO- observed in Los Angeles. Our indepen-
dent treatment and control of point sources may thus render the assump-
tion of uniform region-wide N09 levels from mobile and area NO sources
; " «
more acceptable.
3.3 DERIVING PEAK HOURLY N02 VALUES
Given the sparcity of data on hourly N0? concentrations in most regions,
some means of estimating likely peak NO- concentrations had to be devel-
oped if observed air quality data were to be used in the analysis. One
possibility is to extrapolte 24-hour measurements to hourly estimates.
Alternatively, annual average values could be used to estimate peak
hourly levels.
3.3.1 Relationships Between Continuous and Bubbler Data
The relationship between bubbler (24-hour) and continuous (one-hour)
data for N0» at co-located sites was investigated to determine how well
bubble and continuous data correlate. N02 air quality data for all
co-located bubblers and continuous monitors were obtained from the
SAROAD annual summaries for 1976-1978. A total of 43 site years of data
3-25
-------
were from monitors with valid readings (using the SAROAD criteria) for
both bubblers and continuous monitors.* Six site locations had more
than one continuous monitor in 1976, so a total of 49 records was used
in the analysis.
Annual arithmetic means (AAM's) for bubblers and continuous monitors
were plotted against each other and linear regressions were run, yield-
2
ing R values of 0.75 and 0.64 for Lyshkow colorimetric and chemilumin-
2
escence measurement methods, respectively. Both of these R values were
found to be significant at the 99 percent confidence level. The equa-
tions and graphs for bubbler versus continuous AAM's are presented in
Figures 3-13 and 3-14.
Second high hourly values for bubblers then were estimated from the
bubbler AAM's and standard geometric deviations (SGD's) using assump-
tions of lognormality as per Larsen's method (Larsen, 1971). These
estimated one-hour values were plotted against the corresponding observed
continuous values at the same sites (see Figures 3-15 and 3-16). Linear
2
regressions were run and the correlations proved to be poor (R =0.30
2
for Lyshkow colorimetric and R =0.04 for the chemiluminescence measure-
ment method).
A second method of estimating one-hour second highs for bubblers was
attempted in an effort to obtain an improved correlation. New bubbler
AAM's were calculated from the equation in Figure 3-14. These calcu-
lated AAM's and the original SGD's then were used to derive one-hour
second highs. Comparing these values with the actual second highs from
* There were 20 sites in 1976, 18 sites in 1977, and 5 sites in 1978
with valid bubbler and continuous data. One reason for this decline
was that California, the State with the most NCL monitors, dropped all
bubbler data in 1978. A few other States appear to have done this as
well.
3-26
-------
FIGURE 3-13
OBSERVED N02 BUBBLER VS. CONTINUOUS* ANNUAL ARITHMETIC
MEANS FROM CO-LOCATED MONITORS
BUBBLER READING
140
120-
100-
80-
60-
40-
20-
CONTINUOUS READING (m/m»)*
* Method 11-Lyshkow Colormetric
3-27
-------
FIGURE 3-14
OBSERVED NOz BUBBLER VS. CONTINUOUS* ANNUAL ARITHMETIC
MEANS FROM CO-LOCATED MONITORS
BUBBLER READING
140
120-
100-
60-
40
20-
y =.64x -I- 17.1
20
* Method 14-Chemi luminescence
I
40
I
60
I
80
I
100
I
120
140
CONTINUOUS READING
3-28
-------
FIGURE 3-15
CALCULATED BUBBLER VS. OBSERVED CONTINUOUS* N02
ONE HOUR SECOND HIGHS
ro
vo
ESTIMATED BUBBLER
VALUE (ug/m3)
1400
1200-
1000-
BOO
600-
400-
200-
l
200
400
I
600
i
800
y = .59x+367.2
r2=.3C
1000 1200
OBSERVED CONTINUOUS DATA (ug/m3)*
*Method 11 - Lyshkow Colorimetric
-------
FIGURE 3-16
CALCULATED BUBBLER VS. OBSERVED CONTINUOUS* NOz
ONE HOUR SECOND HIGHS
ESTIMATED BUBBLER VALUE (Mg/m1)
2900-
2700-
2500-
2300-1
1300-
1100-
900-
700-
500-
300-
100-'
0
y =.57x +453
t* =-.04
I I I I
500 600 700 800
OBSERVED CONTINUOUS* DATA
100 200 300 400
900
Method 14-Colofimetric
3-30
-------
the chemiluminescent instrument produced an equally poor correlation
(R2 = 0.03).
One explanation for the lack of correlation between the estimated and
observed values is that the data might not be lognormally distributed.
To test this assumption, the same method used for calculating one-hour
second highs for bubblers was applied to the AAM's and SGD's of con-
tinuous monitors. The second high one-hour estimates for continuous
monitors calculated in this manner bore little resemblance to recorded
values at the same monitors. Therefore, it can be concluded that the
extreme tails of hourly NCL distributions deviate substantially from
lognormality and that using the Larsen approach to estimate second high
values from bubbler AAM's is not justified. Thus, the comparison of
extrapolated bubbler data with continuous observed data, assuming lognor-
maility in the bubbler data, is not valid. In addition, however, the
discrepancies in annual averages between bubbler and continuous monitors
suggests a basic incompatibility between the two types of instruments.
3.3.2 Relationships Between Annual Averages and Peak Hourly Values at
Continuous Monitors
As noted above, developing relationships between peak and mean values
for recorded data at continuous monitors and applying these relationships
to annual means recorded at the more common 24-hour monitors is a convenient
way to extrapolate the observed short-term values. Previous investi-
gations have shown that ratios of peak (observed maximum hourly value)
to mean (observed annual arithmetic average) for approximately 120
continuous monitors are 6:1, on average (median ratio) (Trijonis, 1978).
A more recent study of continuous monitors in urbanized areas conducted
by EPA concluded that the mean ratio of the second high daily maximum
hour-to-annual average concentration is also approximately 6:1.* How-
* Personal communication with Tom McCurdy, ASB, SASD, Office of Air
Quality Planning and Standards.
3-31
-------
ever, the degree of variability in peak-to-mean values with time at any
one site and the possible consistent variation of the ratio with geo-
graphic location (i.e., source influence) were not investigated. In
order to provide additional insight and to develop a more robust esti-
mate of this ratio, 15 well-characterized continuous N02 monitoring
sites were selected for detailed investigation. Several years of re-
corded hourly values at each of these sites were obtained from EPA's
SAROAD data file and curves approximating the distributions for each
year were fitted to the data. Estimates of the predicted second highest
hourly concentration then were compared to predicted annual means,
ratios were computed, and the temporal and within-site variations in
these ratios were examined for 35 station-years of data.* Details of
this analysis are presented in Appendix A.
3.3.2.1 Sites Investigated
The 15 sites were chosen to represent a variety of source influences.
Although the majority are either mobile source dominated or influenced
by a mix of source types, at least two are dominated by emissions from
point sources. Source influence was determined from the site descrip-
tion in the EPA site directory (EPA, 1978a) and through discussions with
EPA personnel.** The sites, locations, years of record, and type of
source influence are shown in Table 3-3. A second criterion was com-
pleteness of the record, with 75 percent of the possible 8,760 hourly
readings considered the lowest acceptable level of coverage. Further-
more, although these tended to be sites with fairly reliable data records,
an EPA computer program to filter probable anomalies from each year's
record was applied to further improve the data sets (EPA, 1978c).
* The expected second highest hourly level per year is somewhat higher
than the second high daily maximum hour. Consequently, the estimated
ratio of this concentration to the annual mean will be slightly
higher than the corresponding ratio for desired concentration.
** Primarily Hal Richter and Don Sennett, MDAD.
3-32
-------
TABLE 3-3
STATIONS USED IN THE STATISTICAL ANALYSIS
Site Code
030600002G01
056800004101
058720001101
060580002F01
180080008F01
222160005F01
234880002F01
234880002F01
334680050F01
050230001101
050500002101
053900001101
054180001101
054180002101
054200001101
Location
Phoenix, AZ
San Diego, CA
Whittier, CA
Denver, CO
Ashland, KY
Springfield, MA
Saginaw, MI
Southfield, MI
Welfare Island, NY
Anaheim, CA
Azusa, CA
Lennox, CA
Los Angeles, CA
Los Angeles, CA
Los Angeles, CA
Years of Record
1976
1976-1977
1975-1977
1975-1976
1975-1977
1977
1975-1977
1976-1977
1975
1975-1976
1975-1977
1975-1977
1975-1977
1975-1977
1975-1977
Source Influence
Mobile
Mobile
Mobile
Mixed
Point
Mixed
Mixed
Mobile
Point
Mobile
Mixed
Mixed
Mixed
Mobile
Mixed
3-33
-------
3.3.2.2 Best Fit Functional Forms of the Distributions
Once the data were assembled and screened, the frequency distributions
of the upper 50 percentile values were fit to two alternative functional
forms: lognormal and Weibull. Two forms were tested because the tradi-
tionally used two parameter lognormal function, while generally adequate
in reflecting the bulk of air quality data, frequently provides a poor
fit for values at the upper tail of the distribution (Curran and Frank,
1975). The Weibull function, on the other hand, is more adaptable to
"light tailed" distributions and has been shown to be a better predictor
of extreme ozone values (Johnson, 1979).
In brief, the analytical procedure used here consisted of transforming
the two distributions into linear forms that then could be evaluated
with standard regression techniques. Using the 44 station-years of
SAROAD data and all NO- values above the 50 percentile level for each
station-year record, the equations were specified and tested for good-
2
ness of fit. The coefficients of determination (R ) for each of the
site years varied from 0.965 to 0.999 for individual years, with no
consistent difference between the two functional forms. Other sta-
tistics, however, were used to judge the goodness of fit for the upper
tails of each distribution.* On this basis, the Weibull distribution
proved superior in 23 of the 35 site-years.
3.3.2.3 Peak-to-Mean Ratios
Once the best fit function was selected, predicted peak and mean values
for each site were estimated and their ratios computed. The results
across all 15 sites were as follows:
* The Durbin-Watson statistic was used.
3-34
-------
Peak:Mean Ratios
6.1
6.5
1.5
23%
5.5-7.5
3.6-9.4
Statistic
Median (Md)
Mean (M)
Standard Deviation (SD)
Coefficient of Variation (SD/M)
50% Confidence Interval
95% Confidence Interval
The variation across sites indicated above is substantial, due pre-
sumably to variations in both meteorology and type of source influence.
To distinguish these effects, variations in peak-to-mean ratios over-
time at individual sites and variations among types of sites were examined
Though the time variation at individual sites was less than the total
variation in the data set, it was still considerable: an average standard
deviation of over 1.0 (or a coefficient of variation of about 15 percent)
for each site. Thus, meteorology would appear to play a significant
role in determining peak-to-mean ratios. On the other hand, the variation
between types of monitors was not as great as anticipated:
Number of
Sites
Average Peak:
Mean Ratio
Standard
Deviation
Point Source
Dominated
7.0
Mixed Source
Influenced
6.3
1.5
Mobile Source
Dominated
6.9
1.5
However, the limited number of point source dominated sites makes this a
considerably less-than-sufficient test. Still, given this outcome, we
can only assume that no statistically significant difference can be
discerned between point and area source-dominated monitors. Thus, the
3-35
-------
average peak-to-mean values for all sites should be used in the area
source component of the regulatory analysis even though most monitors
reflect area source contributions.
3.3.2.4 Implications for the Regulatory Analysis
Despite the lack of a perceptible relationship between peak-to-mean
ratio and site characteristics, the average ratio is broadly consistent
with previous findings. Thus, a 6:1 ratio is employed in the mobile and
area source analysis to estimate peak (second high daily maximum hour)
NO- values for those counties lacking continuous monitors.
3-36
-------
REFERENCES FOR SECTION 3
Chang, T.Y., Norbeck, I.M., and Weinstock, B. 1980. "N02 Air Quality-
Precursor Relationship: An Ambient Air Quality Evaluation in the Los
Angeles Basin." J. Air Pollution Control Association. Volume 30, No.
1. February.
Curran, T. and Frank, N. 1975. "Assessing the Validity of the Lognormal
Model when Predicting Maximum Air Pollution Concentrtions" Paper No.
75.3. 68th Annual Meeting of the Air Pollution Control Association.
Boston, MA.
Iverach, D. 1978. "A Technique for Estimating the Relative Impact of
Mobile and Stationary Sources on N0~ Concentrations." J. Air Pollution
Control Association. Volume 18, No. 8. August.
Johnson, T. 1979. "A Comparison of the Two-Parameter Weibull and Log-
normal Distribution Fitted to Ambient Ozone Data." In Proceedings of a
Specialty Conference on Quality Assurance in Air Pollution Measurement.
New Orleans, LA.
Larsen, R.I. 1971. A Mathematical Model for Relating Air Quality
Measurements to Air Quality Standards (AP-89). U.S. EPA, Research
Triangle Park, NC. November.
Martinez, I.R. and Nitz, K.C. 1979. Analysis of High N02 Concentra-
tions in California, 1975-1977 (EPA-450/4-79-034a). SRI International,
Menlo Park, CA. August. (Prepared for U.S. EPA, Research Triangle Park,
NC.)
Trijonis, I. 1978. Empirical Relationship Between Atmospheric Nitogen
Dioxide and Its Precursors (EPA-600/3-78-018). Technology Service
Corporation, Santa Monica, CA. August. (Prepared for U.S. EPA, RTP,
NC.)
U.S. Environmental Protection Agency (U.S. EPA). 1978a. Directory of
Air Quality Monitoring Sites Active in 1976. (EPA-450/2-78-00). Moni-
toring and Data Analysis Division, OAQPS, Research Triangle Park, NC.
February.
U.S. EPA. 1978b. "Status and Implications of Analyses of Ambient N0_
and Other Air Quality Data" (Draft). Monitoring and Data Analysis
Division, OAQPS, Research Triangle Park, NC. May.
U.S. EPA. 1978c. Screening Procedures for Ambient Air Quality Data
(EPA-450/2-78-037). OAQPS No. 1.2-092. Research Triangle Park, NC.
July.
3-37
-------
REFERENCES FOR SECTION 3 (Continued)
U.S. EPA. 1979. Air Quality Criteria for Oxides of Nitrogen (Draft)
Environmental Criteria and Assessment Office, ORD, Research Triangle
Park, NC.
3-38
-------
4. NATIONWIDE STUDY METHODOLOGIES
The two previous sections underscored the multifaceted nature of poten-
tial NO problems and the limited extent of NO- monitoring in the United
States. These characteristics pose a challenging analytical problem for
any investigation of NO- air quality. Given the objectives of the regu-
latory analysis -- estimating the impacts of the current annual and
alternative short-term NAAQS for NO- on all sources of NO emissions in
Z. X
the United States the task is enormously difficult.
The approach adopted here simplifies the source-receptor relationships
while maintaining a sensitivity to the two major types of NO- problems
into which the four types described in Section 2 can be consolidated.
In essence, point sources are studied from a within-plume 0_ titration
perspective using atmospheric dispersion models, while area and mobile
source emissions are analyzed using currently available NO- monitoring
data and linear rollback. The reliance on dispersion modeling in the
point source analysis reflects the lack of monitoring data at sites
where point source impacts are expected, while the use of monitoring
data in the area source analysis reflects the typical placement of NO-
monitors near mobile and area sources in urban areas.
Although the mobile and area source analysis does not distinguish among
alternative types of episodes (i.e., suppressed mixing, photochemical
production, or mobile source titration as discussed in Section 2), the
fact that the suppressed mixing and photochemical production episodes
imply a linear relationship between ambient NO- and area-wide NO emis-
" a
sions offers substantial support for the selected approach. The mobile
source titration episode is more complex and probably cannot be analyzed
by a proportional modeling technique. However, the occurrence of these
4-1
-------
episodes is much less frequent than the other two, at least in Los Angeles
(Hayes et al., 1980). Therefore, even without a detailed study of the
causes of high NO- levels in each geographic area, the assumption of
linearity between emissions and air quality can be justified.
Although this dichotomous approach does account for different types of
NO- problems, the analytical separation of point sources and mobile and
area sources is not totally satisfying. Some point sources with short
stacks undoubtedly contribute to high monitored NO- levels, just as
certain area sources may contribute to ambient levels where the maximum
point source impacts are expected. Some attempt is made to account for
this interconnectedness, although limitations remain.
Standards are analyzed for three separate time periods: one-hour,
24-hour, and annual average. Although the same basic area and mobile
source methodology is used for all three, two separate point source
approaches are necessary: one for the annual standard and one for the
short-term standards.
The rest of Section 4 discusses the methodologies used to analyze point
source and mobile and area source emissions, including both atmospheric
dispersion and least-cost modeling. The origin of cost and control
effectiveness estimates for alternative control techniques also is dis-
cussed. Finally, the methodology for examining uncertainty (point
sources) or sensitivity (mobile and area sources) in the nationwide
results is presented.
4-2
-------
4.1 POINT SOURCES
4.1.1 Emiss ions Data
4.1.1.1 Processing of NEDS
The National Emissions Data System (NEDS) was used as the source of
point source emissions and operating information. The first step in
processing that information was to select all of the sources in the NEDS
point source file that are potential NO emitters. This was done by
identifying SCC's with non-zero NO emission factors and pulling the
X
records for those SCC's from the NEDS point source file. The list of
SCC's with non-zero NO emission factors was generated using Appendix C
X
to AP-42 (U.S. EPA, 1979a). This screen of the NEDS point source file
identified approximately 85,000 potential NO emitters (sources).
X
The second step in the analysis was to choose a minimum plantwide NO
X
emission level for excluding plants with low emissions from the remain-
der of the analysis. This was done by examining the distribution of
plantwide NO emissions in the NEDS point source file. Table 4-1 de-
tails this distribution. It should be noted that NO emissions in this
x
table are taken directly from NEDS (contrasted with EEA NO emission
X
calculations using emission factors). The first screen was set at 500
tons of NO per year, which reduced the inventory to 1998 plants with
X
18,949 sources that emit 10,497,618 tons per year of NO . However, this
A.
screen eliminated too many sources in the 1C engine and industrial
process categories. While these sources are small emitters, they have
short stacks and could cause high NO- levels. Therefore, it was decided
to lower the plantwide emissions cutoff to 50 tons per year.
Of the SCC's with NO emission factors, it was decided that all commercial/
institutional and space heat boilers should be excluded from the analysis.
These are very small sources and are individually unlikely to contribute
significant NO- concentrations.
4-3
-------
TABLE 4-1
DISTRIBUTION OF NO EMISSIONS
IN THE NEDS POINT SOURCE FILE
Emission Levels
(tons/year)
1-5
5-10
10-15
15-25
25-50
50-100
100-250
250-500
500-1000
1000-2500
2500-5000
5000-10,000
10,000-25,000
25,000-50,000
50,000-100,000
100,000-250,000
>250,000
TOTALS
Number of
Plants
13,410
4,865
1,886
1,104
1,329
1,658
1,410
1,407
853
622
594
315
212
175
61
14
5
0
29,920
Number of
Sources
21,430
9,053
4,478
3,198
4,134
5,826
5,851
7,204
4,752
4,325
5,654
3,633
2,940
1,604
575
196
22
0
84,875
Total NO Emissions
(tons/year)
1,899
11,266
13,426
13,562
25,831
59,751
101,026
224,349
306,379
443,468
945,454
1,112,380
1,490,624
2,808,465
2,147,138
920,345
629,744
0
11,255,107
4-4
-------
In the industrial process categories, a number of SCC's were dropped
from the analysis to eliminate the chance of double counting N0x emis-
sions from those sources. Each industrial process source in NEDS has
two SCC's. One accounts for emissions of the process itself, while the
other accounts for emissions from in-process fuel use. Because it was
important to have a fuel use estimate for each source for calculating
control costs, the in-process fuel use SCC's (3-90's) were used in most
cases to identify a source instead of the process SCC. Process SCC's
were included only if the SCC units were listed in terms of the fuel
used in that process. The result was that the following SCC's were
dropped:
Dropped SCC's
3-01-003-06
3-01-003-07
3-01-005-04
3-01-005-07
3-03-003-02
3-03-003-08
3-03-005-07
3-04-003-01
3-04-007-01
3-04-007-02
3-05-003-11
3-05-003-12
3-05-003-13
3-05-003-14
3-05-003-15
3-05-003-16
3-05-006-06
3-05-012-01
3-05-012-02
3-05-012-03
Process
Ammonia Production
Carbon Black Production
Coke Manufacturing
Copper Smelter
Gray Iron Foundry
Steel Foundry
Brick Manufacturing
Cement Manufacturing - Dry
4-5
-------
Dropped SCC's Process
3-05-012-05 Fiberglass Manufacturing
3-05-012-06
3-05-012-11
3-05-012-12
3-05-012-15
3-05-014-02
3-05-014-03 Glass Manufacturing
3-05-014-04
3-05-016-04 Lime Manufacturing
3-05-016-05
3-06-002-01 Fluid Crackers*
3-06-003-01 Mov-Bed Cat. Crackers*
3-06-004-01 Blow-Down System*
The next step was to screen the NEDS point source file to determine the
number of entries that were not within a reasonable range and to substitute
default values where we believed the original entries were incorrect.
The default value routine is described below:
1. Stack height (h)
a. If h is missing, unreadable, or zero, use the NEDS
default (Median value) for that SCC.
b. If there is no default or the default is zero,
set h = 100 ft.
c. If h <15 ft, use the NEDS default for that SCC.
d. If h >1500 ft, use the NEDS default for that SCC.
* Although these categories do not have a corresponding fuel use SCC,
they were excluded due to the lack of information on appropriate NO
controls. This is discussed in Section 4.1.3. s
4-6
-------
2. Stack diameter (d)
a. If d is missing, unreadable, or zero, use the NEDS
default (median value) for that SCC.
b. If there is no default or if the default is zero,
set d = 4.5 ft.
c. If d is non-zero and <1, use the NEDS default for that SCC.
d. If d >30 ft, use the NEDS default for that SCC.
3. Stack gas temperature (t)
a. If t is missing, unreadable, or zero, use the NEDS
default (median value) for that SCC.
b. If there is no default or if the default is zero, set
t = 350°F.
c. If t is non-zero and <72°F, set t = NEDS default.
d. If t >2000°F, set t = NEDS default.
4. Annual hours of operation (0-hours)
a. If 0-hours are missing, unreadable, or zero, set
equal to 7,000 hours.
b. If 0-hours <8,736, set 0-hours = 7,000 hours.
c. If 0-hours >1,000, set 0-hours = 1,000 hours.
5. Stack Gas Flow Rate (f)
Compute stack gas velocity using Equation (4-1):
v = -^j (4-1)
TUT
where
v = stack gas velocity-(ft/min)
f = stack gas flow (ft /min)
d = stack diameter (ft)
a. If flow is missing, unreadable or zero, use the NEDS
default value for that SCC.
b. If there is no default or if the default is zero, set
v = 1,330 ft/min (7m/sec).
4-7
-------
c. If v <600 ft/min (3 m/sec), use the NEDS default value for
flow.
d. If v >8,000 ft/min (40 m/sec), use the NEDS default value
for flow.
6. For Boilers:
a. If both Boiler Design Capacity and Operating Rate
are missing, drop record.
b. If operating rate is missing, unreadable, or zero
n n _ Boiler Design Capacity * 0-hours * 0.50
°'Rate ~ Heat Content
where heat content is specified by SCC and multiplying
by 0.50 reflects a 50 percent capacity utilization.
c. If Boiler Design Capacity >upper limit value for that SCC
Operating Rate * Heat Content >1 .
Boiler Design Capacity * 8,760
or if Boiler Design Capacity is missing, unreadable or zero,
. D ., _ . ,, .. Operating Rate * Heat Content * 2
set Boiler Design Capacity = *- 6-37-
where multiplying by 2 reflects a 50 percent capacity utilization
7- For Non-Boilers:
a. If both Operating Rate and Maximum Design Rate are missing,
drop record.
b. If Operating Rate is missing, unreadable, or zero:
O.Rate = Max. Design Rate * 0-hours * 0.50
c. If Maximum Hourly Design Rate >upper limit value for that SCC
or Operating Rate ^ Q
Maximum Design Rage * 8,760
or if Maximum Design Rate is missing, unreadable or zero,
M T> n 4. Operating Rate A _
set Maximum Design Rate = r A .* » 2
0-hours
Using the revised NEDS point source file, both annual and hourly NO
X
emission estimates were made for use in the ambient modeling portion of
the analysis. Annual NO emission estimates were made using emission
4-8
-------
factors and annual operating rates.* Hourly emission estimates were
made using the same emission factors and the hourly maximum design rate
(i.e., source capacity) from the NEDS point source file.
One final modification remained before modeling hourly NO could be
X
undertaken. NEDS lists multi-fueled combustors as separate sources, one
record for each fuel. This does not create a problem for the annual
analysis, because the annual operating rate for each fuel is included in
each record and is used to estimate emissions so that total annual
emissions from the combustor are apportioned among the different fuels.
For the hourly analysis, however, emissions are estimated from hourly
design rates that are the same in each record. In essence, a dual-fired
combustor would be considered as two identical but separate sources. To
avoid this duplication, only the record for the predominant fuel (on a
annual basis) was saved. The total annual consumption of all fuels was
converted into an equivalent quantity of the predominant fuel and the
record adjusted accordingly. Obviously, some error is introduced by
this procedure, but it is much less serious than double or triple counting
multi-fueled combustors.
After all of the previously described changes to the NEDS point source
file were made, the file consisted of 28,619 sources emitting over 9.6
million tons of NO per year. This is 34 percent of the initial NEDS
Ji
NO sources and 86 percent of total emissions. The estimated emissions
x
by source category are shown in Table 4-2 for the retained sources. A
comparison of NEDS-reported and EEA-estimated annual emissions also is
shown.
SCC-specific emission factors were developed from information in
AP-42 (U.S. EPA, 1979) and the NO Control Techniques Document (Acurex,
1978).
4-9
-------
TABLE 4-2
NOX EMISSION ESTIMATES BY SOURCE CATEGORY FOR ALL SOURCES
REMAINING AFTER SCREENING NEDS
NEDS Annual
Technology
Group
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Source
Category
Indus . Boiler
Pulv. Coal
Cyclone
Stokers
Resid (Large)
Resid (Small)
Distillate (Large)
Distillate (Small)
Natural Gas
Utility Boiler
Pulv. Coal
Cyclone
Stokers
Oil & Gas
Turbines
1C Engines
Industrial Process Furnaces
Nitric Acid Plants
Adipic Acid
Explosives - TNT
Ammonium Nitrate
Nitrocellulose
Glass Furnace - Residual
Inprocess Fuel Use
Glass Furnace - Distillate
Inprocess Fuel Use
Blast Furnace/Coke Oven
Inprocess Fuel Use
Incinerators
Miscellaneous
Emissions
(tpy)
371,084
69,995
89,642
343,595
1,060
47,381
549
509,861
3,856,483
1,147,904
55,156
1,960,341
125,898
284,272
185,410
56,821
3,037
3,058
2,665
49
0
0
1,749
2,591
16,236
EEA Annual
Emissions
(tpy)
414,954
53,483
98,935
300,742
1,019
29,172
349
332,891
4,108,576
635,785
55,467
2,783,183
82,270
318,049
358,054
9,703
3,037
1,258
3,152
123
0
12
32,573
2,154
24,961
EEA Hourly
Emissions
(Ibs/hr)
224,166
26,997
59,741
202,294
557
26,307
203
187,302
1,708,389
276,088
33,136
2,753,956
138,646
149,367
189,216
4,151
1,450
2,165
2,788
140
0
2
17,290
1,489
15,416
TOTAL
9,134,837 9,649,902 6,021,306
4-10
-------
4.1.1.2 NEDS Coverage of Point Source NO Emitters
X
While the use of default values to correct obvious errors and missing
entries in the NEDS file should produce a more complete data base for
assessing the effect of alternative N0_ ambient standards, there may
still be errors in the analysis due to the under-reporting or inaccurate
reporting of point source NO emitters in the NEDS file itself. The low
X
emissions of industrial NO in NEDS result from incomplete reporting of
natural gas-fired turbines and reciprocating internal combustion (1C)
engines (U.S. EPA, 1980). These sources are significant NO emitters
X
but, because they do not emit many particulates or sulfur dioxide, they
are often ignored in emission inventory surveys. These sources are
operated largely by natural gas pipeline and gas processing companies.
Table 4-3 shows the NEDS population of gas turbines and 1C engines by
State. As expected, natural gas producing States like Louisiana and
Texas have high populations of these sources. Table 4-4 lists the gas
turbine and 1C engine population in the NEDS file by SCC along with the
NEDS estimated NO emissions. The table shows that industrial natural
x
gas-fired 1C engines emit 65 percent of the NO from this group of
X
sources.
An independent but indirect estimate of the U.S. 1C engine population
can be obtained from the Interstate Natural Gas Association of America
(INGAA). INGAA member companies have approximately 800 compressor
stations (INGAA, 1980) that contain an aggregate of 12 million horse-
power in 1C engines and gas turbines, with most individual engines
requiring from 1000 to 100,000 horsepower. These are natural gas-fired
units. If we assume that the average engine is 5,000-10,000 horsepower,
the total population of 1C engines and turbines in these stations is
1,200-2,400. This can be compared with the 2,402 sources in these
categories in NEDS. On this basis alone, it would appear that 1C engines
and turbines are not grossly under-reported in NEDS.
4-11
-------
TABLE 4-3
RECIPROCATING INTERNAL COMBUSTION ENGINE AND GAS TURBINE
POPULATION IN THE NEDS FILE BY STATE
State
1 Alabama
2 Alaska
3 Arizona
4 Arkansas
5 California
6 Colorado
7 Connecticut
8 Delaware
9 D.C.
10 Florida
11 Georgia
12 Hawaii
13 Idaho
14 Illinois
15 Indiana
16 Iowa
17 Kansas
18 Kentucky
19 Louisiana
20 Maine
21 Maryland
22 Massachusetts
23 Michigan
24 Minnesota
25 Mississippi
26 Missouri
27 Montana
28 Nebraska
29 Nevada
30 New Hampshire
31 New Jersey
32 New Mexico
33 New York
34 North Carolina
35 North Dakota
36 Ohio
37 Oklahoma
38 Oregon
39 Pennsylvania
40 Puerto Rico
Number of
Sources
6
221
62
5
404
33
102
0
1
234
1
32
0
135
25
138
616
15
393
15
34
0
61
83
20
13
10
194
5
5
123
600
43
11
3
49
170
5
81
43
NEDS Estimated
NO Emissions
124
27,388
5,493
3,449
32,266
9,750
924
0
188
23,394
1,231
4,265
0
22,185
16,278
1,519
92,986
2,205
92,991
280
3,687
0
6,475
3,259
2,329
149
4,158
9,975
620
128
9,206
98,071
7,866
47
1,873
3,887
30,046
466
3,043
32,193
4-12
-------
TABLE 4-3 (Continued)
RECIPROCATING INTERNAL COMBUSTION ENGINE AND GAS TURBINE
POPULATION IN THE NEDS FILE BY STATE
Number of NEDS Estimated
State Sources NO Emissions
41 Rhode Island 0 0
42 South Carolina 3 262
43 South Dakota 12 193
44 Tennessee 65 / 7,365
45 Texas 630 466,716
46 Utah 10 50
47 Vermont 23 197
48 Virginia 29 1,137
49 Washington 6 8
50 West Virginia 0 0
51 Wisconsin 60 3,706
52 Wyoming 48 7,462
53 American Samoa 0 0
54 Guam 51 3,441
55 Virgin Islands 1_ 0
TOTAL 4929 1,044,932
SOURCE: National Emissions Data System, September 1980.
4-13
-------
TABLE 4-4
RECIPROCATING INTERNAL COMBUSTION ENGINE AND GAS TURBINE
POPULATION IN THE NEDS FILE BY SOURCE CLASSIFICATION CODE (SCC)
Number of NEDS Estimated NO
SCC
Internal Combustion - Electric Gen
Distillate Oil
Sources
Emissions (tpy)'
2-01-001-01
2-01-001-02
Natural Gas
2-01-002-01
2-01-002-02
Diesel
2-01-003-01
2-01-003-02
Residual Oil
2-01-004-01
Jet Fuel
2-01-005-01
Crude Oil
2-01-006-01
Process Gas
2-01-007-01
Kerosene/Naptha
2-01-009-01
2-01-009-02
Turbine
1C Engine
Turbine
1C Engine
1C Engine
Turbine
Turbine
Turbine
Turbine
Turbine
Turbine
1C Engine
598
147
384
233
525
33
88
83,322
4,436
26,413
153,134
34,895
7,827
4,061
1
7
54
5,204
4-14
-------
TABLE 4-4 (Continued)
RECIPROCATING INTERNAL COMBUSTION ENGINE AND GAS TURBINE
POPULATION IN THE NEDS FILE BY SOURCE CLASSIFICATION CODE (SCC)
SCC
Internal Combustion - Industrial
Distillate Oil
Number of
Sources
NEDS Estimated NO
Emissions (tpy)
2-02-001-01
2-02-001-02
Natural Gas
2-02-002-01
2-02-002-02
Gasoline
2-02-003-01
Diesel Fuel
2-02-004-01
2-02-004-02
Residual Oil
2-02-005-01
Jet Fuel
2-02-006-01
Crude Oil
2-02-007-01
Process Gas
2-02-008-01
2-02-008-02
Kerosene/Naptha
2-02-009-01
2-02-009-02
Turbine
1C Engine
Turbine
1C Engine
1C Engine
1C Engine
Turbine
Turbine
Turbine
Turbine
Turbine
1C Engine
Turbine
1C Engine
41
37
367
2035
38
113
35
7
7
0
0
346
1,563
32,689
677,055
233
8,679
483
786
81
0
0
4-15
-------
TABLE 4-4 (Continued)
RECIPROCATING INTERNAL COMBUSTION ENGINE AND GAS TURBINE
POPULATION IN THE NEDS FILE BY SOURCE CLASSIFICATION CODE (SCC)
Number of NEDS Estimated N0
SCC Sources Emissions (tpy)
Internal Combustion - Commercial/
Institutional
Distillate Oil
2-03-001-01 1C Engine 12 71
Natural Gas
2-03-002-01 1C Engine 0 0
Not Elsewhere Classified 203 3,598
TOTAL 4,929 1,044,933
4-16
-------
4.1.2 Air Quality Modeling Approach
4.1.2.1 One-Hour Averaging Time Analyses
As noted above, the point source analysis involves modeling the process
of 00 titration of NO in the plume of each significant source of NO in
j x
the NEDS file. This requires estimating two variables: levels of NO
X
within the plume at ground level and ambient levels of 0_. Once these
quantities are known, the Ozone Limiting Method (OLM) (Cole and Summerhays,
1979) can be applied with suitable modification. The final modeling
technique developed for this application evolved through considerable
testing and evaluation. The end product is a new version of a simple
Gaussian model called the ESTMS (Efficient Short-term Multiple Source)
model.
4.1.2.1.1 Evolution of the Modeling Approach
The initial effort centered on developing a NO modeling technique
X
suitable for the several thousand NO sources in the NEDS point source
X
file. Due to the large number of sources, the interaction of sources
within and between plants was ignored. Instead, EPA's PTMAX model was
used to estimate the maximum ground level ambient contributions of NO
X
from each source. Adding these maximums for all sources within a plant
gave a high-side estimate of the combined effect of all sources. The
resulting concentration of NO then was used with the OLM to estimate
X
N02 levels:
N00 = 0.10 NO + min. | °'90 N°x (4-2)
X I °3
A weakness of this approach is that the OLM assumes that sufficient time
is available for the plume to completely mix. Only at the point of
complete mixing has all of the available NO been exposed to ambient 0 .
Although the point of complete plume mixing is difficult to estimate,
4-17
-------
the point of final plume rise appears to be a reasonable measure because
entrainment of surrounding air is the key to both plume mixing and the
retardation of plume rise. For certain atmospheric conditions, however,
PTMAX will estimate maximum NO concentrations before final plume rise
x
has been reached. In these cases it is not possible to predict the
resulting N0_ levels.
To illustrate the extent of this problem, ground level NO concentra-
X
tions were estimated for three sample sources under a variety of meteor-
ologic conditions. Figure 4-1 depicts the results of three runs for a
sample plant. As shown, the point of maximum ground level concentration
occurs considerably before final plume height is attained under A stability
and the estimated NO concentration at final plume height is much less
X
than the maximum concentration.*
Consequently, the methodology was adjusted to account for the uncertainty
associated with intermediate plume rise by eliminating all computations
of NO before final plume height had been attained. This necessitated
A
the use of algorithms from both PTMAX and PTMTP -- PTMAX to calculate
the horizontal distance to maximum plume height and PTMTP to calculate
the ground level concentration of NO at the point of maximum plume
height.
A major shortcoming remained, however. Recall that each source within
each plant was modeled independently and the concentrations summed.
This overstates the extent of plume interaction for all sources except
those with co-located and totally identical stacks. The greater the
degree of stack separation within plant boundaries or the greater the
difference in effective stack heights,** the greater the overstatement
becomes.
* A detailed discussion of this analysis appears in Appendix B.
** Effective stack height is the sum of physical stack height and plume
rise.
4-18
-------
FIGURE 4-1
RELATIONSHIP OF GROUNDLEVEL NOx, NOz, AND Os
FOR A HYPOTHETICAL SOURCE USING THE MODIFIED OLM*
CONCENTRATION (PPM)
.SO
.40-
.30-
Stack
Final Plume
Height
DISTANCE
» The dashed lines are computed using equation (4-4)
4-19
-------
To examine this problem quantitatively, a systematic sample of 28 power-
plants was selected from the NEDS point source file and modeled 1) ac-
cording to the PTMAX/PTMTP approach described above and 2) with CRSTER
(U.S. EPA, 1977), a multiple source model that realistically accounts
for plume interaction." Powerplants are a good population of sources to
examine because they typically contain a variety of boilers and turbines
or 1C engines with dramatically different stack heights. They also are
likely to account for the bulk of the control costs for meeting at least
the more stringent short-term standards; thus, the degree to which the
simplified source interaction technique overstates NO concentrations
A
from powerplants is more important than the degree of overstatement for
other types of sources. On the other hand, power-plant stacks tend to
be co-located (or nearly so), thus removing one factor responsible for
overestimating NO .
A
The results of this comparison revealed considerable disparity between
the simplified and the more realistic approach: the ratio of PTMAX/PMTP-
to-CRSTER-estimated NO concentrations averaged 1.8 and ranged from 0.6
to 3.7. However, the wide variation in this ratio among plants reflects
not only the influence of varability in effective stack height,
but the effect of assumed meteorology (PTMAX/PTMTP) versus actual meteorology
(CRSTER) as well. That is, wind speed, wind direction, and atmospheric
stability were assumed in the PTMAX/PTMTP runs (the higher concentration
under stability A or C, together with a 2.0 m/s wind of constant direction
was used), while actual hourly data for each of these variables for an
entire year from an example meteorological station (the 1964 record from
St. Louis) were employed in the runs.** This explains the occasional
PTMAX/PTMTP-to-CRSTER ratio of less than 1.0 and contributes in undetermined
* A detailed description of this analysis appears in Appendix C.
** In addition, the treatment of plume rise differs between the modeling
approaches. The PTMAX/PTMTP method uses intermediate plume rise to
screen non-allowable calculations (only NO estimations at final plume
height are allowed), while CRSTER assumes instantaneous achievement of
final plume height, thereby making a plume rise screening moot.
4-20
-------
ways to those greater than 1.0 as well. However, the maximum concentra-
tion from all plumes (in the first approach) appears to be the more
important factor and is undoubtedly responsible for the average ratio
being considerably greater than 1.0.
This finding forced a re-evaluation of the emerging methodology and led
directly to the development of the ESTMS model. Rather than attempting
to adjust the PTMAX/PTMTP results to account for the tendency to over-
estimate concentration, perhaps through universal application of the
average PTMAX/PTMTP-to-CRSTER ratio, a more direct approach was fashioned.
A simple and efficient dispersion model was developed from basic Gaussian
principles to realistically account for plume interaction from co-located
stacks within the same plant.
This still overstates somewhat the degree of source interaction, especially
for plants with widely separated sources. But modeling non-co-located
stacks in over 5000 plants is not feasible, even if reliable data on
exact stack locations were available in NEDS (which is not the case).
Nor can interaction among plants (and between plants and area or mobile
sources) be accurately accounted for in a nationwide analysis. However,
this effect is addressed in an approximate fashion by adding to the
modeled NCL concentrations a value equal to the highest annual average
NCL level among all valid monitors in the same county. This value can
be adjusted upwards or downwards to reflect greater or lesser plant
interaction given, for example, a urban or rural setting. However, in
the absence of a site-specific investigation, the "best" value for any
one plant is speculative.
4.1.2.1.2 ESTMS and Modified OLM Modeling Procedures
The ESTMS Model estimates short-term (one- to three-hour) concentra-
tions of NO at various receptors downwind from multiple co-located
A:
4-21
-------
stacks. It is based on the classical Gaussian dispersion equation
evaluated at ground level (Turner, 1970):
X. .
1J
i
j
Q
2na a
y zu
= 1, ...,
= 1, ..-,
exp
20
6
(4-3)
where
X =
Q =
0 =
z
u =
H =
concentration (g/m )
emission rate (g/s)
horizontal and vertical dispersion coefficients (m)
windspeed
effective stack height (physical height plus plume
rise)
index of downwind distances
index of atmospheric stability classes.
The dispersion coefficients are functions of downwind distance and
atmospheric stability and are estimated according to the technique in
PTMTP- The downwind receptor network is shown in Table 4-5. Atmo-
spheric stability and windspeed are user inputs, with allowable sta-
bility classes shown in Table 4-6. The assumed persistence of the
stability/windspeed combination chosen is matched implicitly with the
averaging time of the estimated pollutant concentration.
For this NO application, the following combinations were selected:
Stability
A
C
D,
Windspeed
2.0 m/s
2.0
1.0
4-22
-------
TABLE 4-5
DOWNWIND DISTANCE INDEX AND ASSOCIATED DISTANCE FOR ESTMS
Distance
x (km)
1 0.1
2 0.2
3 0.3
4 0.4
5 0.5
6 0.6
7 0.7
8 0.8
9 0.9
10 1.0
11 2.0
12 3.0
13 4.0
14 5.0
15 7.5
16 10.0
17 12.5
18 15.0
19 17.5
20 20.0
4-23
-------
TABLE 4-6
STABILITY CLASS INDEX AND ASSOCIATED STABILITY CLASS FOR CIMS
k Stability Class Meterological Conditions
1 A Strongly Unstable
2 B Moderately Unstable
3 C Weakly Unstable
4 Dj Neutral-Daytime
5 D~ Neutral-Nighttime
6 E Stable
4-24
-------
These represent conditions that occur only a few times per year* and
should produce high concentrations for tall stacks (A and 2.0 m/s),
short stacks (D and 1.0 m/s), and combinations (C and 2.0 m/s). The
CRSTER powerplant modeling exercise reported earlier revealed that these
meteorological conditions were equal to or more stringent than those
from the hourly St. Louis meteorological data set for 27 of 28 powerplants
The one exception was a combination of Class E stability and 3.1 m/s
wind.
Atmospheric mixing depth was not included as an input although it may be
a contributing factor to high ground level concentrations. However, the
net influence of mixing depth is a strong function of effective stack
height, making it difficult £ priori to pick values for mixing depth
that would lead to higher concentrations for all sources or even classes
of sources organized by effective stack height. In addition, the CRSTER
modeling analysis indicated that peak concentrations (i.e., highest
hourly level on the second high day) at 25 of 28 powerplants occured
when mixing heights were not constraining (i.e., greater than 1000 m).
Finally, evaluation of the simple Gaussian dispersion equation reveals
that, for any combination of stack height and mixing depth, the maximum
ground level concentration is not caused by mixing depth limitations.
Instead, the effects of limited mixing depths appear further downwind
after the maximum has been achieved (Turner, 1970). However, it is
possible that mixing depth may become important before final plume rise
is achieved.
* To be consistent with the statistical form of the alternative short-
term standards, these conditions should produce the expected highest
concentration for the second high day (i.e., a day other than that on
which the highest value for the entire year was recorded) the
second high daily maximum hour.
4-25
-------
Rules for translating NO into NO, were also developed. Recall that the
A ^
results of a preliminary analysis stressed the desirability of applying
the OLM only at the point where the plume had stopped rising. This is a
reasonable approach for a single stack. Multiple, non-identical stacks,
on the other hand, present a more complex problem because final plume
rise for all stacks will not be achieved simultaneously. Applying the
OLM to evaluate NO- only after all plumes have stopped rising runs the
risk of underestimating the contributions from those plumes that achieve
maximum height rapidly, while evaluating NO- at distances closer to the
co-located stacks will violate the final plume rise rule for some stacks.
Because the NO- contributions from plumes still rising are surely not
zero (i.e., NO- concentrations lie somewhere between 10 percent of total
NO and the concentration estimated by the OLM at final plume rise), a
X
technique for approximating the concentration of NO- before attainment
of final plume height was developed.
This technique is based on the assumption that 10 percent of total NO
A.
in a plume is initially NO- and that the ratio at which the remaining NO
(90 percent of NO ) is made available by plume mixing for reaction with
A
0- is a linear function of distance traveled. At the point of final
plume rise, all of the NO will have been made available for titration by
0-. The following equation is a simple expression of these relation-
ships for a single stack:
NO =0.10 (N0)+min. f 0'90 <*/*» *°x «'V
* X ( U3
where
x = distance from the stack to the maximum NO
x
concentration
FP = distance from the stack to final plume height
This relationship is depicted graphically in Figure 4-1.
4-26
-------
The program as currently written is slightly more complex and allows the
user to specify one of three scenarios: 1) 10 percent initial NC^ plus
0- titration as a function of distance, as depicted above, 2) the OLM
but without modification for intermediate plume rise, and 3) NO- as a
fixed percentage of NO . The first option was employed here.
Once the concentrations of NO- are estimated using the ESTMS model and
the modified OLM according to the above rules, the receptor with the
highest estimated NO- level is used to determine control strategies and
costs. As noted previously, the contributions from other N0_ sources at
this receptor are simulated by adding the highest annual average NO-
concentration recorded by any valid monitor in the same county. Again,
this value can be increased or decreased at the user's option.
Table 4-7 summarizes the input parameters used in the hourly NO- modeling
analysis. The results then were used together with the least cost model
(see Section 4.1.4) to estimate control strategies and costs.
4.1.2.2 24-Hour Averaging Time Analysis
The estimation of 24-hour NO- concentrations from all significant NO
" A
sources in the NEDS point source file is a simple elaboration of the
hourly analysis. In essence, ratios of 24-hour and one-hour concen-
trations are calculated for a representative sample of plants. The
average ratio then is used to translate estimated one-hour NO- into
24-hour values. Ratios of 24-hour and one-hour NO values also must be
X
computed for use with the least cost model.
4.1.2.2.1 Comparison of Estimated 24-Hour and
One-Hour Concentrations
In order to compare both NO and NO- for the two averaging times, a
stratified sample of 32 plants was drawn from the NEDS point source file
using a systematic sampling procedure. The stratification is by major
4-27
-------
TABLE 4-7
SUMMARY OF INPUT PARAMETERS USED IN THE
ESTMS-OLM ANALYSES OF HOURLY N00
Meteorological Conditions
Stability A with 2.0 m/s wind
Stability C with 2.0 m/s wind
Stability DI with 1.0 m/s wind
Modification to the OLM
N0? computed as 10 percent of NO plus ambient 0- or a fraction
of the remaining 90 percent of N$ that increases as a linear
function of distance from the source up to final plume height,
whichever is lower.
0« values are the highest on the second high day recorded
at any valid monitor in the same county as the plant being
modeled.
Contribution from Other Plants and Area Sources
The highest annual average N0~ level recorded at any valid
monitor in the same county is added to the maximum plant-
wide N0? level.
4-28
-------
source category, with representation within each category approximately
proportional to its population in NEDS. Table 4-8 documents the major
characteristics of the sample.
Each of the plants in the sample was modeled using CRSTER and one year's
(1964) record of hourly meteorological data (windspeed, wind direction,
and atmospheric stability) from St. Louis. The estimated hourly NO
X
concentrations at the maximum receptor then were used to derive the
second high 24-hour average value for the summer (July-October) and
winter (December-February) seasons. Ten-day average diurnal hourly 0»
traces then were assembled for the summer and winter seasons for each
plant,* these traces were comprised of average hour-by-hour 0- values
for the 10 highest days recorded at a selected 0_ monitor located in the
J
same county as the plant being analyzed. These values were used to
estimate hourly N09 values for the second high NO day as per the OLM.**
£» A
The 24-hour average NO and N09 values and the maximum one-hour NO and
X £+ X
NO- values appear in Table 4-9. The ratios of one-hour to 24-hour values
range from a low of 2.2 to a high of 12.6, with a median of about 5.0
for both NO, and total NO . Based on the investigation of N0_ records
^ X 2*
for continuous monitors, these ratios appear somewhat high; however, NO.
monitors are sited almost exclusively so that they detect area and
mobile, not point, source influences. Even for the few monitors located
near point sources, it is practically impossible to site them precisely
at the point where the highest one-hour concentrations are created.
Thus, ratios of observed short-term to longer-term concentrations should
be lower than those based on modeled values.
* The compilation of 0_ data was performed by PEDCo, Inc.
** A modification to the OLM for final plume height as discussed in
Section 4.2.2.1 could not be implemented because CRSTER assumes instan-
taneous attainment of final plume rise.
4-29
-------
TABLE 4-8
PLANTS MODELED FOR 24-HOUR N02 CONCENTRATIONS
State County
Plant N0x
SIC* Emissions (g/sec)
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27-
28.
29.
30.
31.
32.
AR
CA
CA
CA
FL
IL
IN
KA
LA
MD
MA
MN
MT
NM
NY
NY
OH
OH
OH
OK
PA
TX
TX
TX
TX
TX
TX
TX
TX
VA
WI
WY
Hot Spring
Kern
Los Angeles
Los Angeles
Putnam
Madison
Warrick
Sedgwick
Orleans
Baltimore City
SE Mass.
Chippewa
Toole
Lea
Chautauqua
Onondaga
Athens
Hamilton
Muskingham
Tulsa
Berks
Brazoria
Brazos
Harris
Hidalgo
McLennan
Morris
Nueces
Pecos
Pittsylvania
Douglas
Converse
4911
1311
2911
2911
4911
4911
4911
4911
4911
2911
4911
4911
2911
2911
4911
3312
4911
2911
3312
4911
3312
1311
4911
2911
1311
4911
3312
2911
1311
4911
2911
4911
851.12
2.71
5t.64
14.48
171.91
522.34
433.45
695.37
46.44
21.14
26.44
25.53
3.61
128.18
84.38
7.41
305.48
12.17
7.55
444.14
2.51
29.16
182.21
45.91
20.72
1545.38
20.26
58.95
23.07
108.45
17.43
2238.66
53.4
18.3
30.5
8.2
45.7
76.2
75.9
43.9
25.9
21.3
38.1
84.7
9.1
6.7
59.4
15.8
152.4
15.2
50.3
53.3
6.1
11.0
15.2
12.2
4.6
51.8
12.2
4.6
4.6
19.2
11.0
76.2
59.7
61.0
25.9
106.7
152.1
45.7
61.0
85.3
30.5
22.3
--
30.5
56.1
12.2
26.5
20.1
34.4
--
60.4
18.3
36.0
11.0
--
13.7
152.4
Range of
Stack Heights (m) No. of
Min. Max. Stacks Modeled
4
1
5
7
2
3
3
2
3
2
1
1
8
7
2
1
1
5
1
3
3
2
3
4
1
2
2
6
2
2
3
3
*SIC Codes:
1311 - Gas Transmission, 2911 - Petroleum Refineries, 3312 -
Iron and Steel, 4911 - Utilities.
4-30
-------
TABLE 4-9
COMPARISON OF HIGH ONE- AND 24-HOUR ESTIMATED CONCENTRATIONS
CO
NO Results
X
I
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
ESTMS
1-Hour
NO
X
0.3345
0.0267
0.1133
0.1825
0.2474
0.1427
0.1658
0.3503
0.0352
0.0498
0.0324
0.0171
0.0460
4.0921
0.0943
0.0229
0.0778
0.0724
0.0075
0.0996
0.0373
0.0788
0.6952
0.1010
0.8577
0.5061
0.0268
0.5841
0.7952
0.3749
0.0863
0.7880
CRSTER
24-Hour
NO
X
0.0411
0.0077
0.0187
0.0309
0 . 0383
0.0156
0.0161
0.0504
0.0059
0.0155
0.0059
0.0020
0.0121
0.6287
0.0138
0.0107
0.0064
0.0235
0.0013
0.0141
0.0105
0.0298
0.2677
0.0238
0.1417
0.0779
0.0097
0.0936
0.1439
0.1482
0.0371
0.1120
Ratio
8.14
3.47
6.06
5.91
6.46
9.15
10.30
6.95
5.97
3.21
5.49
8.55
3.80
6.51
6.83
2.14
12.16
3.08
5.77
7.06
3.55
2.64
2.60
4.24
6.05
6.50
2.76
6.24
5.53
2.53
2.33
7.04
ESTMS
1-Hour
NO,
£.
0.1535
0.0267
0.1133
0.1825
0.1447
0.1427
0.1366
0.1550
0.0352
0.0498
0.0324
0.0171
0.0460
0.5292
0.0943
0.0229
0.0778
0.0724
0.0075
0.0996
0.0373
0.0788
0.1895
0.1010
0.2058
0.1706
0.0268
0.1984
0.1995
0.1575
0.0863
0.1988
N02 Results
CRSTER
24-Hour
NO,
Z.
0.0367
0.0077
0.0187
0.0309
0.0160
0.0128
0.0115
0.0409
0.0057
0.0139
0.0059
0.0020
0.0121
0.0992
0.0124
0.0103
0.0062
0.0157
0.0013
0.0141
0.0100
0.0248
0.0570
0.0192
0.0447
0.0196
0.0097
0.0352
0.0427
0.0517
0.0299
0.0271
Ratio
4.18
3.47
6.06
5.91
9.04
11.15
11.88
3.79
6.18
3-58
5.49
8.55
3.80
5.33
7.60
2.22
12.55
4.61
5.77
7.06
3.73
3.18
3.32
5.26
4.60
8.70
2.76
5.64
4.67
3.05
2.89
7.34
Season of
Maximum 24-Hour
NO,
z.
Summer
Winter
Summer
Summer
Summer
Winter
Winter
Summer
Winter
Winter
Winter
Winter
Winter
Winter
Winter
Winter
Winter
Winter
Winter
Winter
Winter
Winter
Winter
Summer
Winter
Winter
Winter
Summer
Winter
Winter
Winter
Winter
-------
The size of the ratios reported in Table 4-9 is also influenced by
differences between the ESTMS and CRSTER models. Perhaps the greatest
difference is the use of assumed meteorological conditions in ESTMS and
actual meteorological data in CRSTER. Based on a qualitative comparison
of the assumed conditions in ESTMS and those associated with high concen-
trations in the CRSTER analysis, it appears that the assumed conditions
are at least equal to and, in some cases, somewhat more stringent than
those observed, thus producing high one-hour-to-24-hour ratios. It also
implies that the estimated one-hour values may be somewhat high, at
least compared to those produced by "St. Louis type" weather.
4.1.2.2.2 ESTMS and OLM Modeling Procedures
The modeling procedure for the 24-hour analysis is identical to that
used for the one-hour analysis, with one exception: after the one-hour
NO- concentration is estimated at each plant with ESTMS and modified
OLM, the inverse of the median one-hour-to-24-hour ratio (0.2) is used
to convert this concentration to a 24-hour value. From this point
forward, the analyses are identical.
4.1.2.3 Annual Averaging Time Analysis
A hybrid modeling approach also was developed for the annual analysis.
This model, known as the Climatological Interpolative Multi-Source
(CIMS) Model, is a modification of EPA's Climatogical Dispersion Model
(CDM) and has computing procedures that are much more efficient. This
increase in efficiency is obtained by first calculating normalized
concentrations of NO (i.e., concentration times wind speed divided by
X
emission rate) at each of 10 downwind distances, 6 atmospheric stability
classes and 30 effective stack height classes. The desired concentration
then is obtained by specifying all input variables and interpolating
between effective stack height classes. The concentration of N02 is
calculated either as a fixed percentage of NO or by using the OLM with
X
annual average 0~ values.
4-32
-------
4.1.2.3.1 CIMS Modeling Procedure
The CIMS model estimates NO concentrations for individual sources in a
A
three-step process. First, values for normalized concentration in a
three-dimensional array specified by effective stack height classes,
atmospheric stability classes, and pre-selected downwind distances are
calculated once for all sources. To obtain concentration estimates for
a specific source at each receptor, these normalized values are 1) inter-
polated for the appropriate stack height, 2) multiplied by the emission
rate, and 3) divided by the mean windspeed for each speed class and
weighted by the joint frequency of occurrence of each stability class,
wind speed class, and compass direction for the year of record. Finally,
the estimated concentrations from each source within a plant at each
receptor in the network are summed and the highest aggregate concentration
is used as the estimated annual average NO contributed by that plant.
The basic Gaussian dispersion equation employed by CIMS is taken from
Turner (1970). It estimates ground level concentrations averaged over a
22.5 degree geographic sector (corresponding to 16 wind direction classes)
as a function of distance, wind speed, stability, and stack character-
istics of a source:
Xu _ 2.03
Q ~ V
exp
(4-5)
where
_ 2
X = sector-averaged concentration (g/m )
Q = emission rate (g/s)
a = vertical dispersion coefficient (a function of x
Z and stability) (m)
u = wind speed (m/s)
x = downwind distance (m)
H = effective stack height (m).
4-33
-------
Equation (4-5) is used to compute the three-dimensional array of normalized
concentrations (Xu) noted above. The stability, downwind distance, and
effective stack Height classes used for these computations are shown in
Tables 4-10, 4-11, and 4-12, respectively.
The number of receptors at which concentrations are to be estimated is
equal to the number of downwind receptors times the number of wind
directions (16 for this application). The normalized concentrations are
then used to calculate the annual average concentration for each source
at each receptor from the frequency of occurrence of the wind direction,
wind speed, and stability classes at or near the location of the source.
For a specific source:
X' = F.., M^ - i=l,...,10 (4-6)
ijkl jkl \ Q J.v u. j=i;...; 6
k=l,..., 6
1=1,...,16
where
X'= weighted sector-average concentrations
F = frequency of occurrence of wind speed, wind
direction, and stability
i = index of downwind distance
j = index of wind speed class
i k = index of stability class
1 = index of compass direction
Xu = normalized concentration at the interpolated effective
Q stack height
The wind speed classes correspond to those used in CDM and are shown in
Table 4-13 in terms of class means.
4-34
-------
TABLE 4-10
STABILITY CLASS INDEX AND ASSOCIATED STABILITY CLASS FOR ESTMS
fc Stability Class Meteorological Conditions
1 A Strongly Unstable
2 B Moderately Unstable
3 C Weakly Unstable
4 D, Neutral-Daytime
5 D2 Neutral-Nighttime
6 E Stable
4-35
-------
TABLE 4-11
DOWNWIND DISTANCE INDEX AND ASSOCIATED DISTANCE FOR CIMS
Index Distance
i x (km)
1 0.2
2 0.3
3 0.5
4 0.7
5 1.0
6 2.0
7 5.0
8 10.0
9 15.0
10 20.0
4-36
-------
TABLE 4-12
EFFECTIVE STACK HEIGHT INDEX
AND ASSOCIATED HEIGHTS
Index
m
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Stack Heights
H (meters)
0
10
20
30
50
60
80
100
120
140
160
180
200
220
240
Index
m
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Stack tteignts
H (meters)
260
280
300
325
350
400
450
500
600
700
800
1000
1500
2000
20000
4-37
-------
TABLE 4-13
WIND SPEED CLASS INDEX AND ASSOCIATED MEAN WIND SPEED
Index Wind Speed
j (meters/second)
1 1.50
2 2.46
3 4.47
4 6.93
5 9.61
6 12.52
4-38
-------
The estimated annual concentration from a single source at a receptor is
computed by simply summing all weighted concentrations over all wind
speeds and stabilities:
6 6
il -1-1 ijkl (4-7)
j=l j=k J
where
X., = estimated annual concentration at distance i
and wind direction 1 in (g/m )
The meteorological data utilized by CIMS are taken from the weather
station nearest the plant in question. The STAR data summaries at these
stations then are used directly.
To further illustrate the way CIMS operates, a comparison with CDM is
useful. In addition to the use of table look-up and interpolation
schemes for effective stack heights, there are three basic differences
between CIMS and CDM:
Plume Rise CIMS uses the Briggs plume rise equations with
their different formulations for 1) unstable and neutral
conditions (stabilities A-D,) and 2) stable conditions
(stabilities D? and E). This is the approach generally accepted
in the modeling community. CDM employs a simplified equation
that does not differ across stability categories. This differ-
ence will cause CIMS to estimate slightly higher concentrations
under stable conditions.
Vertical Dispersion Coefficients (y )_ -- First, CIMS follows
the standard practice of using different relationships between
Y and x (y = ax ) for different stabilities. CDM uses the
source relationship for stability classes D,, D_, and E.
Second, CIMS assumes the initial coefficient to be zero for
all stack heights, whereas CDM sets a at x = 0 to 50m minus
the physical stack height, but not more than 30m for all
stacks less than 50m. This is designed to account for urban
roughness, but in effect leads to unrealistically high estimates
at short downwind distances. This tendency to overestimate
concentrations is enhanced by the use of intermediate plume
rise in CDM.
4-39
-------
Mixing Depth CDM allows for a user-specified mixing depth
while CIMS uses a fixed mixing height specified to be infinity.
As was explained in the discussion of ESTMS, specifying a
priori mixing depths that increase ground level concentrations
for all sources is difficult. Moreover, the effect of ignoring
this variable probably will not be great in this application.
4.1.2.3.2 Estimating NO^ from NO
^ X
The same basic procedure for estimating NCL from NO developed for ESTMS
£ X
is employed by CIMS. That is, the effect of plume mixing on the extent
of 0- titration is incorporated as a modification to the OLM. In this
case, the annual average CL level recorded at the 0~ monitor used in the
one-hour analysis is the value used in the modified OLM.
However, the OLM should really be applied on an hour-by-hour basis for
an entire year to compute annual average NO- levels. Consequently, the
use of a user-specified fixed NO~-to-NO ratio also is available as an
£» A
alternate approach. Values of 0.5 and 1.0 were used in this analysis.
4.1.3 Control Options and Costs
Control options and the costs for installing and operating these controls
are estimated for each of the major stationary source categories of NO
A
emitters. Costs are broken down into capital, operating and maintenance,
and fuel costs. Each control option has an associated control effi-
ciency (percentage removal capacity). Controls are specified on an SCC
basis in the control cost file. For each source, there are a number of
control options.
For each process or type of combustor, a range of capacities typical of
the process has been determined. By plotting literature estimates of
costs of control equipment against source size, within these ranges
non-linear functions of the following form have been specified:
Y = aXb (4-8)
4-40
-------
where
Y = cost in dollars (total capital or annual O&M)
per unit source size
X = source size (MMBtu/hr or SCC units/yr)
a,b = empirical constants
As noted above, capital costs are typically a function of boiler design
capacity (or maximum hourly design rate for non-boilers). Yearly operating
and maintenance costs are usually a function of the annual operating
rate. In some cases, O&M costs are a function of both capacity and
operating rate which reflects the fixed and variable components of the
O&M cost function.
The "b" constant reflects the degree to which the relationship is non-
linear due to economies of scale. For capital expenses, b is often less
than one due to the savings in materials from building one large piece
of equipment, as as opposed to several small pieces. For O&M costs,
linearity (b = 1.0) is typical because the cost of operating control
equipment is proportional to the production volume. However, a degree
of nonlinearity may be apparent due to the fixed cost component of the
O&M cost function.
Because "Y" in Equation (4-8) gives unit costs, total costs are calculated
as Y times X.
The least cost model assumes that technology selections are based on
after-tax annualized cost, which includes both annualized capital and
annual O&M. An after-tax capital recovery factor of 0.1828, based on a
10 percent discount rate in real terms with a 10-year time horizon and a
2 percent charge for taxes and insurance, was used to annualize the
capital charge. Details of the financial equation are described in
Section 4.1.4.
4-41
-------
The two main sources of information on control effectiveness and costs
for point sources were the NO Control Techniques Document (Acurex,
1978) and the NO Technology Assessment Reports for Industrial Boiler
Applications (U.S. EPA, 1979b,c). Because the ITAR for industrial
boilers presents data specific to different source sizes, non-linear
cost functions are used for industrial boilers. The CTD does not provide
cost estimates based on size, so only linear cost functions are used for
sources other than industrial boilers. The CTD information is supple-
mented with data prepared by Acurex Corporation in a summary of combus-
tion modification NO controls (Evans and Castaldini, 1978).
A
Point source NO controls are of two general types combustion modifica-
a
tion and flue gas treatment both of which were considered in estimating
the most cost-effective control system for each plant. For each source <
category, a number of candidate control systems were evaluated, including:
Low Excess Air (LEA) In this technique, the combustion air
is reduced to the minimum amount required for complete combus-
tion. With less oxygen available in the flame zone, NO
formation is reduced. In addition, the reduced air flow
reduces the quantity of flue gas released per unit of time re-
sulting in an improvement in boiler efficiency.
Overfire Air Staged combustion through overfire air controls
NO by carrying out initial combustion in a primary, fuel-rich
combustion zone, then completing combustion at lower tempera-
tures in a second, fuel-lean zone. Overfire air is effective
for NO reduction and may be used with all fuels.
A
Flue Gas Recirculation (FGR) In addition to LEA and overfire
air, FGR can be used to provide an extra 20 percent reduction
in NO emissions for gas and oil-fired sources. FGR reduces
NO by lowering combustion temperatures.
X
Reduced Combustion Intensity -- This control system generally
lowers thermal NO formation and can be achieved by load
reduction in existing units and by using an enlarged firebox
in new units.
Low NO Burners (LNB) Low NO burners are generally designed
to reduce flame turbulence, delay fuel-air mixing, and establish
4-42
-------
fuel-rich zones where combustion initially takes place. The
resulting lower flame temperatures reduce thermal N0x genera-
tion, plus the reduced availability of oxygen in the initial
combustion zone, inhibits fuel-NO conversion. Low N0x burners
represent a developing technology that promises highly effec-
tive NO control at relatively low cost.
Ammonia Injection This technique reduces NO to N2 and H20
by injection of ammonia (NH.J at flue gas temperatures ranging
from 1Q70K to 1270K. However, the method is very temperature-
sensitive with maximum NO reductions occurring in a narrow
temperature window aroundX1240K + 50K, so that an elaborate
NH, injection, monitoring and control system is required. The
application of this technique, especially to the severe flue
gas environment from coal combustion, is still several years
away.
Selective Catalytic Reduction (SCR) -- This technique utilizes
NH_ to selectively reduce NO to N~. It is capable of achieving
stringent NO control. SCR processes have been applied in
Japan to several residual oil-fired industrial boilers. SCR
has not been demonstrated commercially on coal-fired boilers;
however, pilot units have been operated and some U.S. firms
are offering SCR processes for use on coal-fired boilers.
Therefore, it has been assumed in this study that SCR will be
developed and available for use by 1985.
The above control techniques were assumed to be applicable to boilers
and industrial furnaces. Specific techniques such as water injection
and fine tuning were assumed for gas turbines, and internal combustion
engines, while chilled absorption was applied to nitric acid plants.
Cost and effectiveness values for each control type are reasonably well-
established for boilers and process sources, but limited investigation
of NO reduction has been carried out for furnaces such as cement kilns,
x
petroleum heaters, and glass melting furnaces. Based on the NO control
techniques document (U.S. EPA, 1978) and communications with Acurex,* no
controls were assumed available for process heaters or catalytic crack-
ing units in the petroleum industry, or for glass melting furnaces.
* Personal communication with Mike Evans of Acurex Corp. (Aerotherm
Division).
4-43
-------
Combustion modification and FGR techniques, identical in cost and effec-
tiveness to those used on boilers, were assumed available for other
industrial furnaces and heaters.
The point source control cost algorithms used in this analysis are shown
in Table 4-14. Algorithms are presented separately for capital costs,
operating and maintenance costs, and fuel use. The codes in Table 4-14
can be understood by referring to Tables 4-15 and 4-16. The first two
digits of the code represent the source category and the second two
digits signify the control device.
In updating all costs to 1980 dollars, the price indices in Table 4-17
were used. The GNP implicit price deflators were used to convert capital
and O&M costs to 1980 dollars. The Chemical Engineering Journal indices
were used to convert to 1980 dollars for stationary source capital
costs. Table 4-18 details the yearly fuel prices assumed in the analysis.
It should be noted that the control costs identified in Table 4-14
increased significantly with increasing control effectiveness. While
this relationship is not exponential in all cases, it is for the majority
of control options. Therefore, as more stringent standards are considered
in the nationwide analysis and as an individual source is forced to
decrease emissions more than 50 percent to meet the standard, total
control costs will increase dramatically.
4.1.4 Least Cost Model
The next step in the analysis procedure is to examine each plant that
has a predicted NCL concentration exceeding the standard under considera-
tion and to estimate which control options are needed to bring that
plant into attainment. The control options selected should take into
account both the cost of each control considered and its effectiveness
in reducing NO . The problem of selecting the least expensive set of
A
4-44
-------
TABLE 4-14
POINT SOURCE CONTROL COST ALGORITHMS
Code
P *
101
102
103
104
105
201
204
205
301
304
305
401
402
403
404
405
406
501
502
503
504
505
506
601
602
608
603
607
610
60?
611
701
703
707
711
Capital Cost*
(1980 $)
a
700.000
1500.000
1500.000
273100.000
7600.000
700.000
273100.000
7600.000
4700.000
1 400 . 000
11100.000
5600 . 000
13600.000
18600.000
12000.000
59800.000
71100.000
5600.000
18600.000
18600.000
12000.000
59800.000
71100.000
1200.000
470.000
3000 . 000
470.000
470.000
470.000
470.000
4100.000
800 . 000
2300.000
2300.000
13300.000
b
0.263
0.247
0.247
0.999
0.000
0 . 263
0.999
0.000
0.637
0.000
0.000
0:714
0.774
0.774
0.431
0.436
0.545
0.714
0.774
0.774
0.431
0.436
0.545
0.416
0.000
1.054
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
O&M Costs**
(1980 $)
a
1 . 300
3 . 000
3 . 800
3 . 600
43.700
1.300
3.600
43.700
3.100
3.800
66 . 600
81.300
2.600
4.200
4.100
730.900
1739.000
81.300
5.200
7.000
4.100
730.900
1739.000
3.100
2.500
0.200
3.800
4.000
4.000
4.200
43.100
0.900
6.400
6.700
179 . 200
b
0.387
0.059
0.074
0.000
0.000
0.387
0.000
0.000
0.542
0.000
0.000
1.635
0.000
0.000
0.020
0.616
0.778
1.635
0.000
0.000
0.020
0.616
0.778
0.620
0.049
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
Fuel Penalty"*" Control
(in1 Btu/year) Efficiency (X)
a
0.510
0.24(3
0.1900
0.000
0.190
0.510
0.000
0.180
4.180
0.000
0.150
0.940
0.480
0.000
0.000
0.220
0.220
0.460
0.520
0.000
0.000
0.220
0.220
0.250
0.480
1.480
0.000 '
0.540
1.480
2.010
0.160
0.490
0.000
0.520
0.200
b
0.000
0 . 000
0.000
0.000
0 . 000
0.000
0.000
0.000
-0.632
0.000
0.000
0 . 000
0.000
0.000
0.000
-0.198
-0.198
0.000
0.000
0.000
0.000
-0.198
-0.198
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
10
20
50
50
90
10
50
90
10
50
90
10
30
50
55
90
90'
10
30
50
55
90
90
10
30
40
50
55
70
75
90
10
50
55
90
4-45
-------
TABLE 4-14 CC°ntinued)
POINT SOURCE CONTROL COST ALGORITHMS
301
302
308
303
307
312
310
309
313
311
901
903
914
.1001
1014
1101
1102
1114
1201
1202
1215
1216
1214
170.000
470.000
3000 . 000
470.000
470.000
470.000
470.000
470.000
1400.000
54300.000
79 . 000
£60.000
7900 . 000
79 . 000
7900.000
79.000
170.000
7900.000
40.000
100.000
1200.000
1100.000
7900.000
1313
1416
1414
1501
1517
1514
1600.000
0.000
4000.000
140.000
400.000
6600.000
0.000
0.000
1.054
0.000
0 . 000
0.000
0.000
0 . 000
0.000
0.516
0.000
0.000
0 . 000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.100
2.000
0.200
3.100
4.000
2.100
3.200
4.100
3.900
951.900
0.030
0.280
t 1 . 200
0.030
11.200
0.030
0.170
1 1 . 200
0.000
0.050
0.140
15.800
11.500
1618 62400.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1.400
3.100
12.200
0.140
0.330
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0 . 000
0.000
0.615
0.000
0 .000
0 .000
0.000
0.000
0 . 000
0.000
0.000
0.000
0.000
0.000
0.000
0.490
0.430
1.980
0 . 000
0.550
2.480
1.980
2.540
1.980
0.030
-0.500
0 . 000
3.000
-0.500
3.000
-0.500
0.500
3.000
-0.500
1 .000
1.000
1.000
3.000
2.000
10.000
1.000
0.000
240.000
-l.f
0.1
2.1
0.000
0.000
0.000
0. 000
0 . 000
0. 000
0.000
0 .000
0 .000
0 . 000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0 .000
0.000
0 . 000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
10
35
40
50
60
60
70
75
75
90
11?
40
90
10
90
10
20
90
15
40
60
70
90
60
30
90
25
50
90
65
4-46
-------
TABLE 4-14 (Continued)
POINT SOURCE CONTROL COST ALGORITHMS
FOOTNOTES
* Capital Costs = aX- X~
For boilers:
X1 = boiler design capacity
X_ = boiler design capacity
For non-boilers:
X- = maximum design rate
X- = maximum design rate
The b exponent is always negative.
** O&M Costs (without fuel penalty) = aX1bX2
For boilers:
X- = boiler design capacity
X_ = operating rate * heat content
For non-boilers :
X1 maximum design rate
X- operating rate * heat content
The b exponent is always negative.
+ Fuel Penalty (106 Btu/year) = aX.^
For boilers :
X- = boiler design capacity
X_ = operating rate * heat content
For non-boilers:
X- = maximum design rate
X~ = operating rate * heat content
4-47
-------
TABLE 4-15
POINT SOURCE CATEGORIES AND CODES
Source Category Codes
Industrial Boiler
Pulverized Coal 1
Cyclone 2
Coal Stokers 3
Residual (Large) 4
Residual (Small) 5
Distillate (Large) 6
Distillate (Small) 7
Natural Gas 8
Utility Boiler
Pulverized Coal 9
Cyclone 10
Coal Stokers 11
Oil & Gas 12
Stationary Gas Turbines 13
Reciprocating 1C Engines 14
Industrial Process Furnaces 15
Nitric Acid Plants 16
4-48
-------
TABLE 4-16
CONTROL DEVICE ABBREVIATIONS AND CODES
Codes
LEA = Low Excess Air 01
SCA = Staged Combustion Air 02
LNB = Low NO Burners 03
x
NIL Inj = Ammonia Injection 04
SCR(PF) = Selective Catalytic Reduction (Parallel Flow) 05
SCR(MB) = Selective Catalytic Reduction (Moving Bed) 06
FGR = Flue Gas Recirculation 07
RAP = Reduced Air Preheat 08
RAP + FGR 09
RAP + LNB 10
SCR(FPB) = Selective Catalytic Reduction (Fixed Pack Bed) 11
RAP + SCA 12
RAP + NH3 Inj. 13
Dry SCR 14
Water Injection 15
Fine Tuning and Changing the Air/Fuel Ratio 16
Advanced Design Burner 17
Chilled Absorption 18
4-49
-------
TABLE 4-17
PRICE INDICES USED TO CONVERT TO
1980 DOLLARS (2ND QUARTER)
GNP Implicit
Price Deflator
Multiplier* to
Convert to
Chemical Engr.
* These numbers were used to convert to 1980 dollars for stationary
source operating and maintenance costs and all mobile source costs.
The source of the GNP implicit price deflators is Mr. Bernstein of
the Bureau of Economic Analysis.
** These numbers were used to convert to 1980 dollars for stationary
source capital costs.
SOURCE: Chemical Engineering Journal, July 14, 1980.
Multiplier' to
Convert to
Year
1980
1979
t 1978
01
0
1977
1976
(1972 = 100)
178.9
165.5
152.1
141.6
133.8
1980 Dollars
1.081
1.176
1.263
1.337
Journal Index
253.5
238.7
218.8
204.1
192.1
1980 Dollars
1 . 062
1.159
1.242
1.320
-------
TABLE 4-18
STATIONARY SOURCE FUEL PRICES ($/106 Btu)
Coal*
Residual Oil*
Distillate Oil*
Natural Gas*
Diesel*
Jet Fuel**
Crude Oil**
Process Cas+
Gasoline**
Coke**
Hood-H-
LPG**
Wood/Bark Uaate+h
Fuel
Code
1
2
3
4
5
6
7
8
9
10
11
12
13
1984
1.90
4.85
5.81
4.00
5.81
6.95
5.92
4.00
10.74
2.91
2.35
7.20
2.35
1985
1.97
5.28
6.21
4.42
6.21
7.82
6.02
4.42
11.78
2.95
2.35
8.05
2.35
1986
2.00
5.43
6.38
4.65
6.38
8.00
6.24
4.65
11.98
2.99
2.35
8.33
2.35
1987
2.03
5.59
6.56
4.90
6.56
8.19
6.41
4.90
12.19
3.04
2.35
8.62
2.35
1988
2.07
5.75
6.74
5.15
6.74
8.38
6.58
5.15
12.39
3.09
2.35
8.92
2.35
1989
2.10
5.91
6.93
5.42
6.93
8.58
6.77
5.42
12.60
3.13
2.35
9.23
2.35
1990
2.13
6.08
7.13
5.70
7.13
8.78
6.95
5.70
12.82
3.18
2.35
9.55
2.35
1991
2.18
6.21
7.26
5.83
7.26
8.94
7.10
5.83
13.01
3.26
2.35
9.70
2.35
1992
2.22
6.34
7.41
5.96
7.41
9.10
7.24
5.96
13.20
3.35
2.35
9.85
2.35
1993
2.27
6.47
7.55
6.10
7.55
9.27
7.40
6.10
13.39
3.43
2.35
10.01
2.35
1994
2.32
6.61
7.69
6.23
7.69
9.44
7.55
6.23
13.59
3.52
2.35
10.17
2.35
1995
2.37
6.75
7.84
6.37
7.84
9.61
7.71
6.37
13.79
3.61
2.35
10.33
2.35
1996
2.42
6.82
7.91
6.44
7.91
9.79
7.87
6.44
13.99
3.70
2.35
10.49
2.35
1997
2.46
6.89
7.98
6.51
7.98
9.96
8.04
6.51
14.20
3.80
2.35
10.66
2.35
1998
2.51
6.96
8.05
6.58
8.05
10.15
8.21
6.58
14.41
3.90
2.35
10.83
2.35
1999
2.56
7.03
8.12
6.65
8.12
10.33
8.38
6.65
14.62
4.00
2.35
11.00
2.35
* Source: EEA IFCAM Model foe Industrial Botlere, September 1980.
**Source: DOE MEFS Projections, ARC 1979, Medium price path. Teat results only - not for release (March 1980). Data provided for 1985, 1990, 1995
and, except for crude oil and coke, 1978. Data Interpolated for Intermediate years. Growth rate for 1990-1995 assumed to apply to 1995-1999.
-t Assume all process gas la currently being used; natural gaa assumed to be substitute.
-H-Rough estimate for current price of wood waste - no projection estimate available; coat Is April 1980 dollars. Source: Chemical Engineering,
April 21, 1980, p. 74. Assume wood coat Is equal wood/bark waste coat.
-------
control strategies that will bring the total emissions in the plant
below the maximum allowable level can be formulated as an integer program
as follows:
K
MIN Z f. (x.) (4-9)
subject to Z g. (x.) >b
c.s { 0, 1, . . ., N.}
where
x. = level of control on source i
f.(x) = cost of control level x on source i
g.(x) = emissions reduced by control level x on source i
N. = maximum level of control available for source i
b = emission reduction needed
k = number of sources
Note: g-(x) is assumed to be an increasing function of x.
A heuristic technique is employed to solve the integer program formu-
lated above. Initially, x. is set to zero for all i. The algorithm
then selects one source at a time to control until either the standard
is met or no more control alternatives exist. The selection process is
based on the marginal cost of the various alternatives. At each iteration,
all higher levels of control for each source are considered. The marginal
cost of moving from the present level of control to a higher level of
control is calculated as follows:
f^n) - f.(xi)
MC. = rain"(g± (n) - gi (x^), Delta (4-10)
Q£ x + 1, ..., N
4-52
-------
where
MC. = marginal cost of increasing the level of control on
source i from x. to n
i
Delta = remaining emission reduction needed
Note: initially Delta is set equal to b.
The strategy with the lowest marginal cost is selected, Delta is recal-
culated, and the process is repeated.
where
Delta = Delta - (S± (n*) - g.^ (x..)) (4-11)
Increasing the level of control on source i from x. to n*
had the smallest marginal cost.
If Delta falls below zero then the air quality standard has been met.
If no more alternative strategies exist the process ends and the plant
is identified as violating the standard.
In order to test whether the solution can be improved, the process is
now reversed, the object being to lower the level of control on some
sources while keeping Delta less than zero (i.e., keeping the plant in
attainment of the N0_ standard). For each source the following checks
are made:
Is f± (n) < f (x^ ?
Is g± (n) - gt (x,.) > Delta?
For ne (o,...,x.-l}
4-53
-------
If so, decreasing the level of control on source i from x. to n is
feasible and desirable. For each alternative satisfying the above
inequalities the marginal cost is calculated as follows:
MC.
in
f. (n) - f. (x.)
i 11
g. (n) - g. (x.)
(4-12)
The alternative with the highest marginal cost is selected, Delta is
recalculated, and the process is repeated. When no alternatives satisfy
both inequalities, the process ends.
The least-cost set of control strategies is selected using the estimated
annualized cost to each plant. However, the final costs presented in
this report are social costs, representing the aggregate before-tax
cost. Plant costs are the aggregate after-tax cost to each firm. The
different procedures for estimating plant vs. social costs are described
in detail below.
The equation used to estimate the annualized cost to the firm for each
control option is as follows:
A i - j n *. I After Tax \ /Capital V /, Corporate\/A{,M fc \
Annualized Cost = ( .. n _ . I x f n * \+( 1- , . UO&M Costs \
\Capital Cost/ I Recovery I I Tax Rate II I
\ / \ Factor / \ /\ /
(, Corporate \ /Present Value\
Tax Rate I I of Future 1
where
/Corporate \ /Present Value
After Tax _ Capitalj1.00 - ( Investment\ - [ Tax I ~ [ of Capital
Capital Cost Cost [
V Tax Credit/
\ Rate
\ Equipment
\ Depreciation,
Investment Tax Credit = 10%
Corporate Tax Rate = 50% (Federal tax rate of 46% and typical
State and local tax rate of 4%)
4-54
-------
Capital Recovery Factor = i + MIS
1 - 1
(1 + I)*
where
i = discount rate (10%)
n = life of the control equipment (10 years)
MIS = after tax administrative and insurance charge
including State and local property taxes) (2%)
Implicit in the present value of the capital equipment depreciation cal-
culation is a nominal discount rate, a depreciation schedule and an
equipment lifetime. Assuming a straight-line depreciation schedule, the
present value of capital equipment depreciation is calculated using
Equation (4-14):
Depreciation rate = ^ + V* ~ l ' I (4'14)
jd+j)n n
where
j = nominal discount rate
n = control equipment lifetime.
A 25 percent nominal discount rate and a 10-year equipment lifetime was
assumed in the regulatory analysis. Therefore, the depreciation rate
equals 36 percent. Assuming that the real discount rate is 10 percent
and that the nominal discount rate is 25 percent, the rate of inflation
is 13.5 percent. This may be somewhat high for a long-term rate of
inflation and can be changed for future years. However, the social
costs are almost totally insensitive to the assumed inflation rate. The
social costs would change only to the extent that the value of j causes
firms to change their control strategy.
Social costs are calculated using the equations listed above, but the
corporate tax rate, investment tax credit, and depreciation rate are
assumed to be zero. This is done because it is assumed that these costs
4-55
-------
are being paid by society in one form or another, although a firm does
not make its decisions based on social costs.
4.1.5 Uncertainty Analysis
Any analysis of anticipated responses to government actions by thousands
of firms, units of government, and individuals must, by its very nature,
be highly approximate. An attempt was made in the context of this
regulatory analysis to catalog the various sources of uncertainty and to
estimate how variable the final analytical results for the one-hour
standards may be. This requires specifying the variability in each of
the key parameters on which the analysis is based. Although knowledge
about variability in input parameters is rarely sufficient for a quanti-
tative uncertainty analysis, one can usually make reasonable estimates
based on literature reports or "engineering judgment." At the very
least, an uncertainty analysis makes explicit what is known (or not
known) about variability in the key parameters and, by identifying which
parameters contribute most to overall levels of uncertainty, can act as
a guide in directing future research or data gathering activities.
4.1.5.1 Variability of Input Parameters
The mainstream of the regulatory analysis proceeds by employing point
estimates for each input variable in the dispersion modeling and least
cost estimation steps. These estimates are the most likely values for
each variable considered alone. The uncertainty analysis is focused on
developing probability distributions for each variable reflecting the
frequency of occurrence of alternative values. While these distributions
should be viewed as rough approximations illustrative rather than
definitive they do provide a basis for ranking the parameters with
respect to the uncertainty in the point estimates and they do provide
information for a simulation of the effect of variability in all input
parameters on the total estimated control costs for a sample of plants.
4-56
-------
Table 4-19 lists the key input parameters and the associated frequency
distributions of their values. In most cases the distributions are
based on judgment, although the model accuracy and control costs are
based on literature reports. Some distributions are depicted by five
rather than by the minimum three points, reflecting greater confidence
in the shape of the distribution.
The parameters under "Source Data" refer to NEDS data items. Their
distributions roughly reflect the reliability of NEDS, although the
specific values cited are subjective.
The modeling variables refer to 1) the level of dispersion modeling
accuracy generally acknowledged by the modeling community (plus or minus
a "factor of 2" [AMS, 1978]) and 2) idiosyncracies of the one-hour NO-
application. The latter category includes the use of an NO- "background"
value (i.e., the highest annual average in the county) and the specifi-
cation of an 0- level for the OLM. The distribution for 0- is skewed
toward lower values because the second high daily maximum hour values
for 0- and NO are unlikely to occur simultaneously. The control cost
O A
equations for each type of NO control are derived from point estimates
X
of costs for sources of different sizes. Each of the point estimates is
subject to variability, approximations of which appear in the control
technology literature (Section 4.1.3). Two distributions are employed
here, one for those techniques currently in use and one for those still
under development and testing.
Finally, the emission factors used to estimate NO emissions are subject
A
to considerable uncertainty. Because EPA's AP-42 report (U.S. EPA,
1979a) identifies the level of variability qualitatively by a letter
rating, these ratings were used to estimate the distributions.
4-57
-------
oo
Parameters
Source Data
Stack Heights
Stack Temperature
Stack Flow
Source Design Rate
Modeling
Model Accuracy
Source Interaction
0» Value
Control Costs
T G (1)
E R
C 0 (2)
H U
P
MV±100%
MV- 50%
RV+10%
EV±30%
EV+100%
EV-50%
TABLE 4-19
FREQUENCY DISTRIBUTIONS
Probability of Occurrence
0.10 0.15 0.20 0.25 0.30 0.40 0.50 0.60 0.70 0.80
RV±30%
RV±30%
RV±30%
RV±30%
MV+50%
MV-25%
MV±50%
EV±15%
EV+50%
EV-25%
MV
MV
RV-25% RV
EV
EV
RV
RV
RV
RV
-------
TABLE 4-19 (Continued)
FREQUENCY DISTRIBUTIONS
Probability of Occurrence
Ul
VO
Parameters 0.10 0.15 0.20 0.25 0.30 0.40 0.50 0.60 0.70 0.80
Emission Factors
Emission Factor Rating
A RV±10% RV
B RV±20% RV
C RV±30% RV
D RV±40% RV
E RV±50% RV
Note: RV = reported value
MV = modeled value
-------
4.1-5.2 Sample Selection and Simulation Runs
The uncertainty analysis is conducted with a sample of all plants that
need controls in order to meet the standard. A sample of 2 and 12
percent is employed for the 0.10 ppm and 0.25 ppm standards, respec-
tively. These sampling intensities provide a sample size of several
hundred sources the details of which are discussed in Section 5.
The simulation proceeds by randomly selecting a value for each input
parameter based on the frequency distributions in Table 4-19. Concen-
trations of N02 are computed and a least-cost control strategy is esti-
mated for each plant using the hourly methodology and the costs for all
plants aggregated. This procedure is repeated 50 times and a distri-
bution of costs generated. After a mean and a standard deviation are
computed, they are compared with the simple cost estimate generated by
using the single "point" estimate for all input parameters.
4.2 MOBILE AND OTHER AREA SOURCES
4.2.1 Mobile and Other Area Source Emissions Data
As with point sources, the National Emissions Data System (NEDS) is used
as the source of the emission inventory in the mobile and other area
source analysis. In this case, though, the NEDS area source file is
used as the source of NO county emission inventories, which are assumed
A
to represent 1978 conditions. For use in modeling, the NEDS emissions
within a county are divided into the categories listed in Table 4-20.
Emission inventories for 500 counties are included in the analysis.
4.2.2 Air Modeling Approach
The linear rollback model is used here to relate expected changes in
emissions between 1978 (the base year) and 1985 and 1990 to expected
changes in observed N0~ levels. The basic assumption in the linear
4-60
-------
TABLE 4-20
MOBILE AND OTHER AREA SOURCE EMISSION INVENTORY CATEGORIES
Mobile Source
1. Light-duty vehicles
2. Light-duty trucks
3. Heavy-duty gasoline vehicles
4. Heavy-duty diesel vehicles
5. Miscellaneous mobile sources
Fuel Combustion (Area Sources)
6. Residential coal
7. Residential oil
8. Residential gas
9. Commercial coal
10. Commercial oil
11. Commercial gas
12. Industrial coal
13. Industrial oil
14. Industrial gas
15. Solid waste
16. Miscellaneous area sources
4-61
-------
rollback model is that the air quality in an area is directly propor-
tional to total annual emissions within that area. Therefore, given the
air quality value and total annual emissions in the base year, the air
quality in some future year can be estimated based solely on the pro-
jected total annual emissions in the region. The relationship can be
simply stated as follows:
(4-15)
^n ^o
where
AQ. = air quality value in year i
E. = annual emissions in year i
i = o is the base year
i = n is the evaluation year
With the analysis being performed on a county-by-county basis, E is
simply the NEDS area source inventory for that county. The base year
air quality value is determined according to the form of the standard
and the averaging time being analyzed. All of the NO- values in this
analysis were taken from EPA's SAROADS data base. These values were
selected and reviewed by Robert Faoro of the Monitoring and Reports
Branch, OAQPS. When both 24-hour bubbler and continuous monitors are
present in some counties, continuous data were used to select a base
year NCL value. Base year annual average NCL values are determined by
choosing the highest annual average from the past three years of data
(1976-1978). The form of the standard analyzed for the 24-hour and
one-hour standards is equivalent to the form of the current ozone standard
Because, in general, the average number of days per year above the level
of the standard must be less than or equal to one, the base year concen-
trations for the 24-hour and one-hour standards were chosen to be the
level at each site with the expected number of exceedances per year less
than or equal to one. This is the second high daily maximum hour value.
4-62
-------
The highest value over all sites in each county was chosen as the base
year concentration. For the one-hour analysis, if no one-hour data are
available for a county, the highest recorded annual average NO- level is
used along with the typical relationship (6:1) between the second highest
daily maximum hour and annual average NO- to estimate one-hour concen-
trations. Background NO- is assumed to be zero in the analysis, because
natural and transported NO- levels are typically very low.
The projection year emission levels (E ) for each county are estimated
by accounting for expected NO controls and growth and retirement of
sources over the time period of the analysis. For the purposes of this
analysis, it is assumed that the annual growth rate of all sources in
the inventory is one percent.
In estimating how emissions are likely to change between the base year
and the projection year in the base case, only mobile sources are assumed
to be controlled. The reduction in emissions due to the Federal Motor
Vehicle Control Program (FMVCP) is estimated using EPA's Mobile Source
Emission Factor Program (MOBILE2)- The composite emission factors for
all vehicle types combined are:
Calendar Composite Emission Percentage Reduction
Year Factor (gm/mi) from Base Year
1978 4.32
1985 3.18 26%
1990 2.21 49%
These percentage reductions are assumed to apply to the first four
mobile source categories listed in Table 4-20. The miscellaneous mobile
source category is not assumed to have any controls.
All of the MOBILE2 emission factors applied in the mobile source analysis
are representative of low-altitude 49-State vehicles. While it would
4-63
-------
have been preferable to use high-altitude and California emission factors
for the counties in those particular areas this was not done for two
reasons: 1) MOBILE2 emission factors for high-altitude areas and California
were not released at the same time as the low-altitude emission factors
and 2) the NOY inspection and maintenance credits were only available
A
for the low-altitude case.
No stationary area source controls are assumed to be applied in the base
case; however, for areas where they are needed to attain the standard,
the costs and effectiveness of controls differ for new and existing
sources, with new source controls being imposed at some future date
beyond the base year but before the evaluation year. Therefore, it is
important to differentiate between growth in new sources and retirement
of existing sources. Existing stationary area source emissions in the
evaluation year are estimated using Equation (4-16):
EES.. = EES . (1 + g)m (1 - r)n"m (4-16)
J J
where
EES.. = emissions from existing stationary area source
** category j in year i
EES . = emissions from stationary source category j in the base
^ year
g = annual growth rate
r - annual retirement rate
m = year new source regulations go into effect (base
year)
n = evaluation year (base year)
4-64
-------
This equation accounts for growth (1 + g) until the year that new source
regulations become effective (assumed to be 1984 in this analysis).*
Hereafter, new sources are distinguished from existing sources. The
existing inventory is reduced by the retirement rate (1 - r) over the
remaining years until the evaluation year is reached (1985 or 1990 in
this analysis).
The new stationary area source emissions are calculated from the existing
stationary area source emissions in the base year using equation (4-17):
[([1 + g]n"m -1) + (!-[!- r]n"m)]
vn-m
where
= EES... [(1 + g)u m - (1 - r)u U1] (4-17)
ENS. = emissions from new stationary area source category
j in year i
This equation adds together emissions from new sources responsible for
net growth plus those that replace existing sources retiring over the
years (n-m). Both Equations (4-16) and (4-17) assume that sources emit
the same rate as existing ones at uncontrolled rates.
The equivalent equation for estimating growth and retirement of mobile
sources is much simpler because the retirement of the existing vehicle
fleet is incorporated in the mobile source emission factor calculations.
Equation (4-18) shows how mobile source emissions in the evaluation year
are calculated:
* This is likely to be the first year SIP's become effective given the
time "heeded to promulgate a new NAAQS and the time allowed for SIP
preparation and approval.
4-65
-------
where
EMi. = EMQ. (1 + eg)1" (4-18)
EM.. = emissions from mobile source category j in year i
eg = annual rate of change in emissions due to FMVCP emission
standards and to turnover in the vehicle fleet.
Using the air quality value and the total emissions in the base year and
the N0_ standard in the evaluation year, the maximum allowable countywide
emissions in the evaluation year are calculated as follows:
Emax = A(3S (4'19)
AQo
where
E = maximum allowable emissions
max
AQ = ambient NO- standard.
s z.
In order for the county under study to meet the given NCL standard,
total annual emissions in the evaluation year must be less than or equal
to E .If the total emissions in the evaluation year are less than
max J
the maximum allowable emissions, then the county is assumed to meet the
standard with no emission control necessary. If total emissions are
greater than E then the difference between total emissions and E
0 max max
represents the needed emissions reduction in the county. In this analysis,
controls are applied one at a time, until either the needed reduction is
achieved or all available controls are exhausted.
4.2.3 Control Options and Costs
Besides the FMVCP controls assumed in the base case, three other control
options are evaluated for counties that cannot attain an NCL standard
with baseline controls. These control options are:
4-66
-------
Inspection and maintenance for light-duty vehicles
Transportation control measures
Stationary area source controls.
These controls are assumed to be applied in the order in which they are
listed. Therefore, a nonattainment county is assumed to consider an
inspection and maintenace (I&M) program first and, if attainment is
still not achieved, transportation control measures (TCM's) secondarily.
Then, if the standard is still not met, stationary area source controls
are considered. In addition to this schedule of control strategy imple-
mentation, another constraint is imposed: if any county within an
urbanized area (as defined by the Bureau of the Census) requires an I&M
or TCM program, then all counties in the area must implement the same
type of program.
4.2.3.1 Inspection and Maintenace
Since I&M programs currently monitor only CO and HC emissions, it is
difficult to predict with any certainty what the exact reductions in NO
will be from an I&M program. All of the evidence that can be used to
estimate the potential reductions from a NO inspection is preliminary
A.
and subject to change. The source of the information used in this
analysis is a memorandum prepared by the Inspection and Maintenance
Staff of EPA's Office of Mobile Source Air Pollution Control (U.S. EPA,
1981).
Unlike HC and CO, NO emissions from cars increase as vehicle speed
increases. Consequently, if the idle test presently used in I&M programs
for measuring CO and HC were used to measure NO , test values would not
A
correlate well with actual in-use NO emissions. Instead, an I&M program
A
designed to test for NO would be more complex than any existing I&M
A
program. The two systems currently viewed as practical for NO control
X
4-67
-------
are 1) an inspection of the exhaust gas recirculation (EGR) system and
2) a loaded mode (dynamometer) test under varied driving conditions.
The EGR inspection would simply be added to idle test regime. Light-duty
vehicles with obviously malperforming EGR systems would be identified
and maintenance would be required just on this group. A dynamometer
test would allow emission measurements to be made at high speed load
conditions under which the highest NO levels occur.
A
Cost estimates for NO -related ISM take into account the assumptions
A.
listed in Table 4-21. As noted in the table, it is assumed that States
with decentralized ISM programs planned for HC and/or CO cannot use the
loaded mode test alternative, because it is unlikely that private garages
would buy the dynamometers required for such testing. Assuming the
county is in a State with a centralized program, the loaded mode test is
the first option considered. As can be seen from Table 4-22, it is the
least expensive I&M option and it has the lowest emission reduction
effectiveness. If the loaded test provides enough emission reduction to
bring the county into attainment, then no other control options are
considered. A capital cost of $10,000 (for a dynamometer) is assessed
for every 25,000 LDVs registered in the county.
If the loaded test does not bring the county into attainment, then an
EGR inspection is assumed. There is no capital charge for this method,
but there is an inspection and repair fee per LDV, which is shown in
Table 4-22.
4.2.3.2 Transportation Control Measures
If control beyond I&M is needed, transportation control measures (TCM's)
are considered. TCM's represent a percentage reduction in vehicle miles
traveled (VMT). Present guidance on NO emissions from TCM's deals only
A
with the following programs:
4-68
-------
TABLE 4-21
ASSUMPTIONS USED IN THE NOV I&M COST ESTIMATES
X
1. Investment costs for dynamometers:
Constant load = $10,000
Variable load = $12,000
2. Constant speed dynamometers are the most likely to be used in a
loaded test.
3. The assumption of 25,000 inspections per lane per year is appropriate
for a loaded test.
4. The I&M program for NO in an urbanized area is dependent on the
program in place. A loaded test is not a viable option in an area
with a decentralized program.
5. The loaded test and EGR inspection apply to all post-1972 vehicles.
6. The number of cars in each county is used to estimate costs.
7. County passenger car counts were obtained from the 1979 Commercial
Atlas and Marketing Guide published by Rand McNally and Co.
8. I&M programs are assumed to begin operation in 1984.
9. I&M programs are assumed to apply only to light-duty vehicles.
4-69
-------
TABLE 4-22
NO RELATED INSPECTION AND MAINTENACE
CONTROL EFFECTIVENESS AND COSTS
Control
1985 Composite Emission Factor
(gpm)
1990 Composite Emission Factor
(gpm)
1985 Per Vehicle Cost
I&M already in place (1980 $)
1985 Per Vehicle Cost
New program (1980 $)
1990 Per Vehicle Cost
I&M already in place (1980 $)
1990 Per Vehicle Cost
New program (1980 $)
Loaded Mode
Test
3.02
(5% red.)
2.01
(9% red.)
3.39
11.90
0.89
8.87
EGR Inspection
Centralized Decentralized
2.93
(8% red.)
1.94
(12% red.)
4.16
12.62
1.57
9.57
3.15
10.05
1.39
8.29
SOURCE: U.S. EPA, 1981. Memorandum: Charles Gray to Joseph Padgett,
"Effectiveness of I/M for NO Emission Control." OMSAPC, Ann Arbor,
MI. January 12.
4-70
-------
Bus lanes
Carpool/vanpool
Transit
- Fare reduction
- Service improvements.
The effectiveness of these TCM's is analyzed by a combination of observed
and model-estimated impacts (U.S. EPA, 1978). If these measures were
instituted in a large urban area, it is estimated that regional NO
A
emissions of highway vehicles would be reduced by 1.9 percent. This
percentage is assumed to apply to the emissions of all vehicle types.
Cost-effectiveness values are estimated as follows: these measures are
assumed to result in a 3.3 tons per day reduction in regional weekday
NO emissions in an urbanized area with a population of one million or
X
more. Assuming 260 weekdays in a year yields a 360 tons per year NO
X
reduction from these TCM's. The program costs of these measures (in
1976 dollars) are estimated to be $23.6 million capital cost and $7.2
million annual cost. Converting to 1980 dollars with the GNP deflator,
annualizing and putting on a tons per year NO reduced basis gives a
X
cost of $17,200 per ton per year NO reduced.
X
Urbanized areas on the nonattainment list for CO or 0« are assumed to be
implementing TCM's and no cost for these programs is attributed to NO .
4.2.3.3 Stationary Area Sources
The final control alternative considered is to reduce emissions from
stationary area sources. For each category of area source emissions,
both existing and new, a set of alternative control strategies is speci-
fied. These control options and their costs are detailed in Table 4-23.
The stationary area source cost calculation procedures are explained in
detail in Appendix D.
4-71
-------
TABLE 4-23
NO STATIONARY AREA SOURCE CONTROL COSTS
X (1980 $)
Control
Source Type Fuel Type
New
Small Industrial
Boilers Coal SCA
NH3 Inj.
Distillate LNB
FGR
*^
^ Gas RAP + LNB
RAP + FGR
Commercial
Boilers Coal SCA
NH3 Inj.
Distillate LNB
FGR
Gas RAP + LNB
RAP + FGR
Residential
Furnaces Distillate LNB
Control
Efficiency
20
50
50
55
70
75
20
50
50
55
70
75
55
Capital Cost
($/ton Removed)
3900
3000
8500
7700
5200
4800
15,500
8000
22,900
20,800
19,700
18,400
10,700
O&M
Cost
110
450
800
810
1200
1460
130
450
800
810
1290
1580
0
Fuel Penalty
(10 Btu/ton
Removed)
0
0
0
110
610
730
0
0
0
110
610
730
(1340)
-------
TABLE 4-23 (Continued)
NO STATIONARY AREA SOURCE CONTROL COSTS
X (1980 dollars)
Control
Source Type Fuel Type
Existing
Small Industrial
Boilers Coal SCA
NH3 Inj.
Distillate LNB
FGR
^ Gas RAP + LNB
i
w RAP + FGR
Commercial
Boilers Coal SCA
NH3 Inj.
Distillate LNB
FGR
Gas RAP + LNB
RAP + FGR
Residential
Furnaces Distillate None
Control
Efficiency
(%)
20
50
50
55
70
75
20
50
50
55
70
75
Capital Cost
($/ton Removed)
5100
3000
9700
8900
2700
2500
20,900
8,000
26,300
23,900
7200
6700
O&M
Cost
340
440
1460
1390
980
1180
520
440
1460
1390
980
1180
Fuel Penalty
(10 Btu/ton
Removed)
0
0
0
110
610
730
0
0
0
110
610
730
SOURCE: Adapted from information in the industrial boiler ITAR (U.S. EPA, 1979). Residential furnace control
and cost information was taken from the NO Control Techniques Document (U.S. EPA, 1978).
X
-------
4.2.4 Least Cost Modeling
The problem of selecting the least expensive set of control strategies '
that will bring the total emissions in the county below E is formu-
6 * max
lated in the same manner as the least cost model for point sources
(Section 4.1.5 describes the model). The only difference between the
two least cost routines is that for ^stationary area sources, the model
examines control strategies for source categories, rather than for
individual emitters.
Once an optimal solution has been reached for each county individually,
a check is made to ensure that all counties within each urbanized area
have applied a consistent set of I&M and TCM programs. If any county
within an urbanized area has implemented an I&M program and/or a TCM
program, all counties within the urbanized area are assumed to implement
a similar program.
4-74
-------
REFERENCES FOR SECTION 4 (Continued)
U.S. EPA. 1979b. Technology Assessment Report for Industrial Boiler
Applications: NO Combustion Modification (EPA-600/7-79-178f). Indus-
trial Environmental Research Laboratory. Research Triangle Park, NC.
December.
U.S. EPA. 1979c. Technology Assessment Report for Industrial Boiler
Applications: NO Flue Gas Treatment (EPA-600/7-79-178g). IERL,
Research TriangleXPark, NC. December.
U.S. EPA. 1980. User's Guide to MOBILE2: Mobile Source Emissions Model.
Ann Arbor, MI.
U.S. EPA. 1981. Memorandum: Charles Gray to Joseph Padgett, "Effec-
tiveness of I/M for NO Emission Control." Emission Control Technology
Division, OMSAPC, Ann ^rbor, MI. January 12.
4-76
-------
5. NATIONWIDE ANALYSIS RESULTS
This section presents the estimated expected costs of control to meet or
approach as closely as possible alternative NO- standards. In keeping
with the separate methodologies for estimating nationwide costs and
attainment status for point sources and for mobile and area sources,
costs are presented separately for these source categories in Sections
5.1 and 5.2, respectively; and then combined in Section 5.3. Section
5.4 presents the point source cost uncertainty analysis. The sensitivity
of the mobile and area source costs to changes in some key input variables
is discussed in Section 5.5. Section 5.6 presents the caveats associated
with the nationwide analysis results.
5.1 POINT SOURCE NATIONWIDE ANALYSIS RESULTS
This section presents the results of the nationwide analysis of point
source costs expected to be incurred at alternative NO- standard levels,
with the results of the one-hour, 24-hour, and annual standard analyses
discussed separately. While costs are presented for both 1985 and 1990,
most of the discussion deals with the 1990 costs for ease of comparison.
All costs are expressed in 1980 dollars.
5.1.1 One-Hour Standard Analysis
The point source control cost estimates for alternative one-hour N00
standards are shown in Table 5-1. For each alternative standard, infor-
mation is presented on the number of sources controlled, various costs
of control that standard, and the additional fuel (or fuel savings)
needed to operate the control equipment. The annualized cost represents
the sum of the annualized capital charge, the operating and maintenance
(O&M) cost, and the fuel cost. The number of plants not able to attain
the standard with maximum control is also indicated.
5-1
-------
TABLE 5-1
POINT SOURCE CONTROL ESTIMATES FOR
ALTERNATIVE ONE-HOUR N02 STANDARDS
(costs in millions of 1980 dollars)
One-Hour
NO Standard
(ppm)
0.50
0.35
0.25
0.20
0.15
0.10
Number of
Sources
Controlled
272
875
2,735
4,251
8,289
11,652
Annualized
Cost
$ 29.6
102.7
662.4
1,320.2
4,264.1
7,232.6
Initial
Cost
$ 97.1
316.3
1,930.7
3,985.5
12,500.2
21,018.8
O&M
Cost
$ 5.1
27.9
197.7
337.8
1,060.4
1,827.6
Fuel
Cost
$ 4.8
10.7
73.1
174.1
668.6
1,142.3
Fuel
Use
(10 Btu)
1,089
2,337
19,409
49,677
201,503
333,579
Number of
Nonattain-
ment Plants
4
12
37
73
195
466
-------
As can be seen in Table 5-1, the costs escalate rapidly once the stan-
dard reaches 0.25 ppm, as compared to the costs at the 0.35 ppm and 0.50
ppra levels. Moving from primary to secondary or higher levels of control
causes substantial escalation of costs. This is reflected in the average
cost per controlled source: about $120,000 for the 0.35 ppm standard
and approximately $515,000 for the 0.15 ppm standard. Thus, not only
are more sources controlled at the lower standards, but some sources
that have to control to meet the 0.50 or 0.35 ppm standards must install
more expensive control equipment to meet 0.25 ppm or lower.
Table 5-2 preserits an industry-by-industry breakdown of the estimated
annualized cost of controls needed to meet alternative one-hour NCL
standards. For all standards tested, utilities incur the major cost
burden; for the most stringent standards, they bear 80 percent of the
annualized costs for all point sources. Other industries with signifi-
cant control costs at the lower level standards include iron and steel,
chemical manufacturing, petroleum refining, and oil and gas extraction.
These industries contain a wide variety of sources (see Table 5-3) and
types of stacks, implying that the meteorological conditions specified
in the dispersion analysis create unfavorable conditions for all source
types individually and for various combinations thereof.
A closer look at the utility cost reveals that a large percentage of the
sources needing NO reduction must control to the maximum possible level
a
selective catalytic reduction (SCR). For the 0.10 ppm one-hour N0_
standard, 40 percent of the sources reach maximum control. This produces
extremely high costs because flue gas treatment is several times more
costly than less effective controls.
Even with maximum source control, not all plants can attain the standards,
This is due to the large number of sources in some plants (100 or more),
leading to high ambient levels of NO , in combination with high levels
A
5-3
-------
TABLE 5-2
ANNUALIZED COSTS ESTIMATED TO BE INCURRED BY MAJOR INDUSTRIAL GROUPS
ALTERNATIVE ONE -HOUR N02 S
(Millions of 1980 Dollars)
TO MEET ALTERNATIVE ONE -HOUR N02 STANDARDS
One-Hour N0~ Standards (ppm)
SIC
Code Industry 0.50 0.35 0.25 0.20 0.15 0.10
131
28
29
33
491
Oil and Gas Extraction
Chemical
Petroleum Refining
Iron and Steel
Utilities
Miscellaneous
$ o
2
0
0
23
5
$ 1
7
0
10
76
9
$ 55
14
0
46
400
147
$ 77
21
9
58
940
215
$ 99
72
48
101
3,421
523
$ 130
159
115
151
5,807
871
TOTAL $30 $103 $662 $1,320 $4,264 $7,233
5-4
-------
TABLE 5-3
POINT SOURCE CONTROL ESTIMATES BY SOURCE CATEGORY
TO MEET A 0.25 PPM ONE-HOUR N0£ STANDARD
(Costs in millions of 1980 dollars)
Number of
Source Sources
Category Controlled
Industrial Boilers
01 Coal
ui Oil
Natural Gas
Utility Boilers
Coal
Oil and Gas
Stationary Gas Turbines
Reciprocating 1C Engines
Industrial Process Furnaces
Nitric Acid Plants
60
785
390
177
451
184
921
954
10
Annualized
Cost
$ 30.2
56.8
15.2
214.8
267.1
12.7
22.9
40.6
2.1
Initial
Cost
$ 66.9
125.3
29.5
750.8
696.1
46.8
58.3
150.4
6.6
O&M
Cost
$ 16.6
34.3
13.1
46.2
73.3
0.3
7.4
5.7
0.8
Fuel
Cost
$ o
-3.0
-3.9
16.4
52.6
2.9
3.6
4.5
0
Fuel Use
(10 Btu)
43
-513
283
7988
9552
466
694
892
0
TOTAL (2,932 $662.4 $1,930.7 $197.7 $73.1 19,4o|)
-------
of 0_ near these plants. Recall also that where 0_ is limiting, 10
percent NO control will only lower ambient NO- by one percent (down to
X ^
the point where NO becomes limiting). For the more stringent standards
X
(0.20 ppm and below), the estimated number of plants unable to attain
the standard approaches and then considerably exceeds 100.
5.1.2 24-Hour Standard Analysis
The point source control cost estimates for alternative 24-hour NO-
standards are shown in Table 5-4. The control costs increase rapidly
after the 0.08 ppm standard is reached for the same reasons given for
trends in the one-hour standard results. The use of the same rule for
representing plant interaction as is employed in the one-hour analysis
(i.e., highest countywide annual NO- level) probably overstates the
degree of interaction because the more stringent standards assessed are
close to the average "background" level (about 0.03 ppm). Thus, for a
background level of 0.03 ppm N0_, each plant can contribute no more than
0.03 ppm in order to meet a 0.06 ppm 24-hour standard.
The breakdown of costs by industry to meet a 24-hour NO- standard is
approximately the same as for a one-hour standard. For instance, utili-
ties are expected to incur 70 percent of the costs to meet a 0.06 ppm
24-hour N0_ standard. In fact, since ratios of one-hour to 24-hour
ambient NO- concentrations are estimated to be about 5:1, the costs to
meet a 24-hour standard should approximate those to meet an hourly
standard that is five times less stringent. Actually, the costs for the
24-hour standard should be somewhat higher than for its equivalent
one-hour standard due to the use of the same "background" NO- level in
each analysis.
5.1.3 Annual Standard Analysis
Table 5-5 shows the estimated industry control costs needed to attain
the current 0.053 ppm annual average NAAQS assuming 100 percent conversion
5-6
-------
TABLE 5-4
POINT SOURCE CONTROL ESTIMATES
ALTERNATIVE 24-HOUR N02 STANDARDS
(Costs in millions of 1980 dollars)
Ul
I
24-Hour
NO Standard
(ppm)
0.14
0.12
0.10
0.08
0.06
Number of
Sources
Controlled
247
506
1,227
2,785
4,907
Annual ized
Cost
$ 26.2
75.5
216.2
893.0
2,214.7
Initial
Cost
$ 67.0
181.2
644.6
2,378.4
6,125.9
O&M
Cost
$ 4.1
22.2
48.9
269.7
611.0
Fuel
Cost
$ 8.5
16.6
36.5
140.9
361.4
Fuel
Use
(10 Btu)
1,378
2,641
6,505
32,801
87,369
Number of
Nonattain-
ment Plants
3
5
12
89
223
-------
SIC
Code
13
28
29
33
49
TABLE 5-5
ESTIMATED INDUSTRY CONTROL COSTS FOR
A 0.053 PPM ANNUAL AVERAGE N02 STANDARD*
(Costs in millions of 1980 dollars)
Industry
Oil and Gas Extraction
Chemical
Petroleum Refining
Iron and Steel
Utilities
Miscellaneous
Totals
, '~i~
*'
Annualized
Cost
$ 8.7
0
1.6
1.2
3.7
0.8
Initial
Cost
$23.2
9.1
4.4
4.4
9.0
1.7
$15. r
t-i.
$52.0
* 100% conversion of NO to NO, is assumed.
5-8
-------
of NO to N0~. Most of the costs to meet this standard are incurred by
the gas transmission industry, the most probable explanation being that
gas transmission compressor stations using 1C engines with short stacks
are more likely to produce high annual average concentrations than
plants that emit more NO but have taller stacks. This is because the
meteorological conditions that cause high ground level NO concentrations
from a NO source with short stacks occur more frequently during the
X
year than those producing high N02 levels from tall stacks. However,
the meteorological conditions that produce high ground level NO- from
tall stack sources (such as powerplants) occur often enough to cause
high one-hour concentrations for a tew hours per year. This is borne
out by the distribution of controls and costs by source type: 339.out
of a total of 547 sources needing control and $15-6 million of a total
$15.7 in annualized costs accrue to 1C engines. Interestingly, some
industries and source categories show cost savings, reflecting the fuel
saving effects of combustion modifications.
Assuming 50 percent NO conversion or using the OLM to estimate N0_
levels, substantially reduces the costs. Under the 50 percent conversion
scenario, annualized costs drop to about $2.6 million; OLM brings the
costs to almost zero.
5.2 MOBILE AND AREA SOURCE ANALYSIS RESULTS
5.2.1 One-Hour Standard Analysis
Mobile and area source control cost estimates for alternative one-hour
N0_ standards are presented in Table 5-6 for 1985 and in Table 5-7 for
1990. These tables show that no costs are incurred unless the one-hour
standard is less than 0.35 ppm. On a county-by-county (or urbanized
area) basis, if an I&M program or I&M plus TCM's are sufficient to bring
an area into attainment, then the costs are relatively modest. However
if stationary area source controls are needed, costs escalate rapidly.
5-9
-------
TABLE 5-6
MOBILE AND AREA SOURCE CONTROL ESTIMATES
ALTERNATIVE ONE-HOUR N02 STANDARDS IN 1985
(Costs in millions of 1980 dollars)
One-Hour
Number of
Number of
NO Standard
(ppm)
0.50
01 0.35
S 0.25
0.20
0.15
0.10
Counties
With Controls
0
1
47
116
203
346
Annualized
Cost
$ o
2.6
244.1
681.3
2,709.5
4,244.5
Initial
Cost
$ 0
2.5
214.2
732.4
2,503.7
3,900.2
O&M
Cost
$ o
2.2
102.3
302.3
863.4
1,368.0
Fuel
Cost
$ o
0
107.0
259.7
1,438.5
2,241.6
Nonattainment
Counties
0
0
3
19
66
181
-------
TABLE 5-7
MOBILE AND AREA SOURCE CONTROL ESTIMATES
ALTERNATIVE ONE-HOUR N02 STANDARDS IN 1990
(Costs in millions of 1980 dollars)
One-Hour
Number of
Number of
Cn
i
N02 Standard
(ppm)
0.50
0.35
0.25
0.20
0.15
0.10
Counties
With Controls
0
0
8
70
165
286
Annualized
Cost
0
0
96.3
276.6
1,501.8
3,348.8
Initial
Cost
0
0
58.4
340.5
1,820.7
3,792.4
O&M
Cost
0
0
31.0
101.6
467.7
984.5
Fuel
Cost
0
0
55.8
119.5
737.7
1,746.9
Nona tta inment
Counties
0
0
1
4
32
119
-------
Because mobile source emissions predominate in most areas, the FMVCP,
I&M, and TCM's are relatively effective in enabling those areas to meet
the standard. However, certain urbanized areas have high emissions in
some of the stationary area source categories, especially industrial
natural gas use. In these areas, mobile source controls are relatively
ineffective in reducing NO- levels and the more expensive stationary
area source controls shown in Table 4-14 are needed. It should be noted
that recent NEDS estimates of NO emissions from small industrial natural
x
gas users are much higher than previous estimates. This is because a
change in the procedure for estimating NO emissions from this category
A
was made by EPA to incorporate 1C engine emissions associated with gas
transmission and oil field uses. Note that these sources are similar to
but smaller than the clusters of 1C engines and gas turbines included in
NEDS point source file.
Table 5-8 distributes the estimated costs among the various control
options for mobile and area sources. On average, 75 percent of the
annualized costs of control needed to attain (or approach as closely as
possible) alternative one-hour N02 standards will be incurred by small
industrial sources, which are expected to be predominately natural gas
transmission units (especially in States like Louisiana and Texas).
Another implication that can be drawn from Table 5-8 is that a loaded
mode I&M test rarely provides enough NO reduction to bring a nonattain-
A
ing county into attainment. Therefore, EGR inspection is the required
I&M alternative in most cases.
The estimated control costs fo the most stringent standard drop between
1985 and 1990 from about $4.2 to $3.3 million per year or by about
20 percent per year. This reflects the effect of increasingly stringent
FMVCP emission standards during the early 1980's and the turnover of the
vehicle stock during the entire decade.
5-12
-------
TABLE 5-8
MOBILE AND AREA SOURCE CONTROL COST ESTIMATES BY CONTROL TYPE
FOR ALTERNATIVE ONE-HOUR N02 STANDARDS IN 1990
(Millions of 1980 Dollars)
en
Control
I&M Setup
Loaded Mode Test
EGR Inspection
TCM's
Stationary Area Sources
Residential
Commercial
Industrial
TOTAL
One-Hour NO Standards (ppm)
0.10
$57.4
5.7
177.4
85.0
-118.4
487.3
2,654.5
$3,348.9
0.15
$ 37.3
1.4
110.8
38.5
-53.8
215.6
1,152.0
$1,501.8
0.20
$ 14.9
1.8
38.7
10.0
-7.2
46.5
171.9
$276.6
0.25
$ 1.3
0.4
11.0
1.7
0
0.6
81.3
$96.3
0.35
$0
0
0
0
0
0
0
$0
-------
Few counties are unable to attain the standard until it reaches 0.20 ppm.
Thereafter, nonattainment increases rapidly. However, the FMVCP effects
a substantial decrease in the number of nonattainment counties between
1985 and 1990. For example, the number not attaining the 0.15 ppm
one-hour standard is cut approximately in half.
5.2.2 24-Hour Standard Analysis
Tables 5-9 and 5-10 summarize the results of the mobile and area source
analysis for alternative 24-hour average NO- standards in 1985 and 1990,
respectively. The distribution of costs by control type for the 24-hour
average analysis is essentially the same as for the one-hour standard
analysis (see Table 5-8). The major difference between the estimated
costs for the two averaging times is the distribution of base year NO-
concentrations with respect to the standards. In other words, the
24-hour average base year NO- values are more evenly spread throughout
the range of alternative 24-hour standards considered than are the
one-hour values with respect to their alternative standards. Recall
that most of the one-hour NO- values used in the mobile and area source
analysis are extrapolated from annual averages by assuming a constant
ratio (6:1), while the 24-hour values are taken directly from monitoring
records. Presumably, this at least partially explains the differences
in distributions of NO- levels. The net effect is that the change in
cost between alternative 24-hour NO- standards is generally less than
that for the one-hour standard alternatives.
The costs drop even more dramatically with time then they do for the
one-hour standard. For the most stringent standard, the change is
greater than 30 percent.
A comparison of Tables 5-7 and 5-9 reveals that the equivalently stringent
one-hour standard is far less than five times the 24-hour standard, as
it is in the point source analysis. In other words, the ratio of monitored
5-14
-------
TABLE 5-9
MOBILE AND AREA SOURCE CONTROL ESTIMATES
ALTERNATIVE 24-HOUR N02 STANDARDS IN 1985
(Costs in millins of 1980 dollars)
24-Hour
Number of
Number of
NO Standard
(ppm)
0.14
0.12
0.10
0.08
0.06
Sources
With Controls
34
51
109
158
2.61
Annualized
Cost
$ 240.6
361.3
667.9
1,005.9
2,265.9
Initial
Cost
$ 223.6
330.0
669.3
1,072.5
2,396.0
O&M
Cost
$ 93.1
149.7
306.3
430.5
810.8
Fuel
Cost
$ 111.1
157.9
252.7
400.9
1,065.0
Nonattainment
Counties
5
9
21
44
105
-------
TABLE 5-10
MOBILE AND AREA SOURCE CONTROL ESTIMATES
ALTERNATIVE 24-HOUR N02 STANDARDS IN 1990
(Costs in millions of 1980 dollars)
Ln
24-Hour
Number of
Number of
N02 Standard
(ppm)
0.14
0.12
0.10
0.08
0.06
Counties Annualized Initial
With Controls Cost Cost
16
31
58
133
220
$ 134.4
200.2
338.4
679.8
1,525.1
$ 137.1
214.3
445.9
911.9
2,098.1
O&M
Cost
$ 41.4
61.4
113.7
251.0
495.7
Fuel
Cost
$ 70.7
103.9
152.1
280.4
687.8
Nona t ta inment
Counties
1
4
7
22
62
-------
second high daily maximum hourly concentrations to second high 24-hour
levels at the same site are less than 5:1. This presumably reflects
less variability in records from monitors near mobile and area sources
than those oriented toward point sources alone.*
5.2.3 Annual Standard Analysis
Estimated costs to meet the 0.053 ppm annual standard are summarized in
Table 5-11. Gas transmission units may incur up to 35 percent of the
$39 million annualized cost to meet this standard. Significantly, few
mobile source controls are required and the residential categories of
stationary area sources experience net returns on their investment.
5.3 COMBINED RESULTS OF POINT SOURCE AND MOBILE AND AREA SOURCE ANALYSIS
Figures 5-1 through 5-4 summarize the total capital and annualized costs
needed to meet the alternative short-term NO- standards considered.
These costs represent the point source costs and the mobile and area
source costs combined. For the one-hour standards, the costs escalate
rapidly as the standard is lowered. For the 24-hour standards, this
increase in costs is more gradual.
It should be noted that the slight bulge in each of the one-hour standard
curves (near 0.15 ppm) is due to the fact that 0_ becomes limiting for
many sources at about this level. That is, reducing NO at a plant is
X
only 10 percent effective in reducing NO,, down to the point where NO
« A
drops below 0 (i.e., where 0 is no longer limiting). After this point
is reached, reducing NO is 100 percent effective in reducing N00.
X £
Thus, the steep increase in control costs with increasingly stringent
standards slows before increasing exponentially once again. The bulge
* Recall also that the point source ratio was obtained from modeling
rather than monitoring data.
5-17
-------
TABLE 5-11
MOBILE AND AREA SOURCE CONTROL COST ESTIMATES
FOR THE ANNUAL STANDARD ,(0-053 PPM) IN 1990
(millions of 1980 dollars)
Control Options
Initial
Cost
Annualized
Cost
Mobile and Area Sources
I & M Setup
Loaded Mode Test
EGR Inspection
TCM's
0
0
0
0
0
0
14.7
0
Stationary Area
Residential
Commercial
Industrial
2.5
35.8
33.1
-2.0
12.4
13.4
TOTAL
$ 71.4
38.5
5-18
-------
FIGURE 5-1
TOTAL ANNUAL1ZED COSTS OF CONTROL TO MEET
ALTERNATIVE ONE-HOUR NOz STANDARDS IN 1990
ANNUALIZED COST (BILLIONS $)
16
14-
12-
10-
4-
2-
0.4
0.5 0.6
ONE-HOUR STANDARDS (ppm)
5-19
-------
FIGURE 5-2
TOTAL CAPITAL COSTS OF CONTROL TO MEET ALTERNATIVE
ONE-HOUR N02 STANDARDS IN 1990
CAPITAL COST (BILLIONS $)
45
40
35-
30-
25-
20-
15-
10-
0.2
0.3
0.4 0.5 0.6
ONE-HOUR STANDARDS (ppm)
5-20
-------
FIGURE 5-3
TOTAL ANNUAL1ZED COSTS OF CONTROL TO MEET
ALTERNATIVE 24-HOUR N02 STANDARDS IN 1990
ANNUALIZEO COST (BILLIONS S)
6
5-
3-
2-
1-
i
.06
i
.08
.10
I
.12
.14 .16
24-HOUR STANDARDS (ppm)
5-21
-------
FIGURE 5-4
TOTAL CAPITAL COSTS OF CONTROL TO
MEET ALTERNATIVE 24-HOUR N02 STANDARDS IN 1990
CAPITAL LOST (BILLIONS S)
!£
14-*
12-
10-
4-
2-
.06
i
.08
.10
J2 .14 .16
24-HOUR STANDARDS (ppm)
5-22
-------
is non-existent for the 24-hour average curves because the 24-hour
standards analyzed are not stringent enough to cause CL to be limiting.
(Recall that the 5:1 one-hour-to-24-hour ratio would require the 24-hour
standard to be 0.03 ppm for equivalency with the 0.15 ppm one-hour
standard.)
A summary of the annual average standard analysis results for all sources
combined is shown in Table 5-12, which presents costs by source category.
Small industrial sources are estimated to incur a large portion of the
costs to meet the current annual standard.
5.4 POINT SOURCE UNCERTAINTY AND PLUME BOUYANCY SENSITIVITY ANALYSES
The uncertainty in the costs and effects of meeting alternative one-hour
NO- standards was investigated through a series of simulations, the
methodology for which was described in Section 4.1.5. In addition, the
effect of sulfur dioxide scrubbing on plume bouyancy and, thus, ground-
level NO concentrations was examined through a limited dispersion
X
modeling analysis of sample powerplants.
5.4.1 Uncertainty Analysis
5.4.1.1 Sample Composition
As noted before, a 2 percent sample of all plants with estimated NO
A
levels greater than 0.01 ppm was drawn randomly from the reduced NEDS
point source file of NO emitters for the 0.10 ppm NO- standard analysis;
X £t
a 12 percent sample of all plants with NO levels above 0.25 ppm was
X
selected for the 0.25 ppm NO- standard analysis. These sampling inten-
sities generated samples of sufficient size for estimating cost dis-
tributions.* A distribution of sources by SIC for each sample and the
For the 0.10 ppm standard, the NO "cut-point" is low enough that
virtually all plants that could exceed the standard would be included
in the population to be sampled. The same is true for the 0.25 ppm
standard despite the high NO "cut-point," because very high levels
of NO are needed to boost NO- above the 0_ concentrations, most of
which are below 0.20 ppm.
5-23
-------
TABLE 5-12
NATIONWIDE COST SUMMARY (1990)
ANNUAL STANDARD = 0.053 PPM
(millions of 1980 dollars)
Initial Annualized
Control Options Cost Cost
Mobile and Area Sources
I & M Setup $ 0 $ 0
Loaded Mode Test 0 0
EGR Inspection 0 14.7
TCM's 0 0
Stationary Area
Residential 2.5 -2.0
Commercial 35.8 12.4
Industrial 33.1 13.4
Point Sources
Industrial Boilers 3.0 -1.6
Utility Boilers 0 0
Reciprocating 1C Engines 38.3 15.6
Industrial Processes 10.6 1.7
TOTAL $123.3 $54.2
5-24
-------
TABLE 5-13
SIC
Code
13
20
22
24
26
28
29
33
37
49
TOTAL
DISTRIBUTION OF SOURCES BY SIC GROUP IN THE SAMPLE
AND FOR ALL PLANTS NEEDING CONTROL TO MEET THE 0.10 PPM
ONE-HOUR NO STANDARD
Industry
Number of Sources Needing Control
Oil & Gas Extraction
Food Processing
Textiles
Lumber
Paper
Chemicals
Petroleum Refining
Primary Metals
Transportation Equipment
Utilities
All Others
Sample
3
0
2
0
38
61
32
48
14
58
32
All Sources
1,659
293
117
187
592
1,091
1,979
637
215
3,786
1,096
288
11,652
5-25
-------
TABLE 5-14
DISTRIBUTION OF SOURCES BY SIC GROUP IN THE SAMPLE
AND FOR ALL PLANTS NEEDING CONTROL TO MEET THE 0.25 PPM
ONE-HOUR NO STANDARD
SIC
Code
13
20
28
29
33
37
49
TOTAL
Number of Sources Needing Control
Industry
Oil & Gas Extraction
Food Processing
Chemicals
Petroleum Refining
Primary Metals
Transportation Equipment
Utilities
All Others
Sample
32
16
4
31
72
1
82
9
All Sources
847
51
127
436
234
65
837
111
247
2735
5-26
-------
TABLE 5-15
RESULTS OF THE UNCERTAINTY ANALYSES
One-Hour NO-
0.10 ppm Standard
Initial Cost Annualized Cost
($ Million) ($ Million)
A A
X = $358 X = $141
X = 410 X = 155
S.D. = 64 S.D. = 22
95% C.I. = 283 - 536 95% C.I. = 113 - 197
0.25 ppm One-Hour NO,, Standard
A A
X = $170 X = $74
X = 220 X = 87
S.D. = 81 S.D. = 28
95% C.I. = 61 - 376 95% C.I. = 32 - 143
A
X = Single "point" estimate for the sample
X = Mean for all 50 simulations
S.D. = Standard deviation for 5,0 simulations
95% C.I. = 95% confidence interval for 50 simulations
5-27
-------
FIGURE 5-5
DISTRIBUTION OF INITIAL COSTS TO MEET A 0.10.PPM ONE-HOUR
N02 STANDARD FOR A SAMPLE OF PLANTS
NUMBER OF RUNS
14-
12 -H
10-
8-
6-
4-
2-
0 Single "Point" Estimate for the Sample
Mean Estimate for 50 Simulation Runs
150 175
250
I
275
I
300
I I
325 350
I
375
I " I
400 425
I
450
I
475
I I
500 525 550
COSTS (S MILLION)
5-28
-------
FIGURE 5-6
DISTRIBUTION OF ANNUALIZED COSTS TO MEET A 0.10 PPM ONE-HOUR
N02 STANDARD FOR A SAMPLE OF PLANTS
NUMBER OF RUNS
14
12-
10-
8-
6-
4-
2-
Single "Point" Estimate for the Sample
Mean Estimate for 50 Simulation Runs
70
80
I
90
1
100
110
I
120
I
130
140
I
150
I
160
I
170
I
180 190 200
COSTS (5 MILLION)
5-29
-------
FIGURE 5-7
DISTRIBUTION OF INITIAL COSTS
TO MEET A 0.25 PPM ONE-HOUR NOz STANDARD
FOR A SAMPLE OF PLANTS
NUMBER OF RUNS
14-
10 H
8-
6-
4-
2-
Single 'Point' Estimate for the Sample
Mean Estimate For 50 Simulation Runs
75
100
I
125
1
150
175
I
ZOO
Z25 Z50
275
300 325
l
350
375 406 425
($ MILLION)
-------
FIGURE 5-8
DISTRIBUTION OF ANNUALIZED COSTS
TO MEET A 0.25 PPM ONE-HOUR N02 STANDARD
FOR A SAMPLE OF PLANTS
NUMBER OF RUNS
14-
12-
10-
8-
6-
4-
2-
Single 'Point' Estimate for the Sample
Mean Estimate For 50 Simulation Runs
frr
20 30 40 50 60 70
90 100 110 120 130 140 150 160
(5 MILLIONS)
5-31
-------
complete population of sources needing additional control to meet the
0.10 ppm and 0.25 ppm NC- standards are shown in Tables 5-13 and 5-14.
In the main, the samples capture those industries that shoulder the
primary cost burden.
5.4.1.2 Results
The results of the simulation analyses are shown in Table 5-15 and
illustrated in Figures 5-5 through 5-8. In all cases the simple cost
estimates for the sample generated by using the best estimate for each
parameter (the single "point" estimate for the sample) is less than the
mean for all simulation runs. This reflects the fact that more parameter
distributions are skewed toward higher, rather than lower values. The
second important observation is that the cost variability for the 0.10
ppm standard is less than that for the 0.25 ppm standard, in both absolute
and relative (to the mean) terms. This reflects the heightened importance
of 0_ for the latter standard (NO rather than 0,, is controlling for the
j X j
0.10 ppm standard) and the presumed large variability around the selected
0- values.
In order to apply these results to the nationwide cost totals, they are
best expressed as percentages. First, a better estimate of the most
probable nationwide costs could be obtained by increasing the single
point estimates by the following percentages:
One-Hour Standard Initial Costs Annualized Costs
0.10 ppm 15% 10%
0.25 ppm 29% 18%
5-32
-------
Once this is accomplished, the 95 percent confidence intervals, in
percentage terms, are as follows:
One-Hour Standard Initial Costs Annualized Costs
0.10 ppm +31% +27%
0.25 ppm +72% +64%
These figures are taken directly from the uncertainty analysis and
assume perfect transferability to the nationwide analysis. They are
best used as a guide for judging variability in the estimated nationwide
costs rather than as definitive statements regarding total variability.
Moreover, some additional variability would be introduced by considering
terrain impacts, although the direction of the effect would be to increase
costs.
5.4.2 Plume Bouyancy Analysis
A limited study of plume bouyancy and the effect of SCL scrubbers acting
through reduced bouyancy on ground level NO concentrations also was
investigated. Two powerplants were used in the study because SO- scrubbing
is most likely to be employed in the utility sector. The source parameters,
the modeling conditions, and the specific PTMTP results at various
distances from each source are described in detail in Appendix F.
The results are summarized in Table 5-16. Based on an assumed reduction
flue gas temperatures of 159°C and 142°C for the two powerplants, the
ground level concentrations increase substantially. Considering only
stability A, ground level concentrations increase by 6 and 70 percent
for the two plants, respectively. Increases for the other stabilities
are relatively greater but, due to better dispersion conditions, the
absolute concentrations are lower.
5-33
-------
TABLE 5-16
EFFECTS OF SC>2 SCRUBBING ON GROUND LEVEL
CONCENTRATIONS FOR TWO POWERPLANTS
Plant
#1
With Scrubber (ppm)
Without Scrubber (ppm)
Percent Change
#2
With Scrubber (ppm)
Without Scrubber (ppm)
Percent Change
Maximum NO Concentrations
x
Stability
B
0.191
0.180
6
0.052
0.030
70
0.077
0.044
75
0.023
0.010
132
0.045
0.022
100
0.016
0.006
156
5-34
-------
Based solely on this limited analysis, one must conclude that the reduc-
tion in effective stack height caused by the use of flue gas scrubbers
for SO- could substantially increase ground level NO concentrations.
b X
However, the variability in the results of the two plants suggests that
the effect is highly plant-specific.
5.5 MOBILE AND AEEA SOURCE SENSITIVITY ANALYSIS
While an uncertainty analysis was performed for the point source modeling
results, a sensitivity analysis is more appropriate, where variation in
the modeling parameters is less well characterized. This is the case
for the mobile and area source analysis. A sensitivity analysis is
performed by arbitrarily varying values for a few of the key parameters
to test the sensitivity of the modeling results to those changes. These
variables included growth rate of new sources, the ratio of default peak
to mean N0_, and the assumed percentage reduction in N0_ due to volatile
organic compound (VOC) control. The effects of changes in these parameters
in the model results are discussed separately in the sections below.
For simplicity, all of the sensitivity runs were made for the 0.15 ppm
one-hour standard in 1990. The base case assumptions are a 6:1 peak-to-
mean ratio, a one percent growth rate, and no effect of VOC control on
N02 levels.
5.5.1 New Source Growth
Using the base case assumptions, the effect of assuming zero and 3
percent annual growth rates is compared for a 0.15 ppm one-hour N09
standard. In the base case, 32 counties are estimated not to attain the
0.15 ppm standard. With no growth, only 17 counties do not attain
while, with a 3 percent growth rate, 74 counties cannot attain this
standard. The estimated annualized costs for these alternatives are:
5-35
-------
Assumptions Annualized Costs (x 10 )
No Growth $ 713
1% Growth 1,502
3% Growth 3,318
5.5.2 Peak-to-Mean Ratio
In the one-hour standard analysis, for counties with valid annual averages
and no continuous data, a 6:1 ratio was used to estimate one-hour NO-
levels given the annual average. In order to test the sensitivity of
the model results to changes in this ratio, both 5:1 and 7:1 ratios were
used. In the base case, 32 counties are estimated not to attain the
0.15 ppm standard. With a 5:1 peak-to-mean ratio, 25 counties are
estimated not to attain while, with 7:1 ratio, 44 counties do not attain.
The estimated annualized costs for these alternatives are:
Ratio Annualized Costs (x 10 )
7:1 ratio $1,656
6:1 ratio 1,502
5:1 ratio 1,311
5.5.3 The Effect of VOC Control on NO,, Levels
Although the atmospheric chemistry of NO and VOC is complex and poorly
understood, recent empirical evidence suggests that VOC control yields
slight to moderate benefits in reducing yearly maximum one-hour N0?
concentrations (Trijonis, 1978). It is estimated that reducing VOC
emissions by 50 percent should reduce yearly peak NO- by about 10 to 20
percent. For this sensitivity analysis, it was assumed that VOC emis-
sions may be reduced by 50 percent from 1978 to 1990 and that this VOC
reduction will reduce N0? levels by 15 percent. The effects of a 15
percent reduction in NO- due to VOC controls on the base case are shown
below:
5-36
-------
Assumptions
No effect of VOC control
on NO- levels
15% reduction in NO- due
to VOC controls
Number of Annualized Cost
Nonattaining Counties
32
11
(x 106)
$1,502
590
When the results of the sensitivity analysis of this variable are compared
with the growth rate and peak-to-mean ratio changes, assuming a 15
percent reduction in NO- due to VOC controls produces the biggest change
in the final result (on the low side). Assuming a 3 percent growth rate
increases the 1990 annualized costs by a factor of two when compared
with the base case. Variations in the peak-to-mean ratio produce the
smallest variations in the modeling results, because most areas with
high NO- levels have continuous monitors and because no default value is
used. Table 5-17 shows the combined effect of the three variables
(growth, peak-to-mean ratio, and VOC control) on the 1990 estimated
annualized costs for mobile and area sources. The variation in costs is
dramtic, although the probability that the low and high parameter values
will occur simultaneously is undoubtedly less. Strictly speaking,
however, the probability is unknown.
5.5.4 The Effect Of FMVCP Waivers For Diesels
Section 202(b)(l)(B) of the Clean Air Act as amended establishes the
emission standards for light-duty vehicles for 1977 and later model
years. This section of the Act specifies that NO emissions for light-
A
duty vehicles may not exceed 1.0 gram per mile for 1981 and later model
years. However, Section 202(b)(6)(B) of the Act gives EPA the authority
to delay implementation of the 1.0 gram per mile (gpm) NO standard by
A
waiving it under certain conditions to permit the use of diesel engine
technology (45 FR 5480). The waiver can raise the FMVCP standard from
5-37
-------
TABLE 5-17
SENSITIVITY OF THE MOBILE AND AREA SOURCE MODELING RESULTS
ANNUALIZED COSTS (MILLIONS OF 1980 DOLLARS)
Peak: Mean Ratio
Growth Rate
Control Options Effect of VOC Control
Mobile and Area Sources
I&M Setup
Loaded Mode Test
EGR Inspection
TCM's
Stationary Area
Residential
Commercial
Industrial
TOTAL
5:1
0%
15% red.
$11.4
6.8
28.8
5.0
-4.8
26.5
169.4
$243.1
6:1
1%
None
$37.3
1.4
110.8
38.5
-53.8
215.6
1152.0
$1,501.8
7:1
3%
None
$65.6
6.6
196.9
85.8
-168.4
491.7
2921.4
$3,599.6
5-38
-------
1.0 up to 1.5 gpm for no more than four model years. The major con-
sideration in providing for light-duty diesel NO waivers was the poten-
tial fuel economy benefits of diesel technology. However, in applying
for waivers, manufacturers had to show that the technology has a potential
for long-term air quality benefit.
As of this writing, the EPA Administrator has issued three decisions
regarding light-duty diesel NO waivers. A summary of waiver applica-
X
tions granted is shown in Table 5-18. It should be noted that waivers
have only been granted for the 1981 and 1982 model years.
The guidance for waiver submissions established several NO -related air
X
quality effects to be assessed by the applicants. These were:
Annual regional NO- levels
Short-term regional NO- levels
Peak NO, exposure levels near roadways.
Modeling was based on the assumption that light-duty diesels would be
produced in 1981-1984 model years at a '.
assumed to be a "worst case" situation.
produced in 1981-1984 model years at a 1.5 gpm NO level. This is
GM estimated that a total change in regional NO and NO- concentrations
would be no more than 0.5 percent in late 1984, while Daimler-Benz and
VW predicted a change of less than 0.3 percent from granting the waiver.
The EPA analysis showed that future ambient levels of NO would continue
A:
to decrease even if NO waivers were granted to the 1.5 gpm level for
X
four years. Only if it is assumed that diesel market penetration would
be 25 and 50 percent in 1985 with mid-range and high NO deterioration
A
rates, respectively, would the progress in annual average NO- level
reductions be slowed by one to two percent.
5-39
-------
TABLE 5-18
LIGHT-DUTY DIESEL NO WAIVER APPLICATIONS GRANTED
x
Ln
i
Manufacturer
Daimler-Benz
General Motors Corp.
Volvo
Peugeot
VW
Engine Family
2.4 liter
3.0 liter naturally aspirated
3.0 liter turbocharged
5.7 liter
6 cylinder, 2.4 liter naturally aspirated
2.3 liter turbocharged
1.6 liter naturally aspirated,
2375 Ibs. inertia weight
1.6 liter naturally aspirated, 2500 I.W.
2.0 liter naturally aspirated, 3250 I.W.
1.6 liter turbocharged, 2375 I.W.
1.6 liter turbocharged, 2625 I.W.
2.0 liter turbocharged, 3250 I.W.
Model
Years
1981
1982
1981,1982
1981,1982
1981,1982
1981,1982
1981,1982
1981,1982
1981
1981,1982
1982
1982
1982
Interim Standard
(gpm)
1.5
1.25
1.5
1.5
1.5
1.5
1.5
1.3
1.4
1.5
1.3
1.4
1.5
SOURCE: 45 FR 5480, 45 FR 34719, 45 FR 65490.
-------
GM concluded that total short-term regional change in NO- concentrations
would not exceed 0.5 percent in late 1984, the date of estimated maximum
impact, even if all applicants were granted full four-year waivers.
EPA's analysis shows that short-term N0_ levels generally will decrease
through 1990 and that this trend would be slowed by no more than one
percent even under a high waiver diesel market penetration estimate of
25 to 50 percent.
The waiver applicants' analyses of peak one-hour maximum N02 exposures
from light-duty diesels granted the maximum waiver for model years
1981-1984 show that there would be insignificant increases over a "no
waiver" situation. EPA's comparison of waiver/no waiver cases indicates
a range in NO- levels from no change to an increase of less than 5
percent by 1982 if a two-year waiver is granted.
The air quality modeling studies performed by both the manufacturers and
EPA indicate that there would be a limited effect on ambient NO- levels
due to a full four-year diesel NO waiver. Therefore, the potential
A
effect on N09 levels resulting from a NO waiver for two model years
£* A
would not be significant.
5.6 CAVEATS
The results of the nationwide analysis described above are subject to a
variety of limitations but to two principal types. One is generated by
data shortcomings; the other derives from assumptions and simplifications
incorporated in the methodology.
NEDS and SAROADS comprise the primary data sets used in the analysis.
Despite extensive efforts to correct obvious errors and inconsistencies
and to replace missing values in NEDS, substantial inaccuracies may
still remain. Under-reporting of sources or inclusion of already-retired
plants are serious problems for certain States. These concerns place
5-41
-------
the results for particular industries and geographic regions in in
doubt. Unfortunately, the degree of data reliability among industries
and among regions is not well characterized. One the positive side, a
rough comparison of 1C engines and gas turbines used in the gas trans-
missions industry with an independent survey suggests that the popula-
tion of these sources in NEDS may be approximately correct.
The limitations imposed by SAROADS are probably not as serious. Although
measurement and reporting errors are incoporated in SAROADS, the annual,
24-hour, and one-hour values were screened by EPA before use in the
regulatory analysis, so that the NO- values are probably valid. The
more serious question concerns the way these values are used to depict
air quality.
With respect to limitations imposed by the methodology, the major limita-
tions can be characterized as follows:
The dispersion modeling analyses of point sources are subject
to all the conventional limitations of dispersion models.
In addition, the representativeness of the meteorological data
(from the nearest station) in the annual analysis remains to
be established.
The hourly analysis is even more uncertain the because meteoro-
logical conditions that lead to the second high daily maximum
hourly concentrations in flat terrain are assumed.
Yet another element of uncertainty is introduced in the 24-hour
analysis by the use of a constant 24-hour to one-hour ratio.
Interactions among plants in terms of ambient NO- contributions
are captured in a highly inexact manner through the use of an
NO- "background" value.
The mobile and area source analysis relies extensively on the
assumption of countywide homogeneity of air quality and of
emissions within each source category.
The use of a constant 6:1 peak-to-mean NO- ratio where continuous
monitors are not available is a simplifying but error-introducing
assumption.
5-42
-------
The application of generalized NO., control cost and effectiveness
curves is a highly approximate approach for individual sources.
The use of a least-cost criterion for selecting alternative
controls does not represent the existing nor perhaps the
future method of control strategy design in many regions.
Many of these sources of potential error were addressed in the point
source uncertainty analysis and the area source sensitivity analysis.
The results of these analyses, especially the uncertainty analysis, can
be used to judge the degree to which the nationwide results may differ
given different assumptions and different values for input parameters.
However, no attention has been given to the growth of point sources,
which, for the most stringent short-term standards, may further increase
the nationwide cost estimates. Nor has the effect of terrain, especially
on powerplants, been considered. Finally, the results of the point
source and mobile and area source analyses are probably not completely
additive because the interaction of point sources with the other source
categories has not been considered explicitly.
5-43
-------
REFERENCES FOR SECTION 5
Trijonis, I. 1978. Empirical Relationships Between Atmospheric Nitrogen
Dioxide and Its Precursors (EPA-600/3-78-018). Technology Service
Corporation, Santa Monica, CA. February. (Prepared for U.S. EPA,
Research Triangle Park, NC.)
5-44
-------
6. CHICAGO CASE STUDY
The nationwide point and area mobile source analyses are reasonable
attempts to assess the impact of alternative NAAQS's on thousands of
sources in hundreds of areas. However, the somewhat artificial separa-
tion of point from mobile and area sources is not totally suitable for
large metropolitan areas where, under appropriate conditions, all types
of sources could interact to produce high NO- levels. For point sources
with relatively short stacks, interaction with mobile and area sources
is likely under any of the generic episode categories while, for elevated
point sources, interaction is likely only during inversion break-up at
the end of a suppressed mixing scenario. In order to analyze these
situations more realistically, a multiple source dispersion modeling
study should be performed. Ideally, such a study would feature a photo-
chemical model in which chemical reactions involving NO- are explicitly
considered. If applied within the context of the regulatory analysis to
one or more metropolitan regions, the result should be a more accurate
estimate of control strategies and costs and of attainment status.
Comparing these to the results of the nationwide analyses for the same
geographical area should provide information on the degree of bias in
the nationwide methodologies.
A modified version of such a study was undertaken for the Chicago AQCR.
The modification lies with the dispersion model employed: due to the
considerable expense of using a photochemical model, both in terms of
developing the requisite VOC emission inventory and of exercising the
model, a dispersion model (RAM) appropriate for nonreactive pollutants
(e.g., NO ) was employed.* Because the use of the OLM is not appropriate
A
for urban regions or in general areas where sources interact extensively,
* NO can be considered "nonreactive" in this context because the major
reactions simply interconvet NO and N0_, leaving total NO unchanged.
* A
6-1
-------
N09 was estimated by assuming fixed N09-to-NO ratios. This introduces
&, L* X
considerable artificiality into the analysis, but is the only practical
approach under the circumstances. Moreover, the NCL-to-NO ratio is
^ A
varied to show how sensitive the results are to this assumption.
In addition to a comparison of the nationwide methodologies to the
source-interactive modeling approach, the Chicago case study affords the
opportunity to investigate an alternative to the use of least-cost
control strategies . Many States and local areas employ uniform regu-
lations, at least for all sources within the same category. Although
this approach is administratively convenient, it is not economically
efficient, as this analysis reveals.
6.1 SELECTION OF A CASE STUDY
Chicago is a logical choice for this analysis because:
It is one of the five AQCR's in the nation recording violations
of the annual average NCL standard over the last few years,
suggesting that hourly levels may be high as well.
The point and area source NO emission inventory for Chicago
is relatively complete as a result of several recent studies
conducted there, although accuracy of the inventory data
remains to be established.
Chicago, a classic urban- industrial area, presents a good
opportunity to study interactive effects of both point and
area sources. The N02 problem in Chicago can be assumed to be
representative of many urban AQCR's in the country.
6.2 AMBIENT N02 CONCENTRATIONS IN CHICAGO
Ambient N0~ data in the Chicago AQCR are inadequate, on their own for
characterizing and assessing the short-term N02 problem. There are only
five continuous N00/N0 monitors in the Chicago region, far below what
Z. X
would be required for an area of approximately 150 km by 80 km. However,
the five sites can be considered representative of some of the most
significant hot spots in the region: the peak one-hour N0~ levels
6-2
-------
normally observed at these sites are among the highest concentrations
observed in the country. Table 6-1 shows the highest and second highest
one-hour NO- levels measured annually at the five continuous monitors in
the region from 1975 through 1978.
Figures 6-1, 6-2, and 6-3 show the diurnal variation in the NCL and NO
^ A.
levels for the summer and winter seasons at the Plymouth Court (CAMP),
West Polk (Medical Center), and Joliet monitoring stations. These
patterns reflect the influence of mobile source NO emissions (morning
X
and evening rush hours) typical of urban areas. The N0_ levels at the
West Polk and Joliet stations show a less pronounced double peak that is
also typical of mobile sources influences, while the CAMP station reflects
an early afternoon bulge more characteristic of point source influences.
6.3 CHICAGO EMISSIONS INVENTORIES
The point and area/mobile source inventories were originally obtained
from the Radian Corporation.
6.3.1 Point Sources
The point source inventory is based on the Illinois EPA (IEPA) inventory
for the Illinois portion of the Chicago AQCR (circa 1975), while the
Indiana portion was derived from NEDS. Rather than providing estimates
of annual source operating characteristics, Radian estimated character-
istics for a typical summer morning, a summer afternoon, and a winter
afternoon, which reflect diurnal and seasonal variations in operating
loads. The summer morning inventory is selected for use in the case
study because emissions tend to be highest at this time. Actual emissions
are estimated using the emission factors described in Section 4.1.1.
Inspection of this inventory reveals that certain data needed to esti-
mate emissions, dispersion, and/or control costs are obviously erroneous
or missing from many records. Default values for these variables are
6-3
-------
TABLE 6-1
YEARLY HIGHEST AND SECOND HIGHEST ONE-HOUR N02 LEVELS
(ppm)
1975
1976
1977
1978
ON
Monitoring
Site Location
Camp
Medical Center
Joliet
La Salle
Northwestern
Second
Highest Highest
0.210
0.361
0.205
0.328
Second
Highest Highest
0.261 0.251
0.210 0.188
0.139 0.133
Second
Highest Highest
0.263
0.321
0.260
0.279
Second
Highest Highest
0.225 0.200
0.176 0.175
0.134 0.122
0.323 0.303
0.255 0.247
SOURCE: Illinois EPA.
-------
FIGURE 6-1
WINTER
12H
«P
IZM
SUMMER
NO,
8P
1ZM
PLYMOUTH COURT (CAMP STATION) UTM. (450. 4630)
02-2398-»
Seasonal-Diurnal Variation in N0x and N02 for
Plymouth Court Monitoring Station
SOURCE: Epright, et al., 1978.
6-5
-------
FIGURE 6-2
WINTER
ISM
SUMMER
12M
WEST POLK (MED CENTER)
UTM « (4.45, 4635)
Seasonal-Diurnal Variation in N0x and N02 for
West Polk Monitoring Station
SOURCE: Epright, et al., 1978.
6-6
-------
FIGURE 6-3
SEASONAL/DIURNAL VARIATION IN NOX AND N02 AT JOLIET STATION
CONCENTRATION (PPM)
0.3
0.2
0.1
WINTER (JANUARY 24,1977)
12M
6A
12N
i
6P
12M
CONCENTRATION (PPM)
0.3
0.2
0.1 -
SUMMER (JULY 8,1977)
12M 6A
SOURCE: EEA, Inc.
12N
6P
12M
6-7
-------
substituted based on what constitutes normal operating practice in a
particular process category. This defaulting procedure is a simplified
version of the one used in the nationwide point source analysis.
Table 6-2 shows the distribution of hourly emissions (in units of grams/
second) among 10 key source categories for the typical summer morning.
Utility boilers account for well over half of all NO emissions in
A.
Chicago, with industrial boilers and gas turbines in both the utility
and industrial sectors responsible for the bulk of the remainder. The
total lack of 1C engines is also noteworthy.
Total, region-wide annual emissions (tons/year) also are estimated
assuming constant emission rates for each hour and 6,000 hours of opera-
tion per year.* The annual rates are not used in the analysis, but can
be used to compare total regional emissions with those contained in
NEDS. The 1976 NEDS point source NO total for the 10-county AQCR is
X
approximately 490,000 tons per year (U.S. EPA, 1979). Given the differ-
ent sources of the two inventories and the approximate manner in which
the annual values were derived from the hourly estimates for the Chicago
(IEPA) inventory, the agreement is fairly good.
Table 6-2 also shows the distribution of emissions after small sources
(10 Ibs of NO /hour or less) are eliminated from the inventory. This
A
screening significantly reduces the number of sources but reduces emissions
by only one percent.
6.3.2 Area and Mobile Sources
Hourly NO emission estimates for mobile and stationary area sources in
X
the Chicago region were computed by Radian Corporation. Total vehicular
* Annual hours of operation are available for most sources but were not
used to compute annual emissions.
6-8
-------
TABLE 6-2
DISTRIBUTION OF NO SOURCES AND EMISSIONS AMONG
SOURCE CATEGORIES IN THE CHICAGO AQCR
All Sources
Source
Type
Utility Coal Boilers
Utility Oil & Gas
Boilers
Industrial Coal
Boilers
Industrial Oil & Gas
Boilers
Gas Turbines
Internal Combustion
Industrial Furnaces
Nitric Acid Plants
Incinerators
Commercial &
Institutional
Boilers
TOTAL
Sources Emissions (g/s)
21 8,086
11
70
797
820
645
15,538
(370,000 t/y)*
Sources >10 Ibs/hr
Sources Emissions (g/s)
21 8,086
7 817
59
472
637
504
11
0
111
6
4
59
2,372
2,505
0
968
31
1
112
264
11
0
81
6
0
23
2,254
2,505
0
948
31
0
95
15,373
(366,000 t/y)*
Assumes constant emission rates for 6,000 hours of operation per year at
every source.
6-9
-------
travel and speed data from the Chicago Area Transportation Study (CATS)
were used by Radian to apportion the 1975 annual average mobile source
emissions (taken from NEDS) among 541, 5 x 5 km grid cells in the Illinois
portion of the Chicago AQCR. Uniform distributions were assumed in the
Indiana counties (102 grid cells). Hourly emissions for typical summer
mornings (as well as summer and winter afternoons) were obtained by
further apportioning daily NO totals according to diurnal traffic
X
counts in each of 1,700 traffic zones (Illinois portions only).
The NEDS stationary area source emissions were apportioned among the 643
cells, using population and economic data for both portions of the
AQCR.* Diurnal and seasonal variations in the annual emissions were
based on natural gas demand data obtained from local gas companies.
Unfortunately, the subcategorization of these sources by residential,
commercial/institutional, and industrial class and by fuel type was not
maintained.
The summer morning hourly NO emission estimates for the region are
A
summarized below and compared with NEDS emissions:
Chicago NEDS
Sources g/s t/y t/y
Mobile 6,244 152,000 241,000
Stationary Area 2,819 65,000 70,000
* Details of the procedures used in apportioning the county-wide annual
emissions spatially and temporally can be found in the Radian report
(Eppright et al., 1978).
6-10
-------
The conversion from hourly to annual estimates employs the average of
county-specific conversion factors listed in the Radian report for all
counties. The stationary area source estimates match the NEDS totals,
but the mobile source estimates appear somewhat low given that they are
reported to have been derived from NEDS originally.
6.4 MODELING SPECIFICATIONS
As noted previously, RAM (Turner and Novak, 1978) was selected as an
appropriate model for estimating short-term NO concentrations in a
X
large metropolitan area. RAM is an approved EPA model which, given
simplifying assumptions and some modification to the solution algorithms,
can be run efficiently for an area of this size.
6.4.1 Meteorological Assumptions
Rather than estimating hourly concentrations for an entire year, this
analysis employs point estimates of hourly emissions for a summer morning,
as described above, together with assumed one-hour meteorological condi-
tions that could produce high ground level concentrations. For consis-
tency with the nationwide methodology, the conditions selected are
atmospheric stability class A, 2 m/s wind speed and stability class C, 2
m/s wind speed. In addition, RAM requires the specification of mixing
depth but, because mixing depth was not considered in the nationwide
analysis, fairly large values are used in the case study -- 2500 m for A
stability and 800 m for C stability. Since the first set of conditions
will create unfavorable dispersion for elevated sources, while the
second set is unfavorable toward sources that have short stacks or emit
at ground level, both sets are run for each receptor and the one producing
the highest NO concentration is selected for all sources affecting that
receptor. In other words, the modeling results represent the combina-
tion of the two sets of meteorologic conditions across all sites which
lead to the highest concentrations everywhere.
6-11
-------
Wind direction is the final meteorological variable of importance.
Based on the results of preliminary model runs, five wind directions are
selected (90°, 180°, 215°, 225°, 275°) and the one producing the highest
concentrations is selected for each receptor individually. These correspond
to the five predominant wind directions in the Chicago region.
Selecting specific combinations of meteorologic conditions, as opposed
to evaluating several thousand hours of observed conditions, greatly
reduces the computing requirements and, thus, the cost of the analysis.
However, the interpretation of the results becomes more problematic, as
is the case in the nationwide analysis. The levels of NO estimated are
A
certainly high, but their probability of occurrence is not known.
Instead, the probability must be assumed. In keeping with the nationwide
analysis, we assume that the combinations of meteorological conditions
used as input parameters will produce second high daily maximum hourly
concentrations of NO and, by extension, N0~.
6.4.2 Receptor Network
A total of 5,500 receptors was initially selected for the entire AQCR
and located in a regular hexagonal pattern throughout the region.
Seventy receptors corresponding to locations of the continuous and
24-hour NO- monitoring sites in the Chicago AQCR also were included.
This receptor network was used to identify the most significant point
and area sources and a reduced receptor network was respecified. Two
receptors downwind of each major point source and one for each major
area source were selected. The first receptor for each point source was
positioned at the estimated point of maximum ambient concentration; the
second receptor, in the same direction but twice as far away as the
first, was designed to capture the interaction of overlapping plumes.
The single receptor for each area source reflects the maximum impact
from each area source. The reduced network contains a total of 203
receptors.
6-12
-------
6.4.3 NO -to-N02 Conversion
Because RAM estimates total NO , the results must be translated into
A
N00. The OLM does not apply to situations where VOC and NO emissions
£ X
from numerous sources interact and photochemically produce N02. Thus,
the only alternative is to assume that N00 is a fixed fraction of NO at
z, x
each receptor. The following combinations are assumed for different
runs:
Point Sources Area/Mobile Sources
0.5 0.5
1.0 0.5
The ratio for area and mobile source contributions was not altered
because a review of NO_-to-NO ratios at the five Chicago continuous
^ A
monitors indicated that N0» was rarely greater than 50 percent of total
NO .
6.4.4 Calibration of RAM Results
Once estimated ambient concentrations are generated, they are typically
compared to observed concentrations at each monitoring site and a linear
calibration curve is estimated. However, the limited number of continuous
monitors precluded such a calibration in the case study. Instead, the
NO values were used without adjustment. On the other hand, the mon-
A
itored N0» values can be used in a general way to judge how reasonable
the modeled NO- values are.
6.4.5 Least-Cost Modeling
A least-cost algorithm similar to the one employed in the nationwide
analysis (see Section 4.1.4) was developed for this application. The
cost minimization principle is identical but the mechanics are, by
6-13
-------
necessity, somewhat different. In addition, a "de minimus" ambient NO
/
contribution of 5.3 x 10 ppm (1.0 |Jg/
cant sources from the costing analysis.
/ O
contribution of 5.3 x 10 ppm (1.0 |Jg/m ) was used to screen insignifi-
The files of available point and area/mobile source NO controls and the
A
cost of these controls are identical to the nationwide control file (see
Table 4-14). Fuel costs likewise are identical, as are the equations
for computing before- and after-tax costs.
The use of uniform Reasonably Available Control Technology (RACT) controls
in one scenario is a departure from the nationwide study. Table 6-3
presents the definitions of RACT controls for the major source categories.
These definitions are the joint product of EPA and EEA engineering
j udgment. *
The controls and costs for area sources are composite values for the
undifferentiated category "stationary area sources." They reflect the
relative NO emissions from the various subclasses and fuel types deli-
x J^
neated in NEDS. The distribution of NO emissions for these categories
A
comes from NEDS summary statistics for Chicago (U.S. EPA, 1979).
6.5 TREATMENT OF GROWTH
Although growth is not treated explicitly in the nationwide analysis, it
is an integral part of the Chicago case study. Components of the growth
analysis are as follows:
Production levels (and therefore uncontrolled NO emissions)
are changed at each plant based on two-digit SIC growth rates
estimated for Illinois by the Bureau of Economic Analysis
(BEA), U.S. Department of Commerce.
Additional industrial emissions due to growth are controlled
to Best Available Control Technology (BACT) levels, estimated
roughly at 50 percent for all industries
John Copeland (ESED, OAQPS) contributed to these definitions.
6-14
-------
TABLE 6-3
RACT DEFINITIONS FOR NO CONTROL
B.
C.
D.
E.
Source Type
Utility Boilers
Cyclones
Stokers
Pulverized Coal
Gas
Oil
Industrial Boilers
Stokers
Pulverized Coal
Gas
Oil
Residual
Distillate
Gas Turbines
1C Engines
Industrial Process Heaters &
Furnaces
Nitric Acid Plants
Definition
No RACT
LEA & OSC
Low NO Burner
A
LEA, OSC, & FGR
LEA, OSC, & FGR
LEA & OFA
Low NO Burner
x
FGR
LEA & OFA
FGR
Water Injection
A/F Adjustment
No RACT
Currently at RACT
LEA = Low Excess Air
OSC = Off-Stoichiometric Combustion
OFA = Over-fire Air
FGR = Flue Gas Recirculation
6-15
-------
Half of the controlled additional emissions is added to existing
sources (in-place growth) and half to model plants specified
for the major growth industries (greenfield growth)
Model plants accounting for the greenfield growth are located
in industrial parks with undeveloped acreage. These parks are
stratified by the types of industries they can accommodate.
Area and mobile sources grow at a rate of 1.0 percent per year
to reflect population increases. Mobile source emissions are
reduced to reflect FMVCP emission standards for new vehicles.
6.6 RESULTS
6.6.1 Estimated Air Quality Levels
Judged on the basis of a general comparison between estimated and observed
peak NO _ values, the RAM results appear high. The estimated values for
the two conversion scenarios are as follows:
NCL/NO Ratio 10 Highest Receptors
^ A
1.0 0.90-0.96 ppm
0.5 0.37-0.48 ppm
While the highest concentrations in Table 6-1 are between 0.3 and 0.4
ppm, the estimated concentrations are not totally unreasonable.
6.6.2 Estimated Costs and Attainment Status
Table 6-4 summarizes the cost and attainment results for the various
projection years, standards, and NCL conversion factors. Also shown for
comparison purposes are the results for the 10-county Chicago AQCR using
* Four of the 10 counties in the AQCR have no annual average NO- values
which can be used in the nationwide area source analysis. Thus, the
area source costs (but not the point source costs) estimated by the
nationwide methodology in Table 6-4 include only the six largest
.counties.
6-16
-------
ON
I
Year
1985
Point Sources
Area/Mobile Sources
1990
Point Sources
Area/Mobile Sources
TABLE 6-4
CONTROL COSTS AND ATTAINMENT SUMMARIES FOR CHICAGO
CASE STUDY
ONE-HOUR STANDARD AND NO CONVERSION ASSUMPTION
0.
50%
10 ppm
Conversion
Annualized
Costs
($1000)
"I
103J
10 3\
107J
Violations
10
10
0.25
50% Conversion
Annualized
Costs
($1000) Violations
-9\ 0
oj
-8\ 0
of
ppm
100% Conversion
Annualized
Costs
($1000) Violations
24\ 4
103J
54\ 4
108J
0.50
ppra
100% Conversion
Annualized
Costs
($1000)
-9\
OJ
-8\
oj
Violations
0
0
NATIONWIDE ANALYSIS
One-Hour Standard
1985
Point Sources
Area/Mobile Sources
1990
Point Sources
Area/Mobile Sources
0.10 ppm
Annualized Annualized
Costs Costs
($1000) Violations ($1000) Violations
386
96
\
/
386\
85/
14
14
151\
°J
0.50 ppm
Annualized
Costs
($1000) Violations
-------
the nationwide point and area source methodologies.* The case study
trends are clear: no costs are incurred to meet the 0.50 ppm standard,
the 0.10 ppm standard cannot be attained, and the implications of a 0.25
ppm standard depend entirely on the NO-to-NCL conversion assumption. In
terms of total costs, the 0.50 ppm standard is associated with cost
savings of over $9 million (that is to say; it was assumed on this study
that firms would adopt combustion modifications on boilers and furnaces
if it saves money, irrespective of air quality standards), while total
costs to approach the 0.10 ppm standard as closely as possible (assuming
50 percent NO conversion) are approximately $180 million to $210 million
per year. The latter cost reflects the least cost of attaining the
standard at all but 10 receptors and maximum controls on all sources
affecting those 10 receptors. Interestingly, the area source control
costs are greater than those on point sources for all standards, which
reflects the necessity of establishing an expensive regionwide mobile
source ISM program or a countywide area source control program whenever
point source controls are insufficient at even a single receptor in the
region (mobile source controls) or in the county (area source controls).
The impact of growth is also apparent, increasing costs by between 15
and 30 percent for the 0.10 and 0.25 ppm standards, respectively, despite
the countervailing effect of the FMVCP-
Comparing these results to those obtained with the nationwide method-
ologies reveals that the area and mobile source cost estimates are in
agreement,* but that the point source costs obtained with the nationwide
approach far exceed those estimated in the case study. This is true
despite the close agreement in attainment status (e.g., 10 violating
receptors versus 14 violating plants for a 0.10 ppm standard). It is
unlikely that the difference in dispersion modeling methodology would
explain these large discrepancies. Certainly the OLM can never produce
* The costs estimated by the nationwide methodology are slightly lower,
reflecting the exclusion of four out of ten countries in the analysis.
6-18
-------
estimates of NCL which exceed total NO at any receptor. Perhaps the
Z. X
use in the nationwide approach of an N0? "background" equal to the
highest annual average NO- in each county overstates somewhat the degree
of interaction among plants. However, even putting maximum control on
all point sources in the inventory produces costs that are far less than
the estimates generated for the NEDS inventory of the 10-county region
($211 million versus $416 million). This suggests that the source
inventories are considerably different in terms of numbers of sources
and production rates though apparently not in terms of total regionwide
emissions.
Table 6-5 disaggregates the results for the 0.10 ppm and 0.25 ppm stan-
dards by source type for each of the approaches. Marked differences are
seen both in the number of sources and the distributions of costs.
Coal-fired utility boilers predominate in the case study and a sur-
prisingly large cost also accrues to just two gas turbines. The nation-
wide approach, on the other hand, selects utility boilers of all fuel
types for the major cost burden. Industrial boilers and furnaces shoulder
an important secondary cost burden. Clearly, NEDS contains more sources
and a different distribution of sources among key categories than does
the IEPA inventory. In addition, the control cost per source of the
same type varies between the two analyses, as illustrated most dramati-
cally by the gas turbines. Although the IEPA inventory is reputed to be
of a more recent vintage, a detailed comparison of both inventories will
be required to more fully understand the cause of the discrepancies.
Despite these differences, the two analyses of the Chicago region are in
agreement on the need for NO control in the utility sector. This
A
suggests that with unstable atmospheric conditions (stability class A),
even tall stacks can emit plumes that have the potential to cause substan-
tial ground level concentrations of NO-. Because the bulk of regionwide
NO emissions are due to these sources, the majority of the control
A
costs accrue to them as well.
6-19
-------
TABLE 6-5
CONTROL COSTS FOR CHICAGO IN 1985 BY SOURCE TYPE
CASE STUDY
Source Type
Utility Boilers
Coal
Oil and Gas
Industrial Boilers
Coal
Oil and Gas
1C Engines
Gas Turbines
Industrial Furnaces
Nitric Acid Plants
Maximum
Annualized
Costs
($ Million)
$ 95
22
11
34
0
35
14
-------
In order to demonstrate the effect of uniform, source category-wide
technology standards on all sources in Chicago, RACT controls were
applied and modeled in the case study. The results indicate that RACT
costs approximately $44 million in annualized dollars and does not
contribute to attainment of the standards. Alone, RACT cannot achieve
the 0.10 ppm or 0.25 ppm (100 percent NO conversion) standards and
achieving the best least-cost solution (i.e., closest approach to the
standards) with RACT applied first raises the total costs by between $12
million and $42 million. Thus, RACT is largely superfluous. Likewise,
applying maximum control to all 430 point sources in the final inventory
raises the costs to about $211 per year, but cannot improve air quality
at any of the violating receptors.
6.7 CONCLUSIONS
The results of the Chicago case study raise questions about the adequacy
of the nationwide analysis. Specifically, discrepancies between the two
Chicago AQCR inventories suggest that either NEDS contains too many
sources and data of questionable accuracy, or that the IEPA inventory is
inadequate. Certainly, questions about the reliability of NEDS have
been raised previously by others. However, if this comparison is valid
and representative of the other regions and if NEDS is the less accurate
data source, then the control cost estimates for the nation as a whole
may be too high. On the other hand, both of the Chicago analyses suggest
that powerplants should bear the major cost burdens, though this is
accentuated in the case study analysis.
The case study also underscores the futility and inefficiency of apply-
ing uniform controls (RACT) to all sources in the region. RACT is
costly, ineffective in achieving ambient standards alone, and largely
superfluous when used in combination with a least-cost strategy.
6-21
-------
7. NEW SOURCE CONTROLS
The costs and impacts of meeting alternative ambient standards of NO-
are the primary focus of the regulatory impact analysis. However,
another perspective on the magnitude of these costs can be gained by
comparing them to the costs of meeting the emission standards for new
sources the New Source Performance Standards (NSPS's) for stationary
sources and the Federal Motor Vehicle Control Program (FMVCP). This
section will focus on these costs.
7.1 COST OF MEETING NEW SOURCE PERFORMANCE STANDARDS
Total costs of new source controls to meet the NSPS's for NO are estimated
X
for the following major categories of stationary sources:
Utility Boilers: Coal-fired
Utility Boilers: Oil- and gas-fired
Industrial Boilers: Coal-fired and oil- and gas-fired
Stationary Gas Turbines
Reciprocating 1C Engines
Nitric Acid Plants.
The capital costs are summed for the period beginning with the promulga-
tion of the appropriate NSPS (see Table 7-1) and ending in 1990. Annual
costs are for 1990 and include an annualized capital charge and operating
and maintenance (O&M) costs. All costs are in 1980 dollars.
7.1.1 Utility Boilers; Coal-fired
Based on the uncontrolled emission factor in AP-42 (U.S. EPA. 1978b),
new coal-fired utility boilers would have to institute control procedures
7-1
-------
TABLE 7-1
NEW SOURCE PERFORMANCE STANDARDS FOR NO,
Source Category
NSPS
Date of
Proposal
1. Utility Boilers:
Coal-fired
0.70 lb/10 Btu
0.60 lb/106 Btu
1971
1978
2. Utility Boilers:
Oil-fired
0.30 lb/106 Btu
1971
3. Utility Boilers:
Gas-fired
0.20 lb/106 Btu
1971
4. Nitric Acid Plants
5. Gas Turbines*
1,000-10,000 hp
>10,000 hp
3 Ib/ton of 100%
acid produced
150 ppm NO^
75 ppm NO
1971
1977
The proposed standards apply to all new, modified, and reconstructed
stationary gas turbines with greater than 1,000 hp heat input. Gas
turbines with a heat input at peak load from 1,000-10,000 hp are
exempt from the NO emission limit for five years from the date of
proposal.
7-2
-------
to meet the 0.7 lb/10 Btu NSPS proposed in 1971. However, the lag time
between NSPS proposal date and boiler start-up date has averaged about
five years for utilities, so that boilers constructed between 1971 and
1975-1976 were not affected by this regulation.* Furthermore, boiler
fire-box designs appearing after 1975 are reported to have reduced
uncontrolled NO levels to at least 0.7 lb/10 Btu, at no additional
A
cost* (U.S. EPA, 1978c). Thus, the cost of this NSPS is assumed to be
zero.
Similarly, recent studies have indicated that the current boiler design
employed by each of the four major manufacturers of coal-fired utility
boilers should be sufficient to comply with the proposed 0.6 lb/10 Btu
emission limits (U.S. EPA, 1978c). Where these designs prove to be less
than adequate, minor modifications to current designs should alleviate
the problem. Thus, coal-fired utility boilers should experience no
increase in capital or O&M costs as a result of the NSPS for NO .
7.1.2 Utility Boilers: Oil- and Gas-Fired
Based on the AP-42 emission factors for oil and gas utility boilers,
units installed since 1971 should have been able to meet NSPS levels
without additional control (U.S. EPA, 1978b). Where control has been
found necessary for some residual fuel oil boilers, combustion modifi-
cation (low excess air) at negligible capital and no annual expense has
undoubtedly been the preferred option (Evans and Castaldini, 1976).
7.1.3 Industrial Boilers: Coal-Fired and Oil- and Gas-Fired
There currently are no NSPS's for industrial boilers, although a study
of the cost and economic efforts of proposing NSPS's for industrial
* Personal communication with John Copeland, U.S. EPA, OAQPS-ESED,
Durham NC, February 1979.
7-3
-------
boilers is underway. Based on that study, the following NO limits
appear the most likely to be proposed:
Fuel Type NO Emission Limits
A
Natural Gas 0.2 Ib/MMBtu
Distillate Oil 0.2 Ib/MMBtu
Residual Oil 0.3 Ib/MMBtu
Coal
- stokers 0.6 Ib/MMBtu
- pulverized coal 0.7 Ib/MMBtu
These emission limits are assumed to apply to all industrial boilers
with heat input capacity greater than 50 10 Btu/hr and should require
some form of combustion modification for most boilers. Natural gas
firetube boilers and coal underfeed stokers may be exceptions, however.
Total 1990 industrial boiler fossil fuel consumption is predicted to be
6.72 x 10 Btu. Of this amount, new industrial boiler fossil fuel
demand is assumed to be 38 percent or 2.58 x 10 Btu.
Using the above assumption, and assuming that new boilers coming on-line
in 1982 and later are subject to the NO emission limits listed above,
A
NO control costs estimates for industrial boilers are as follows:
A
Evaluation Capital Cost Annualized Cost
Year (millions of $) (millions of $)
1985 6.9 26.0
1990 23.9 45.7
7.1.4 Stationary Gas Turbines
On September 10, 1979, EPA promulgated regulations designed to reduce
gas turbine emissions of nitrogen oxides and sulfur dioxides. The
regulations established performance standards for new, modified, and
7-4
-------
reconstructed gas turbines larger than 1,000 horsepower (hp). These
NSPS's require that the NO emission levels for gas turbines between
X
1,000 and 10,000 hp not exceed 150 ppm. This 150 ppm level applies to
gas turbines that are larger than 10,000 hp that are used in oil and gas
production or transport and that are located outside SMSA's. For all
other gas turbines larger than 10,000 hp, the NO concentration of
exhaust gases cannot exceed 75 ppm.
NSPS's for stationary gas turbines were originally proposed on October 3,
1977. Owners and operators who began construction after October 3,
1977, on gas turbines larger than 10,000 hp must comply with the regula-
tions, but exemptions are granted until October 1982 for gas turbines
between 1,000 and 10,000 hp.
The total nationwide cost of the NO NSPS emission standards for sta-
x
tionary gas turbines is calculated on the basis of 1) expected sales
from the time the proposed standard takes effect on different sized gas
turbines through 1990 and 2) the estimated costs for meeting the standard.
Sales and cost estimates are taken from the Standards Support and Environ-
mental Impact Statement (SSEIS) report (U.S. EPA, 1977), from which the
sales projections used here are extrapolated. Because the timing of the
NSPS differs by the size of the gas turbine, the average unit size was
determined for each application listed in the SSEIS. Based on this
information, it was assumed that stationary gas turbines used by electric
utilities have an average unit size of 80 MW (100,000 hp). Therefore,
all stationary gas turbines built after October 3, 1977 for use by
electric utilities will have to comply with the 75 ppm NO standard.
Because stationary gas turbines greater than 20,000 hp heat input are
produced on a custom order basis only, the lead time for delivery is
approximately two years. Therefore, it was assumed that costs of the
NSPS would not be incurred until 1979 for turbines of this size. This
is a much longer lead time than that required for the smaller mass-
produced models.
7-5
-------
For the other applications of stationary gas turbines listed in Table 7-2,
the average unit is assumed to be between 1,000 and 10,000 hp and is
exempt from the proposed NSPS for five years from the date of proposal.
Therefore, costs of the NSPS for 1,000 to 10,000 hp gas turbines are
assumed to be incurred starting in 1983.
The total capital and annualized costs estimated to be .incurred between
the NSPS proposal date and 1990 are shown in Table 7-2. Annualized
costs are a function of assumed hours of operation.
7.1.5 Reciprocating Internal Combustion Engines
The NSPS for stationary 1C engines proposed on July 23, 1979 was as
follows:
Fuel Type NSPS (ppm NO )_
X
Gas 700
Diesel 600
Dual Fuel 600
Emission limits take effect in January 1982, 30 months after the proposal
date. The:
1982-1990.
date. Therefore, costs of NO control for 1C engines are summed from
Based on comparisons of uncontrolled emisson rates for new 1C engines
with the estimated standards, the percentage reductions necessary to
meet the NSPS's are estimated as follows:
Percentage Emission Reduction Needed
Fuel Type to Meet the Estimated NSPS's
Gas 60
Diesel 45
Dual Fuel 25
7-6
-------
TABLE 7-2
STATIONARY GAS TURBINES - TOTAL CAPITAL AND ANNUAL1ZED COST IN 1990
(1980 Dollars)
1.
2.
3.
4.
Applications
Utilities
Oil and Gas
Industry
Private Industry
Electric Power
Generation
a. oil and gas
b. other industry
Other Industry
Year Costs
Of NSPS Incremental Sales
Begin to
Be Incurred High (Med) Low
1979 47,830 29,300
1983 7,970 1,880
1983 600 55
1983 13.80
1983 8.35
Capital
Costs
($/kWh)
$3.51
6.52
6.51
6.51
6.51
Total Capital Cost
Through 1990 (10 $)
High (Med) Low
$167.9 $102.8
52.0 12.3
4.0 0.4
9.0
5.4
$238.3 $129.9
Assumed Ave.
Annual Usage
(hr/year)
500
8000
8000
2000
2000
Total Annual-Cost
Annual! zed in 1990 (10 $)
Cost per kWh
(mills/kWh) High (Med) Low
2.1 $49.1 $30.2
0.9 52.1 12.2
1.6 8.5 0.6
1.4 3.8
1.4 2.3
$115.8 $49.1
SOURCE: Derived from data in U.S. EPA, 1977.
-------
The costs for new 1C engine controls, as they appear in the SSEIS (U.S.
EPA, 1979b) are shown below:
Annualized Costs of Alternative Standards,.in the
Fifth Year after Implementation (10 $)
Alternative NO Reductions
Application 20% 40%X 60%
Gas Production and $6.9 $16.0 $20.9
Transmission
Electric Generation 9.3 10.5 23.4
Other Applications 2.6 4.0 8.3
All Applications $18.8 $30.5 $52.7
Because the cost information for alternative NO reductions in the SSEIS
x
is broken down by application rather than by fuel, a conservative assump-
tion is made that a 60 percent reduction in NO is needed for all applica-
A
tions. Based on this assumption, and on the assumption that the constant
annual 1C engine sales estimated in the SSEIS will continue through
1990, the following costs are estimated for the 1982-1990 period:
Capital cost = $19 million
Annualized cost in 1990 = $84 million.
The capital costs for the 1982-1990 period are estimated by multiplying
the $2 million per year capital cost (inflated to 1980 dollars) given in
the SSEIS by eight. The annualized cost in 1990 is estimated by multi-
plying the $52.7 million fifth-year cost associated with a 60 percent
NO reduction by 1.6.
A
7-8
-------
7.1.6 Nitric Acid Plants
Capital and operating costs for control of NO in new nitric acid plants
X
are taken from the NO Control Techniques Document (CTD) (U.S. EPA,
X *
1978a). These costs (representative of a 270 Mg/day plant) are used
along with growth estimates provided by MITRE Corporation to calculate
the total cost of attaining an NSPS of 3 Ibs of NO /ton of 100 percent
X
acid produced. MITRE estimates that an average of four nitric acid
plants of 420 Mg/day capacity will be built each year.* Actual growth
in nitric acid plants from 1971 to 1978 has been somewhat less than four
plants per year; however, the four new plants per year assumption is
used in this analysis.
Costs for six different NO abatement systems for nitric acid plants are
shown in Table 6-4 of the CTD. In order to determine a high and a low
cost estimate for new source controls, costs were developed for both a
molecular sieve control process and a chilled absorption process.
Chilled absorption is used primarily for retrofit of existing plants,
however, and cost estimates for this control technique may be unreal-
istically low. The catalytic reduction process has higher costs than
the molecular sieve process and was used to establish the NSPS for
nitric acid plants originally. However, since that time, fuel costs
have risen to the point where catalytic abatement is not economically
attractive for new nitric acid plants. Therefore, the molecular sieve
process is analyzed here to generate a high cost estimate. The esti-
mates of total cost of all nitric acid plants built from 1971 to 1990
which attained an NSPS of 3 Ibs of NO /ton are as follows:
* Personal communication with Linda Duncan, MITRE Corporation, McLean,
Virginia, December 1978.
7-9
-------
Control , Annualized,.
Technique Capital Cost (10 $) Total Cost (10 $)
Molecular Sieve 14.0 44.1
Chilled Absorption 6.0 17.2
(CDL/Vitok Process)
7.1.7 Summary of NSPS Costs
Table 7-3 summarizes the cumulative capital and annualized costs for
stationary source NO control to meet the NSPS levels. The ranges in
A
control costs indicate the uncertainty in both the number of new sources
that will be constructed between now and 1990 and the costs of control
techniques.
7.2 FEDERAL MOTOR VEHICLE CONTROL PROGRAM COSTS
The costs of the Federal Motor Vehicle Control Program (FMVCP) are
estimated separately for each of the following vehicle categories:
Light-duty vehicles
Light-duty trucks
Heavy-duty vehicles.
The costs of NO control for motorcycles are negligible and are not
X
included here. Although EPA-OMSAPC has prepared an analysis of the
cost-effectiveness of alternative aircraft emission control strategies
for use in promulgating emission standards, it now appears that there is
little likelihood of NO controls for existing aircraft engines. Apparently,
X
the technology is such that controls must be built into new engines.
Therefore, it will be assumed in this analysis that no NO controls will
A.
appear on aircraft engines until after 1990.
7-10
-------
TABLE 7-3
CUMULATIVE CAPITAL AND ANNUALIZED COSTS FOR STATIONARY SOURCE
NO CONTROL TO MEET THE NSPS LEVELS IN 1990
x
(millions of 1980 dollars)
Source Categories
Utility Boilers
Industrial Boilers
Stationary Gas Turbines
Reciprocating 1C Engines
Nitric Acid Plants
TOTAL
Capital Cost
$ 0
23.9
129.9-238.3
19.0
6.0-14.0
$178.8-295.2
Annualized
Cost
$ 0
45.7
49.1-115.8
84.0
17.2-44.1
$196.0-289.6
7-11
-------
7.2.1 Light-Duty Vehicles
7.2.1.1 Initial Cost of Emission Control Systems
The schedule of NO emission standards for light-duty vehicles, the
applicable control devices and their costs attributable to NO are as
follows:
Emission
Standard Starting in Control
(gpm) Model Year Device Cost
3.0 1973 EGR $ 9.10
2.0 1977 EGR $ 9.10
1.0 1981 3-way catalyst $101.00
As indicated, the technology used to meet the 2.0 and 3.0 grams/ mile
standards for NO was exhaust gas recirculation (EGR). The cost of an
X
EGR valve ($8.42 in 1979 dollars or $9.10 in 1980 dollars) is as estimated
by Lindgren (1977). Because this control technology is specific to NO ,
A
the entire cost of EGR can be attributed directly to NO .
More sophisticated technology is needed in order to meet the 1.0 gram/
mile NO emission standard mandated to start with the 1981 model year.
A
This technology is a three-way plus oxidation catalyst system. Because
the three-way catalyst needed to meet the 1.0 gram/mile NO standard is
A
also needed to meet the 0.41 gram/mile HC and 3.4 gram/mile CO stan-
dards, the initial cost of installing this three-way catalyst must be
apportioned between the three pollutants.
The cost of components in a three-way plus oxidation catalyst system are
shown in Table 7-4. Where applicable, the costs in this table are noted
as being for HC, CO, or NO control only. The cost of this system
A
attributable to NO control is estimated by adding one-third of the
A
costs of the components that control all three pollutants to the total
7-12
-------
TABLE 7-4
COST OF COMPONENTS IN A THREE-WAY PLUS OXIDATION CATALYST SYSTEM
Components
Throttle Position Sensor
PCV Valve
HEI (less breaker point distributor)
TVS (spark)
Electric Choke
EFE
EGR (back pressure)
TVS (EGR)
Stainless Steel Exhaust Pipe (less steel pipe)
Air Injection System
Air Switching System
Feedback Carburetor (less open loop carburetor)
Three-way Plus Oxidation Catalyst
ECU
Qy Sensor
H90 Temperature Sensor
Inlent Air Temperature Sensor
Engine Speed Sensor
Crank Angle Position Sensor
EGR Pintle Position Sensor
Evaporative System
TOTAL
Cost in
Minimum
$ 1
7
1
4
7
9
30
2
8
157
30
3
10
1978 Dollars
Maximum
$ 2
1
7
2
1
4
7
2
9
30
2
8
157
30
3
2
2
2
2
2
10
Pollutants
Controlled
All
HC
All
. All
HC, CO
All
NO
X
NO
X
All
HC, CO
HC, CO
All
All
All
All
All
All
All
All
All
HC
$269
$285
-------
cost of the components that control only NO . In 1978 dollars, the cost
X
attributable to NO is $86; inflated to 1980 dollars, it is $101.
The capital costs of light-duty vehicle (LDV) NO control in 1985 and
X
1990 are estimated by multiplying the projected or actual new LDV sales
for each model year back to 1973 by the increased cost per LDV due to
NO control. These products are then summed to determine the total
X
capital cost of NO control. The cumulative capital cost is estimated
to be $3.8 billion by 1985 and $7.1 billion in 1990. The calculation
procedure used for estimating the 1990 capital costs is shown in Table
7-5. The 1985 capital costs were estimated by summing the yearly capital
costs in Table 7-5 from 1973 to 1985. The annualized capital costs for
1985 and 1990 were calculated by multiplying the capital cost attributed
to each model year by the scrappage rate (determined by vehicle age) and
a capital recovery factor of 16.28 percent. This capital recovery
factor assumes a 10 percent discount rate and a 10-year time horizon.
The 1985 annualized capital cost for light-duty vehicle NO control is
estimated to be $559 million, while the 1990 annual cost is estimated to
be $916 million. The calculation of the costs is shown in Tables 7-6
and 7-7, respectively, for the two years.
7.2.1.2 Maintenance Costs
The assumed yearly operating and maintenance savings or costs are taken
directly from U.S. EPA (1979a). For pre-1981 LDV's, it is assumed that
there is no change in O&M costs attributable to NO control. In other
A
words, it is assumed that the EGR valve does not need replacement. For
the LDV's with three-way catalysts, it is assumed that a $2.94 per year
maintenance cost will be accumulated by each vehicle on the road. The
specific costs and savings included in this estimate are shown in Table
7-8. The annual O&M costs due to NO controls for 1985 and 1990 are
x
$165 million and $302.8 million, respectively. The O&M cost calculation
for 1990 is shown in Table 7-9. Costs for 1985 were calculated in the
same manner.
7-14
-------
TABLE 7-5
THE CAPITAL COST INCREASES FOR LIGHT-DUTY VEHICLES
DUE TO NO CONTROLS IN 1990
(COSTS IN 1980 DOLLARS)
Model
Year
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
TOTAL
New Car,
Sales (x 10°)*
12.5
12.6
12.5
12.4
12.3
12.3
12.0
11.5
11.0
11.1
11.5
10.9
11.3
11.2
9.9
8.2
8.8
11.5
Initial Cost Increase
Per Car
$101.00
101.00
101.00
101.00
101.00
101.00
101.00
101.00
101.00
101.00
9.10
9.10
9.10
9.10
9.10
9.10
9.10
9.10
Total Capital Cost
(x 108)
$12.63
12.73
12.63
12.52
12.42
12.42
12.12
11.62
11.11
11.21
1.05
0.99
1.03
1.02
0.90
0.75
0.80
1.04
$128.99
* SOURCE: McNutt et al., 1979. Table A-l.
7-15
-------
TABLE 7-6
ANNUALIZED CAPITAL CHARGES FOR NO CONTROL
IN 1985 FOR LIGHT-DUTY VEHICLES
(1980 Dollars)
Model
Year
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
TOTAL
Total Capital Cost
(x 108)
(From Table 7-5)
12.42
12.12
11.62
11.11
11.21
1.05
0.99
1.03
1.02
0.90
0.75
0.80
1.04
Scrappage
Rate*
1.000
0.992
0.968
0.951
0.925
0.884
0.824
0.750
0.656
0.550
0.447
0.356
0.279
Capital
Recovery
Factor**
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
Annualized
Capital o
Charge (x 10b)
$2.02
1.96
1.83
1.72
1.69
0.15
0.13
0.13
0.11
0.08
0.05
0.05
0.05
$9.97
* SOURCE: EEA, Inc., 1980. Table 3-3.
** Assumes a 10 percent discount rate and a 10-year time horizon.
7-16
-------
TABLE 7-7
ANNUALIZED CAPITAL CHARGES FOR NO CONTROL
IN 1990 FOR LIGHT-DUTY VEHICLES
(1980 Dollars)
Model
Year
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
TOTAL
Total Capital Cost
(x 108)
(From Table 7-5)
$12.63
12.73
12.63
12.52
12.42
12.42
12.12
11.62
11 ..11
11.21
1.05
0.99
1.03
1.02
0.90
0.75
0.80
1.04
Scrappage
Rate*
1.000
0.992
0.968
0.951
0.925
0.884
0.824
0.750
0.656
0.550
0.447
0.356
0.279
0.219
0.170
0.119
0.083
0.058
Capital
Recovery
Factor**
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
Annualized
Capital
Charge (x 108)
$2.06
2.06
1.99
1.94
1.87
1.79
1.63
1.42
1.19
1.00
0.08
0.06
0.05
0.04
0.02
0.01
0.01
0.01
$17.23
* SOURCE: EEA, Inc., 1980 (Table 3-3).
** Assumes a 10 percent discount rate and a 10 year time horizon.
7-17
-------
TABLE 7-8
ESTIMATED PER VEHICLE MAINTENANCE COSTS OVER 100,000 MILES FOR
LIGHT-DUTY VEHICLES WITH THREE-WAY CATALYSTS
(1980 dollars)
Total Cost of Cost Related to
Maintenance Maintenance NO Control
, x
SOURCE: U.S. EPA, 1979.
Change Oxygen Sensor Three Times 3 x $18 = $54 $18
Misc. Emission System Repairs $59 $20
Save One Muffler Change -$24 -$8
$30 per 10 years
7-18
-------
TABLE 7-9
ANNUAL OPERATING AND MAINTENANCE COSTS DUE TO THE
FMVCP IN 1990 FOR LIGHT-DUTY VEHICLES
(1980 Dollars)
Model
Year
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
Annual O&M
Costs Per
Vehicle*
$2.94
2.94
2.94
2.94
2.94
2.94
2.94
2.94
2.94
2.94
0
0
0
0
0
0
0
0
New LDV,.Sales
(x 10°)**
12.50
12.60
12.50
12.40
12.30
12.30
12.00
11.50
11.00
11.10
11.50
10.90
11.30
11.20
9.85
8.20
8.79
11.48
Scrappage
Rate
1.000
0.992
0.968
0.951
0.925
0.884
0.824
0.750
0.656
0.550
0.447
0.356
0.279
0.219
0.170
0.119
0.083
0.058
Total Annual
Savings (xlO )
$36.8
36.7
35.6
34.7
33.4
32.0
29.1
25.4
21.2
17.9
0
0
0
0
0
0
0
0
TOTAL
* SOURCE: U.S. EPA, 1979.
** SOURCE: McNutt et al., 1979. Table A-l.
+ SOURCE: EEA, Inc., 1980. Table 3-3.
$302.8
7-19
-------
7.2.1.3 Differences in Fuel Economy
Fuel economy differences due to NO controls on LDV's are estimated
using 1978 weight-mix fuel economy data (Murrell et al., 1980). The
fuel economy penalty associated with an EGR system is estimated by
comparing 1978 weight-mix fuel economy values for the combined city/
highway driving cycle for pre-control LDV's with 1973 and 1974 model^
year values. These values show an average fuel economy penalty due to
EGR of 1.0 percent. This penalty is assumed to apply to the 1973 through
1980 model years.
Murrell et al. (1980) shows the composite fuel economy of 1978 California
vehicles (meeting 0.41/9.0/1.5 g/m standards for HC, CO, and NO ) to be
about 7 or 8 percent better than that of uncontrolled vehicles at a
constant weight mix. This is the fuel economy credit assumed for post-
1974 vehicles in both the ozone (U.S. EPA, 1979a) and carbon monoxide
regulatory analyses (EEA, 1979). For NO control, this fuel economy
A
credit is not assumed to apply until the 1981 model year.
Athough the 7 percent fuel economy benefit is consistent with previous
regulatory work, more recent evidence shows a significantly different
fuel economy benefit for three-way catalyst systems. If the data in the
Murrell et al. (1980) paper are used with the same methodology employed
in the EGR analysis noted above to determine the fuel economy impact of
a three-way catalyst system (i.e., comparing a 1978 weight-mix, pre-control
fuel economy with the 1980 fuel economy for California for LDV's meeting
0.41/9.0/1.0 O&M standards), a 26.5 percent increase is estimated. The
problem with assuming that this number represents the fuel economy
benefit of a three-way catalyst system is that many other changes to LDV
engines were made over this period to maximize fuel economy that cannot
be credited to the three-way catalyst. Because it is not possible to
isolate the fuel economy benefit solely due to NO controls, the 7
X
percent fuel economy credit is assumed. One-third of this (2.3 percent)
is then attributed to the NO standard.
x
7-20
-------
Using this assumption, Table 7-10 shows the annual fuel economy differ-
ences due to the FMVCP in 1990 for IDV's. In this table, annual vehicle
miles of travel (VMT) per LDV are divided by the average fuel consump-
tion to estimate annual fuel usage per LDV. This fuel usage estimate is
then multiplied by estimated LDV sales, scrappage rate, fuel economy benefit
(loss), and the. 1980 price of gasoline ($1.30 per gallon) to estimate
annual savings (cost) in 1990. The estimated fuel economy benefit of
FMVCP controls for NO in 1990 is $996.1 million.
x
7.2.1.4 Costs of Unleaded Fuel
Light-duty vehicles equipped with oxidation catalysts or three-way cata-
lysts must use unleaded fuel, so that the price differential between
leaded and unleaded fuel must be considered for model years when the
cost of these controls can be attributed to NO . Therefore, the added
x
cost of unleaded fuel for NO controls is estimated for 1981-1990 model
x
years. Table 7-11 illustrates the cost calculation for 1990 assuming a
$0.04/gal penalty for unleaded fuel. The annual cost to LDV's due to
the unleaded fuel price differential is estimated to be $1.9 billion in
1990. However, only one-third of this cost can be attributed to NO .
X
The 1985 annual cost is estimated to be $980 million (for all pollutants).
7.2.2 Light-Duty Trucks
7.2.2.1 Initial Cost of Emission Control Systems
Initial costs of emission control systems for light-duty trucks (LDT's)
will be assumed to be the same as for LDV's. Because the schedule of
NO standards for LDT's is somewhat different, the total capital cost
A
attributable to NO control in 1985 and 1990 will be somewhat lower than
x
that for LDV's. The assumed standard schedule for LDT's will be:
Starting in LDT1 LDT2
Model Year <6,000 Ibs GVWR 6,001-8,500 Ibs 6W?R
1973 3.1 g/m
1979 2.3 g/m 2.3 g/m
7-21
-------
TABLE 7-10
ANNUAL FUEL ECONOMY DIFFERENCES DUE TO THE FMVCP IN 1990 FOR LDV'S
to
N5
Model
Year
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
TOTAL
Annual VMT
Per LDV*
14,436
13,903
13,371
12,838
12,306
11,773
11,240
10,708
10,176
9,643
9,110
8,577
8,045
7,513
6,980
6,447
5,927
5,382
Fuel
Consumption
(mpg)**
27.5
27.5
27.5
27.5
27.5
27.5
26.0
24.5
23.0
21.5
20.0
19.0
18.0
17.0
16.0
15.0
13.0
13.0
Gallons of
Fuel Used
525
506
486
467
447
428
432
437
442
449
456
451
423
442
436
430
456
414
New LDV,
Sales x 10 **
Scrappage
Rate
* SOURCE: EEA, Inc., 1980. Table 2-1
** SOURCE: McNutt et al., 1979.
+ The price of gasoline is assumed to be $1.30 per gallon (1980 dollars).
Fuel Economy
Benefit (Loss)
12.50
12.60
12.50
12.40
12.30
12.30
12.00
11.50
11.00
11.10
11.50
10.90
11.30
11.20
9.85
8.20
8.79
11.48
1.000
0.992
0.968
0.951
0.925
0.884
0.824
0.750
0.656
0.550
0.447
0.356
0.279
0.219
0.170
0.119
0.083
0.058
0.023
0.023
0.023
0.023
0.023
0.023
0.023
0.023
0.023
0.023
-.010
-.010
-.010
-.010
-.010
-.010
-.010
-.010
Annual Savings
in 1990 (xlO )
$150.9
145.5
135.3
126.7
117.0
107.0
98.2
86.7
73.4
63.0
-30.5
-22.8
-17.3
-14.1
-9.5
-5.5
-4.3
-3.6
$996.1
-------
TABLE 7-11
INCREASE IN THE ANNUAL COST OF LDV'S DUE TO THE ADDED
COST OF UNLEADED FUEL IN 1990
Gallons of
OJ
Model
Year
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
Fuel Used
(From Table 7-10)
525
506
486
467
447
428
432
437
442
449
New LDV 6
Sales x 10 *
12.5
12.6
12.5
12.4
12.3
12.3
12.0
11.5
11.0
11.1
Scrappage
Rate**
1.000
0.992
0.968
0.951
0.925
0.884
0.824
0.750
0.656
0.550
4 Cents/Gallon
Penalty
$0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
Annual Cost
x 10B
$ 262.5
253.0
235.2
220.3
203.4
186.2
170.9
150.8
237.6
109.6
TOTAL
$1,919.5
* SOURCE: McNutt, et al., 1979.
** SOURCE: EEA, Inc., 1980.
-------
Prior to 1975, LDT's up to 6,000 Ibs GVWR were classified as LDV's.
Therefore, the LDV emission standards for NO apply to this class of
A
LDT's for the 1973 and 1974 model years. As shown above, until the 1979
model year when the Federal LDT class was expanded to include vehicles
up to 8,500 Ibs GVWR, there was no NO emission standard for trucks
X
between 6,001 and 8,500 Ibs GVWR. Therefore, in order to accurately
estimate costs of NO control for LDT's, LDT1 and LDT2 controls must be
x
considered separately. Although no specific level has been set for the
post-1984 NO emission standard for LDT's, the Clean Air Act requires a
A
75 percent reduction in NO (from pre-controlled) in 1985 for vehicles
X
over 6,000 Ibs GVWR. It is assumed here that three-way catalyst systems
will be needed for LDT's to meet this requirement.
The expected initial cost to the consumer for NO control for 1985-1990
X
model years is $154 (U.S. EPA, 1981), which includes $3.19 for develop-
ment and testing, $1.38 for certification, $0.84 for in-use durability
testing, $1.46 for emission control hardware, and $2.26 for allowable
maintenance provisions (gasoline-powered LDT's only).
The capital costs of LDT NO control in 1985 and 1990 are estimated by
multiplying the projected or actual new LDT sales for each model year
back to 1973 by the increased cost per LDV due to NO control. New LDT
sales information is taken from McNutt et al. (1979). It should be
noted that these numbers are somewhat higher than those used by EPA in
its LDT regulatory analysis (U.S. EPA, 1980). The sales data from
McNutt were used here primarily because the EPA analysis only presents
sales data from 1973-1978 and 1983-1988 model years and data for 1973-
1990 model years were needed as well.
The cumulative capital cost for LDT NO control is estimated to be $990
X
million by 1985 and $4.6 billion in 1990. The calculation procedure
used for estimating the 1990 capital costs is shown in Tables 7-12 and
7-13. The 1985 capital costs were estimated by summing the yearly
7-24
-------
TABLE 7-12
CAPITAL COST INCREASES FOR LIGHT-DUTY TRUCKS (0-6,000 Ibs)
NO CONTROLS II
x
(1980 Dollars)
DUE TO NO CONTROLS IN 1990
A
Model
Year
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
New LDTl.-Sales
(x 10°)*
2.54
2.52
2.48
2.44
2.40
2.37
2.28
2.13
1.99
2.02
2.04
1.87
1.94
1.34
1.47
1.21
1.62
1.85
Initial Cost Increase
Per LDT1
$154.0
154.0
154.0
154.0
154.0
154.0
9.1
9.1
9.1
9.1
9.1
9.1
9.1
9.1
9.1
9.1
9.1
9.1
Total Capital Cost
(x 106)
$391.2
388.1
381.9
375.8
369.6
365.0
20.7
19.4
18.1
18.4
18.6
17.0
17.7
12.2
13.4
11.0
14.7
16.8
TOTAL
$2,469.6
SOURCE: McNutt et al., 1979. Table A-l.
7-25
-------
TABLE 7-13
CAPITAL COST INCREASES FOR LIGHT-DUTY TRUCKS (6,001-8,500 Ibs)
(1980 Dollars)
Model
Year
1990
1989
1988
1987
1986
1985
198A
1983
1982
1981
1980
1979
TOTAL
New LDT2,Sales
(x 106)
2.26
2.24
2.21
2.17
2.14
2.13
2.03
1.90
1.76
1.81
1.84
1.66
Initial Cost Increase
Per LDT2
$154.0
154.0
154.0
154.0
154.0
154.0
9.1
9.1
9.1
9.1
9.1
9.1
Total Capital Cost
(x 10b)
$348.0
345.0
340.3
334.2
329.6
328.0
18.5
17.3
16.0
16.5
16.7
15.1
$2,125.2
* SOURCE: McNuttetal., 1979. Table A-1.
7-26
-------
capital costs in Tables 7-12 and 7-13 from 1973-1985. The annualized
capital costs for 1985 and 1990 were calculated by multiplying the
capital cost attributed to each model year by the scrappage rate (deter-
mined by vehicle age) and a capital recovery factor of 16.28 percent.
This capital recovery factor assumes a 10 percent discount rate and a
10-year time horizon. The 1985 annualized capital cost for LDT NO
X
control is estimated to be $150 million, while the 1990 annual cost is
estimated to be $656 million. The calculation of the costs for 1990 is
shown in Table 7-14.
7.2.2.2 Differences in Fuel Economy
Fuel economy differences due to NO controls on LDT's are estimated
using the same assumptions made for LDV's for pre-1985 model years.
Therefore, an average fuel economy penalty due to EGR of 1.0 percent is
assumed. In addition, it is assumed that three-way catalysts have no
effect on fuel economy. Tables 7-15 and 7-16 show the calculations of
increased annual costs due to fuel economy effects of NO controls on
A
LDT's in 1990. Estimated annual costs in 1990 due to fuel economy
differences are $89 million.
7.2.3 Heavy-Duty Vehicles
The Clean Air Act calls for a heavy-duty vehicle NO emission standard
A
that represents a 75 percent reduction from pre-controlled levels starting
with the 1985 model year. Although EPA has not yet set the standard, it
has prepared an analysis of alternative standards and the most probable
control systems to be used to meet those standards (U.S. EPA, 1981).
Because the NO control options differ for heavy-duty gasoline and
diesel engines, these two vehicle types are analyzed separately.
EPA expects that manufacturers of heavy duty gasoline-fueled vehicles
(HDGV's) will adopt three-way and three-way plus oxidation catalyst
systems together with EGR and air injection to comply with the emission
7-27
-------
TABLE 7-14
ANNUALIZED CAPITAL CHARGES FOR NO CONTROL
IN 1990 FOR LIGHT-DUTY TRUCKS (0-^,500 Ibs)
(1980 Dollars)
Total Capital Cost
Model (x 10 ) (From)
Year (Tables 7-12 & 7-13)
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
TOTAL
$739.2
733.1
722.2
710.0
699.2
693.0
39.2
36.7
34.1
34.9
35.3
32.1
17.7
12.2
13.4
11.0
14.7
16.8
$4,594.8
Scrappage
Rate*
1.00
0.97
0.93
0.88
0.83
0.78
0.73
0.68
0.63
0.59
0.55
0.51
0.47
0.43
0.40
0.36
0.32
0.28
Capital
Recovery
Factor
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
Annualized
Capital ,.
Charge (xlO&)
$120.34
115.77
109.34
101.72
94.48
88.00
4.66
4.06
3.50
3.35
3.16
2.67
1.35
0.85
0.87
0.64
0.77
0.77
$656.30
* SOURCE: EEA, Inc., 1980. Table 3-7.
7-28
-------
--J
I
TABLE 7-15
ANNUAL FUEL ECONOMY DIFFERENCES DUE TO THE FMVCP IN 1990 FOR LDTl'S
Model
Year
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
Annual VMT
Per LDT1*
15,610
14,049
12,488
11,239
10,147
9,366
8,585
7,961
7,493
7,025
6,556
6,244
5,932
5,463
5,151
4,996
4,683
4,370
Fuel Consumption
(mpg)**
24.5
24.5
24.5
24.5
24.5
23.5
24.2
23.3
21.4
19.9
18.9
18.4
18.0
18.2
16.6
13.1
11.0
11.2
Gallons of
Fuel Used
637
573
510
459
414
399
355
342
350
353
347
339
330
300
310
381
426
390
New LDT1 ,
Sales x 10
2.54
2.52
2.48
2.44
2.40
2.37
2.28
2.13
1.99
2.02
2.04
1.87
1.94
1.34
1.47
1.21
1.62
1.85
Scrappage
Rate
1.00
0.97
0.93
0.88
0.83
0.78
0.73
0.68
0.63
0.59
0.55
0.51
0.47
0.43
0.40
0.36
0.32
0.28
Fuel Economy
Benefit (Loss)
0
0
0
0
0
0
- .01
- .01
- .01
- .01
- .01
- .01
- .01
- .01
- .01
- .01
- .01
- .01
Annual Savings
in 1990 (x 10 )
$0
0
0
0
0
0
-7.68
-6.44
-5.70
-5.47
-5.06
-4.20
-3.91
-2.25
-2.37
-2.16
-2.87
-2.63
$-50.74
* SOURCE: EEA, Inc., 1980.
** SOURCE: McNutt et al, 1979.
The price of gasoline is assumed to be $1.30 per gallon (1980 dollars).
-------
TABLE 7-16
ANNUAL FUEL ECONOMY DIFFERENCES DUE TO THE FMVCP IN 1990 FOR LDT2'S
co
o
Model
Year
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
1979
TOTAL
Annual VMT
Per LDT2*
15,800
14,220
12,640
11,376
10,270
9,480
8,690
8,058
7,584
7,110
6,636
6,320
Fuel Consumption
(mpg)**
20.2
20.2
20.2
20.2
20.2
20.2
19.8
.1
.5
19.
17.
16.4
15.4
15.1
Gallons of
Fuel Used
782
704
626
563
508
469
439
422
433
434
431
419
New LDT1 ,
Sales x 10
2.26
2.24
2.21
2.17
2.14
2.13
2.03
1.90
1.76
1.81
1.84
1.66
Scrappage
Rate
1.00
0.97
0.93
0.88
0.83
0.78
0.73
0.68
0.63
0.59
0.55
0.51
* SOURCE: EEA, Inc. 1980.
** SOURCE: McNutt et al, 1979.
The price of gasoline is assumed to be $1.30 per gallon (1980 dollars).
Fuel Economy
Benefit (Loss)
0
0
0
0
0
0
- .01
- .01
- .01
- .01
- .01
- .01
Annual Savings
in 1990 x 10°)
$0
0
0
0
0
0
-8.46
-7.09
-6.24
-6.03
-5.67
-4.61
-$38.1
-------
standards proposed for 1985. EPA estimates that 10 percent of the
heavy-duty engines will be able to meet the revised emission standards
with the use of three-way catalyst/EGR technology and the remaining 90
percent will require three-way plus oxidation catalyst/EGR technology.
Using these percentages, the initial cost of emission control hardware
for HDGV's is estimated by EPA to be $267. Including the added cost to
*
manufacturers for research and development certification .and allowable
maintenance yields an expected initial price increase of $284 for HDGV's.
This $284 per vehicle cost estimate is used in Table 7-17 along with
HDGV sales data to estimate the total capital cost increase due to NO
controls, which is estimated to be $140.3 million in 1985 and $829.6
million in 1990.
For heavy-duty diesels (HDD's), the control techniques expected for 1985
and later model year vehicles could include:
Electronic engine controls
Exhaust gas recirculation
Intercooling/aftercooling
Injector combustion chamber and other engine modifications
Retarded injection timing
High pressure fuel injection.
EPA estimates the hardware costs for the most probable combination of
these controls to be $717, but qualifies this estimate by saying that it
could be off by as much as 50 percent for some engine families. By
including research and development, certification, and in-use testing
costs, the expected increased cost to the consumer for HDD NO controls
is $741.
7-31
-------
TABLE 7-17
CAPITAL COST INCREASES FOR HEAVY-DUTY GASOLINE VEHICLES
DUE TO NO CONTROLS IN 1990
x
Model
Year
1990
1989
1988
1987
1986
1985
TOTAL
Estimated
HDGV Sales*
459,000**
474,000
489,000
504,000
501,000
494,000
Initial Cost Increase
Per HDGV
$284
284
284
284
284
284
Total Capital
Cost (x 10 )
$130.4
134.6
138.9
143.1
142.3
140.3
$829.6
* SOURCE: U.S. EPA, 1981.
** Extrapolated from the U.S. EPA sales estimates for the 1985-1989 model
years.
7-32
-------
The fuel economy of HDD engines may suffer as a result of the proposed
regulations. However, at this point, EPA cannot accurately quantify the
full fuel economy effects on HDD engines. Therefore, it is assumed here
that no fuel economy effects will occur with the new NO controls.
Table 7-18 shows the estimated capital cost increase due to NO controls
for HDD's in 1990 to be $2.1 billion. The equivalent cost in 1985 is
estimated to be $287.5 million. Adding these costs to the HDGV costs,
gives 1990 total costs for HDV's of $2.92 billion and 1985 capital costs
of $1.12 billion.
Annualized capital charges for NO control in 1990 for HDV's are estimated
A
as shown in Table 7-19. The annualized costs in 1990 are estimated to
be $391 million and the 1985 annualized costs are estimated to be $70
million.
7.2.4 Summary
Tables 7-20 and 7-21 summarize the total annualized costs of the FMVCP
which can be attributed to NO for 1985 and 1990. Where the costs or
x
benefits of NO versus HC or CO control could not be differentiated, the
x '
total cost of the FMVCP was divided by three to estimate the portion
attributable to NO .
7-33
-------
TABLE 7-18
CAPITAL COST INCREASES FOR HEAVY-DUTY DIESEL VEHICLES
DUE TO NO CONTROLS IN 1990
Model
Year
1990
1989
1988
1987
1986
1985
TOTAL
Estimated
HDDV Sales*
539,000**
514,000
489,000
465,000
427,000
388,000
Initial Cost Increase
Per HDDV
$741
741
741
741
741
741
Total Capital
Cost (x 10 )
$399.4
380.9
362.3
344.6
316.4
287.5
$2,091.1
* SOURCE: U.S. EPA, 1981.
** Extrapolated from the U.S. EPA sales estimates for the 1985-1989 model
years.
7-34
-------
TABLE 7-19
ANNUALIZED CAPITAL CHARGES FOR NO CONTROL IN 1990
FOR HEAVY-DUTY VEHICLES
Model
Year
1990
1989
1988
1987
1986
1985
TOTAL
HDGV
Total Capital
Cost (x 10 )
$130.4
134.6
138.9
143.1
142.3
140.3
Scrappage*
Rate
1.00
0.95
0.88
0.79
0.69
0.61
Capital
Recovery
Factor'
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
Annualized
Capital ,
Charge (x 10 )
$21.23
20.82
19.90
18.40
15.98
13.93
$829.6
$110.26
Model
Year
1990
1989
1988
1987
1986
1985
TOTAL
HDDV
Total Capital
Cost (x 106)
$399.4
380.9
362.3
344.6
316.4
287.5
Scrappage^
Rate
1.00
0.91
0.84
0.77
0.71
0.64
Capital
Recovery
Factor**
0.1628
0.1628
0.1628
0.1628
0.1628
0.1628
Annualized
Capital ,
Charge (x 10 )
$65.02
56.43
49.55
43.20
36.57
29.96
$2,091.5
$280.73
* SOURCE: EEA, Inc., 1980.
** Assumes a 10 percent discount rate and a 10-year time horizon.
7-35
-------
TABLE 7-20
TOTAL ANNUALIZED COST OF THE FEDERAL MOTOR
VEHICLE CONTROL PROGRAM IN 1985
(millions of 1980 dollars)
Total Annualized Cost
Vehicle Type of NO Control
** x
Light-Duty Vehicles
Hardware $997
O&M 165
Fuel Economy (342)
Unleaded Fuel Cost 327
Subtotal $1147
Light-Duty Trucks
Hardware 150
Fuel Economy 96
Subtotal 246
Heavy-Duty Vehicles 70
TOTAL $1,463
7-36
-------
TABLE 7-21
TOTAL ANNUALIZED COST OF THE FEDERAL MOTOR VEHICLE CONTROL PROGRAM
IN 1990
(1980 dollars)
Total Annualized Cost
Vehicle Type of NO Control
i£ x
Light-Duty Vehicles
Hardware $1,723
O&M 303
Fuel Economy (996)
Unleaded Fuel Cost 633
Subtotal $1,663
Light-Duty Trucks
Hardware $ 656
Fuel Economy 89
Subtotal 745
Heavy-Duty Vehicles 391
TOTAL $2,799
7-37
-------
REFERENCES FOR SECTION 7
Energy and Environmental Analysis, Inc. 1980. Techniques For Esti
mating MOBILE2 Variables. Arlington, VA. July. (Prepared for U.S.
EPA, Ann Arbor, MI.)
Evans, R. and Castaldini, C. 1978. Summary of Combustion Modification
NO Controls - Emission Levels, Costs, and Fuel Impacts. Acurex Corpora-
tion, Mountain View, CA. March.
Lindgren, L.H. 1978. Cost Estimations For Emission Control Related
Components/Systems and Cost Methodology Description (EPA-460/3-78-002)
Rath and Strong, Inc., Lexington, MA. March. (Prepared for U.S. EPA,
Ann Arbor, MI.)
McNutt, B.; Dulla, R.; Lax, D. 1979. "Factors Influencing Automotive
Fuel Demand," SAE Technical Paper Series No. 790226. SAE Congress and
Exposition, Detroit, MI. February.
Murrell, J.D.; Foster, J.A.; Bristor, D.M. 1980. "Passenger Car and
Light Truck Fuel Economy Trends through 1980." SAE Technical Paper
Series No. 800853. Passenger Car Meeting, Dearborn, MI. June.
U.S. Environmental Protection Agency (U.S. EPA). 1977 Stationary Gas
Turbines - Standard Support and Environmental Impact Statement Volume I:
Proposed Standards of Performance (EPA-450/2-77-017a). Emission Standards
and Engineering Division, OAQPS, Research Triangle Park, NC. September.
U.S. EPA. 1978a. Control Techniques for Nitrogen Oxides Emissions
from Stationary Sources, Second Edition (EPA-450/1-78-001). Research
Triangle Park, NC. January.
U.S. EPA. 1978b. Compilation of Air Pollutant Emission Factors, Third
Edition (AP-42). Research Triangle Park, NC. May.
U.S. EPA. 1978c. Electric Utility Steam Generating Units - Background
Information for Proposed NO Emission Standards (EPA-450/2-78-005a).
Research Triangle Park, NC.X July.
U.S. EPA. 1979a. Cost and Economic Impact Assessment For Alternative
Levels of the National Ambient Air Quality Standard for Ozone
(EPA-450/5-79-002). Strategies and Air Standards Division, OAQPS,
Research Triangle Park, NC. February.
U.S. EPA. 1979b. Stationary Internal Combustion Engines (Draft EIS)
Standards Support'and Environmental Impact Statement Volume I: Proposed
Standards of Performance (EPA-450/2-78-125a). Emission Standards and
Engineering Division, OAQPS, Research Triangle Park, NC. July.
7-38
-------
REFERENCES FOR SECTION 7 (Continued)
U.S. EPA. 1981. Draft Regulatory Analysis, Environmental Impact State-
ment and NO Pollutant Specific Study for Proposed Gaseous Emission
RegulationsXFor 1985 and Later Model Year Light-Duty Trucks and 1986 and
Later Model Year Heavy-Duty Engines. Office of Mobile Source Air Pollution
Control, Ann Arbor, MI. January.
7-39
-------
8. ECONOMIC IMPACT ANALYSIS OF THE ANNUAL STANDARD FOR
8.1 INTRODUCTION
This section of the report estimates the economic impact of the current
annual N0» standard (0.053 ppm) in 1990. The results are presented in
two parts to facilitate the proper distribution of control costs by
economic sector: stationary source (including point and area) impacts
and mobile source impacts. Stationary source costs are divided into the
industrial sector (by SIC), residential sector, and commerical sector.
Mobile source costs are disaggregated by government and household sectors.
The economic impacts are derived from "costs to the firm" as distin-
guished from "costs to society." The procedures for estimating these
costs in terms of initial investments and annualized expenditures are
discussed in Section 4.1.4. Strictly speaking, these costs are "engineer-
ing" rather than "economic" costs. The distinction may be important
when considering mobile source control because the value of time spent
waiting for auto inspection, for example, is not considered.
Second, only impacts resulting from primary costs are evaluated. There
is no attempt to incorporate any change in the price of the control
technology due to increased demand. Also not included is the admini-
strative cost of control strategy implementation or enforcement. No
monitoring costs or inspection/audit costs, for example, are estimated.
8.2 STATIONARY SOURCE IMPACTS
8.2.1 Introduction
The costs imposed on stationary sources include costs to point sources
(see Section 5.1) and stationary area sources (see Section 5.2). All of
8-1
-------
the point source costs as well as the majority of the area source costs
are borne by the industrial sector. The point source costs and certain
of the area source industrial costs are distributed by SIC. These are
added together and analyzed concurrently using SIC-specific industry
data to estimate the potential impact of control costs. The remainder
of the industrial costs, generated from county fuel consumption data,
cannot be attributed to any specific industry. Impacts from these costs
cannot be examined fully because the incidence of cost burden cannot be
determined.
The remaining stationary source costs are distributed between the commer-
cial and residential sectors of the economy. In addition to the cost
totals, the number of counties bearing these costs is presented. The
average impact per county then is estimated by dividing the costs by the
number of counties. Finally, the impact on the average county incurring
the cost is estimated using typical income data by economic sector for
counties within various population groups.
8.2.2 Economic Impact to the Industrial Sector: Approach
The economic analysis of industrial control costs is divided into two
sections: capital availability to cover the initial expenditure for
control equipment and annual impact on the cost of production.
The following steps were used to determine the ability of a firm to
shoulder the burden of capital expenses incurred by pollution control
requirements:
Screen for industries significantly affected
Compare capital required to historical capital expenditures of
a typical company
Compare capital cost to total long-term debt of a typical
company
8-2
-------
Examine debt/equity ratios and beta values* to determine the
ability of a typical company to raise additional capital.
Screening consists of examining the capital requirements for 1) the
number of plants requiring control and 2) the percentage of recent
capital expenditure for each industry represented by controls to meet
the standard. If the number of sources is limited (e.g., one plant) and
capital costs are less than one percent of recent expenditures, the
industry is removed from the study. (Fluctuations of one percent and
less are assumed to be normal.) However, an industry comprised mostly
of small firms is retained in the study because smaller changes in cost
can still affect small companies significantly.
If the screening shows the possibility of significant impact, capital
costs are compared to total long-term debt for a typical firm to indicate
the ability of a company to absorb the expense. Small changes in debt
will not affect the posture of a firm; large changes are examined for
the effect on the debt/equity ratio. An increase in the debt/equity
ratio or an already high ratio together with a high beta value indicates
that raising capital may prove difficult.
The increase in the annual costs of production is determined from an
annualized cost calculation. Figure 8-1 presents a diagram of the
components of the annualized cost analysis. A screen for significant
impact within the industries is used to eliminate industries whose price
impact is less than 5 percent.** The estimator used for the screening is
the percentage increase in the cost of goods sold within the industry.
* The bet\a value is a financial indicator and is the co-variance between
risk and the cost of equity.
** Five percent was chosen for consistency with the EPA's "Criteria for
Conducting Regulatory Analysis" (Federal Register, 1978).
8-3
-------
FIGURE 8-1
ANNUALIZED COST METHODOLOGY
Annualized
Costs
Adjust for
After-Tax Cost
Screen for
5 Percent
DO
I
Price Impact
Full Cost
Pass-Through
Domestic
Competition
Impact to Firm
Import
Competition
-------
If the screening indicates the possibility of significant impacts, the
average cost impact is calculated by dividing the tax-adjusted aggregate
annualized cost for each industry by the number of sources controlled
under each standard. It is assumed that this cost is representative of
the cost faced by a typical firm.
By further assuming full pass-through of cost, the increase in the
average cost of production corresponds to the price increase; that is,
increases in production cost would be offset by increases in price.
Accounting for the effect of taxes (because 50 percent of the marginal
revenue goes to corporate income tax), the price impact is twice the
increase in the average cost of production. The magnitude of the cost
impact, viewed in terms of a percentage price increase, can be identi-
fied by comparing the price increase to a weighted average product
price.
The price impact is examined to determine if the firm's competitive
posture within its own industry would be affected. Foreign competition,
transportation costs, and availability of sources of substitute products
are considered.
8.2.3 Economic Impact to the Industrial Sector: Results
The total initial cost for the industrial sector is estimated to be $85
million. Over 28 percent or $23.9 million is attributed to the oil and
gas extraction industry (SIC 13). This is the largest portion paid by
one industry. Other major contributors of total initial cost are the
chemical industry (SIC 28), $9.1 million or 11 percent, and electrical
utilities (SIC 49), $9.0 million or 10.6 percent. Over $32 million in
industrial initial cost, or 38 percent of the total, results from control
of stationary industrial area sources which, because of data constraints
imposed by NEDS, cannot be attributed to any industry. Other industries
that contribute more than a million dollars to the total are petroleum
8-5
-------
refining (SIC 29), $4.4 million, and the primary metal industry (SIC 33),
$4.5 million.
Table 8-1 presents historic aggregate annual capital expenditures expressed
in 1980 dollars for each industry incurring initial cost as well as the
percentage of capital expenditure that the initial expenditure for the
annual NO- standard represents. None of the industrial initial investment
costs exceed six-tenths of one percent of the expected capital budgets
of the industry. Therefore, since fluctuation in capital expenditures
of one percent are considered normal, no industry should have a difficulty
in raising the necessary capital.
The total annualized cost of the control for the annual standard is
estimated as $23 million dollars nationwide. Unfortunately, over 57
percent of this total or $13.2 million cannot be attributed to any
industry because of the data limitations in the NEDS inventory. The
remaining costs of $9.8 million are divided primarily between the oil
and gas extraction industry ($5.8 million), and electric utilities ($2.2
million). *
Table 8-2 presents historical annual revenue for each industry and the
percentage indicating cost of goods sold. The total cost of goods sold
is estimated by using these data. The percentage increase in cost of
goods sold resulting from the control costs for the NO- annual standard
is then calculated. None of the estimated increases exceed 0.025 percent
of the estimated cost of goods sold. Therefore, no industries indicate
any significant impact resulting from annualized cost.
8.2.4 Economic Impact on the Commercial Sector
The commercial sector of the economy bears $35.8 million in initial
investment and $12.4 million in annualized cost. These costs are incurred
by just two counties nationwide, both of which are in large urban areas.
8-6
-------
TABLE 8-1
INITIAL COST BURDEN FOR THE
INDUSTRIAL SECTOR BY SIC
Historical
Initial Cost of NO, Capital Expenditure* Percent of Capital
($10 ) (%) Expenditure
5,697 0.006
4,222 0.566
2,510 0.017
1,610 0.013
8,956 0.102
36,688 0.012
6,654 0.059
3,339 0.002
29,901 0.030
* Historical annual capital expenditure expressed in 1980 dollars
Capital expenditure data obtained from U.S. Industrial Outlook 1979"
U.S. Department of Commerce and "Statistical Abstract of the United
States 1979" U.S. Department of Commerce.
SIC
10
13
14
24
28
29
33
46
49
Control ($000)
349
23 , 900
431
210
9,117
4,427
4,504
78
8,992
8-7
-------
DO
TABLE 8-2
ANNUALIZED COST IMPACT IN THE INDUSTRIAL SECTOR
Annualized Cost
SIC
10
13
14
24
28
29
33
46
49
Revenue*
($iob)
6,742
37,132
7,770
43,064
64,879
372,916
128,196
4,711
96,205
Percent Cost of**
Goods Sold
55
65
60
82
65
71
74
56
65
Estimated Cost
of Goods Sold
3,708
24,136
4,662
34,312
42,171
264,770
94,865
2,638
62,533
Cost for NO
Annual ($000)
82
5,800
76
47
201
969
760
26
2,157
Percent Increase in
Cost of Production
0.002
0.024
0.002
0.0
0.0
0.0
0.001
0.0
0.003
* Historical revenue expressed in 1980 dollars in "Almanac of Business and Industrial
Financial Ratios", Leo Troy, 1978.
** Dun & Bradstreet, Inc., 1977. "Cost of Doing Business."
-------
The control costs within these counties would be paid directly from each
county's commercial receipts. Typically, such receipts average between
$10 billion and $30 billion annually for counties with a population of
500,000, or more than $750 billion annually nationwide (U.S. Department
of Commerce, 1977).
For the sake of a preliminary impact estimate, the commercial receipts
in the two counties are assumed to be equal to $20 billion annually.
The initial control cost represents 0.18 percent of average commercial
receipts and the annualized cost represents 0.06 percent of the average
receipts for a typical set of urban counties.
Due to data limitations within the NEDS inventory, the cost to the
commercial sector cannot be distributed within the sector. Because we
do not know how many firms bear the cost, it is impossible to determine
the extent of any impact beyond the level of gross receipts. From these
calculations, barring an unusual distribution of costs, no significant
impact can be expected.
8.2.5 Economic Impact on the Residential Sector
The residential sector of the economy incurs a capital cost of $2.5
million in the same two large urbanized counties. Typical counties
within urbanized areas have populations of 500,000 to 3,000,000 (U.S.
Department of Commerce, 1977). Even if the counties involved are at the
low end of the range, the average cost per capita would only be $2.50.
The corresponding annualized cost, however, is negative due to the
"imposition" of cost-saving controls. The savings amount to almost $2
million per year. Thus, there is no significant impact expected in the
residential sector.
8-9
-------
8.3 MOBILE SOURCE CONTROL IMPACTS
8.3.1 Approach
Although all of the costs are borne by the motorist directly, the costs
are not borne equally by all motorists. The cost of EGR inspection must
be divided into two areas: the cost of inspection and the cost of
repair/re-inspection.
The total initial inspection cost is calculated by multiplying the
estimated number of cars inspected under the program by $1.00. The
inspection cost then is divided into the income groups of car owners who
pay for the inspection, using Department of Transportation statistics.
The resulting distribution of cost by income group is examined to deter-
mine where the burden of cost is heaviest. It is already known that the
average car owner pays $1.00 but this distribution tells what, if any,
redistributive effect results from the program.
The remaining cost of the mobile source control measure is borne directly
by the motorist whose car fails the inspection. This, in turn, is a
function of vehicle age. Failure rate was related to age through the
cumulative mileage-failure rate relationship in MOBILE2 and EPA's esti-
mated annual mileage for cars of various ages. The results are shown in
Table 8-3.
Using these data, a statistic indicating likelihood of failure by age
was created. The probability of failure for each age group was divided
by the probability of failure of a car of mean age. This age-failure
index is presented in Table 8-4. The statistic then can be applied to
determine the increased or decreased probability of failure as a function
of age.
8-10
-------
TABLE 8-3
FACTION OF AUTOMOBILE FLEET FAILING EGR BY
MODEL YEAR IN 1990
Year
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
Age
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Cummulative
Mileage
14.4
28.3
41.7
54.4
66.8
78.6
89.9
100.5
110.7
120.3
129.4
138.0
146.0
153.5
160.0
166.4
172.3
177.7
Failure Rate by
Model Year
0.5
0.8
1.1
1.5
1.8
2.1
2.4
2.7
3.0
3.2
3.5
3.7
3.9
4.2
3.4
3.7
3.8
3.9
8-11
-------
TABLE 8-4
AGE FAILURE INDEX
Index
Year Failure Rate Observed t Mean
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Mean
0.5
0.8
1.1
1.5
1.8
2.1
2.4
2.7
3.0
3.2
3.5
3.7
3.9
4.2
3.4
3.7
3.8
3.9
2.73
0.183
0.293
0.403
0.549
0.659
0.769
0.879
0.989
1.099
1.172
1.282
1.355
1.429
1.54
1.26
1.355
1.391
1.429
1.000
8-12
-------
A distribution of age of car by income of the owner was found in the
National Personal Transportation Study (U.S. Department of Transporta-
tion, 1974). These data are presented in Table 8-5. For each income
level and corresponding mean age of car, the table also shows the age-
failure index, which is calculated by linear interpolation between whole
years.
A distribution of repair costs is determined from these statistics by
multiplying income group ownership by the corresponding age-failure
index. The resulting distribution is normalized by dividing by the
summation of the age-failure index times the percent of cars by income.
This provides a distribution of cost by income group. In functional
form, the statistic is calculated as follows:
IG. AF. (8-1)
11
I (IG. AF.)
i
where
IG = percentage of cars by income group
AF = age-failure index
i = income group.
Each factor is presented in Table 8-5. The normalized distribution of
repair cost is multiplied by the total repair cost to indicate the
relative burden of repair cost by income group.
It should be noted that implicit within this calculation is the assump-
tion that the mean failure rate of a car whose owner is part of a par-
ticular income group can be estimated by the failure rate of a car of
mean age if distribution within each income group were normal. The re-
sulting distribution therefore provides only an indicator of the burden
of cost.
8-13
-------
TABLE 8-5
DISTRIBUTION OF REPAIR COST FACTORS
Income Level*
<$4,800
4,800-6,400
6,400-8,000
8,000-9,600
9,600-12,100
12,100-16,000
16,000-24,100
>24,100
Unreported
Age
7.0
6.1
6.2
6.0
5.6
4.8
4.6
4.0
5.5
Age-Failure
Index
0.879
0.780
0.791
0.769
0.725
0.637
0.615
0.549
0.714
Percent of
Cars
5.91
4.74
4.68
7.40
12.60
18.02
24.53
12.50
9.61
Index
x Percent
0.051949
0.05772
0.037018
0.056906
0.09135
0.114787
0.150859
0.068625
0.0686154
Distribution
Factor
0.0744
0.0827
0.0531
0.0815
0.1309
0.1645
0.2162
0.0978
0.0976
^Expressed in 1980 dollars.
8-14
-------
8.3.2 Results
The total annual cost of mobile source control NCL is $14.7 million
dollars distributed among eight counties within two urbanized areas.
There is no initial expenditure because ISM programs are assumed to
already be in place within the counties likely to require control of
mobile sources for N0_.
The EGR inspection affects an estimated 8.23 million cars within the
eight counties. Each motorist within the county is required to pay
$1.00 for the incremental cost of EGR inspection. Therefore $8.23
million dollars in cost are distributed among all motorists for the
program. Table 8-6 indicates the distribution of these costs by income
group.
The distribution indicates that the largest portion of the cost is borne
by the $16,000-$24,000 income group, which owns approximately 1.6 cars
per household and pays 24.7 percent of the toal cost, or $2.02 million.
The smallest portion of the cost is paid by the $6,400-$8,000 income
group, which pays only 4.7 percent of the total cost of $385,000.
The number of cars per household increases with income as illustrated in
Figure 8-2. Once the $4,800 mean level is reached, the marginal increase
in number of cars per increased dollar in income diminishes with increasing
income. This can be seen from the distance between the midpoint of the
distribution and the dashed line. Therefore, in proportion to income, a
low income household will face a larger burden than will a high income
family. This does not imply that the total burden on the low income
group is disproportionate to its segment of the population: in 1979,
approximately 12.8 percent of all households earned less than $6,400 per
year; by comparison, the same group pays only 10.6 percent of the total
inspection cost.
8-15
-------
TABLE 8-6
DISTRIBUTION OF EGR INSPECTION COSTS BY INCOME
Income Group
<$4,800
4,800-6,400
6,400-8,000
9,600-12,100
12,100-16,000
16,000-24,000
>24,000
Unreported
TOTAL
Percent
of Cars
5.91
4.74
4.68
12.60
18.02
24.53
12.50
9.61
Average Number of
Autos Per Household
0.4
0.8
0.9
1.2
1.3
1.6
1.9
1.2*
Inspection Costs
($000)
$ 486
390
385
609
1,483
3,019
1,029
791
$8,230
*Mean number of autos per household
8-16
-------
FIGURE 8-2
CARS PER HOUSEHOLD VS. INCOME
CARS PER HOUSEHOLD
2.5 -i
2.0 -
1.5-
1.0 -
0.5-
0.0
2,000
6,000 10,000 14,000 18,000
20,000 24,000
INCOME IN DOLLARS
8-17
-------
The remaining annual costs ($6.47 million) are distributed among the
motorists who failed the inspection. This amount pays for repair of the
EGR as well as inspection. The cost incidence was calculated using the
age-failure index technique discussed earlier and the results are shown
in Table 8-7.
The largest portion of the cost nearly $1.14 million is borne by
the $16,000-24,000 income group and represents 22 percent of the $6.47
million total for repair and re-inspection.
It should be remembered that the distribution of cost by income group
represents a distribution based on estimated failure rates from MOBILE2.
The actual distribution of the cost could differ from the estimates
presented depending on the individual threshold values of emissions used
to determine failure in each program.
The effect of increased failures in the older cars tends to exacerbate
the problems of the low income group. As Table 8-7 shows, the inclusion
of the age-failure index causes the distribution of repair costs to be
skewed towards low income groups compared to the original distribution
of car ownership.
It is also important to note the limitation placed on the accuracy of
the distribution by the small number of counties affected. The data for
age of cars in each income group represent national averages. By reducing
the sample size to a small number of counties, the validity of using
national data is greatly reduced.
The shift in distribution, while never more than 3 percent for any
income group, adds to the adverse impact of the mobile source costs.
Again, the total burden on the income group does not appear dispropor-
tionate; but cost as a percentage of income is greater for the low
8-18
-------
oo
TABLE 8-7
DISTRIBUTION OF MOBILE SOURCE COST FACTORS
Cost of Repair
Income Level
<$4,800
4_, 800 -6, 400
6,400-8,000
8,000-9,600
9,600-12,100
12,100-16,000
16,000-24,100
>24,100
Unreported
Age
7.0
6.1
6.2
6.0
5.6
4.8
4.6
4.0
5.5
Age-Failure
Index
1.079
1.079
1.079
1.079
1.073
1.060
1.057
1.049
1.069
Percent of
of Cars
5.91
4.74
4.68
7.40
12.60
18.02
24.53
12.50
9.61
Distribution
by Income
0.0744
0.0827
0.0531
0.0815
0.1309
0.1645
0.2162
0.0978
0.0976
and Re-Inspection
($000)
481
535
344
527
847
1,064
1,399
633
631
-------
income groups. However, because the total magnitude of the costs is
low, the impacts should be low as well.
8.4 URBAN AND COMMUNITY IMPACT ANALYSIS FOR THE ANNUAL STANDARD
The urban and community impact analysis (UCIA) is intended to highlight
anticipated consequences of major Federal policies, programs, and regula-
tory changes for different types of jurisdictions. The focus of such an
analysis is the socio-economic impacts that may adversely affect the
distressed communities. The prime indicators for these impacts are
fiscal conditions of governments, income, and unemployment.
One of the primary purposes of the UCIA is to identify those cities or
types of cities likely to incur an additional economic burden due to the
increased costs. There have been many studies identifying the most
economically depressed cities in the United States. A recent study by
the Urban Institute (1978) examined the country's 153 largest cities on
the basis of three distress indicators:
Population decline between 1979 and 1976 of 2 percent or more
Per capita increases less than the all-city average in 1970
Unemployment rates greater than the all-city average in 1976.
The cities were grouped into four categories, with Category 0 comprising
cities not exceeding any of the distress indicators and Category 3
comprised of cities exceeding all of the distress indicators. The
counties requiring mobile source and stationary area source control are
in Category 2. However, due to the small number of counties affected,
it is difficult to draw conclusions regarding any adverse effect that
NO- control would have on distressed cities. If, indeed, only two urban
areas must expend funds to meet the current annual NAAQS, then whatever
impacts are experienced will be highly limited geographically.
8-20
-------
The UCIA also is concerned with the effects of control on disposable
income, which could arise from two primary causes: increased costs of
goods and services or reduced employment levels within the community,
either of which will have an adverse impact on real income. However, as
was seen in the economic analysis, none of the areas of the economy
shows an indication of any adverse impact. No price impacts were seen
and there was no indication of employment reductions resulting from the
cost.
The methodology used to estimate the costs does not allow sufficient
disaggregation of cost to determine any impact to a specific city.
While that was never the intent of the report, this limitation in-
herently prevents a positive statement from being made regarding urban
and community impacts to the two communities incurring costs. However,
the low magnitude of the costs as well as the lack of any significant
impact in any economic sector in terms of costs or income distribution
leads to a relatively certain conclusion that no significant adverse
socio-economic impacts result from the annual standard of 0.053.
8-21
-------
REFERENCES FOR SECTION 8
Dun & Bradstreet, Inc. 1977- "The Cost of Doing Business."
Federal Register. 1978. "Improving Government Regulations: Proposal
for implementing Executive Order 12044." July 11.
Troy, Leo. 1978. Almanac of Business and Industrial Financial Ratios.
New York: Prentice-Hall.
U.S. Department of Commerce (U.S. DOC). 1977. City-County Databook.
Washington, D.C.
U.S. DOC. 1980a. "Statistical Abstract of the United States, 1979."
Washington, D.C.
U.S. DOC. 1980b. "U.S. Industrial Outlook, 1979." Washington, D.C.
U.S. Department of Transportation, Federal Highway Administration.
1974. National Personal Transportation Study. Report II. "Automobile
Ownership." Washington, D.C.
U.S. Environmental Protection Agency. 1980. User's Guide to MOBILE2:
Mobile Source Emissions Model. Ann Arbor, MI.
The Urban Institute. 1978. "Distressed City Indicator." Cited in the
President's National Urban Policy Report, 1978. Prepared for the U.S.
Department of Housing and Urban Development.
8-22
------- |