United States
Environmental Protection
Agency
Office of Air Quality
Planning and Standards
Research Triangle Park NC 27711
EPA-450/2-80-072
June 1980
Air
-------
-------
EPA-450/2-80-072
of
De
for Criteria Pollutants
by
PEDCo Environmental, Inc.
Durham, North Carolina 27701
Contract No. 68-02-3173
Task. No. 8
EPA Project Officer: James Weigold
New Source Review Office
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Control Programs Development Division
Research Triangle Park, North Carolina 27711
June 1980
-------
-------
CONTENTS
Figures
Tables
Acknowledgment
1. Introduction 1
1.1 Proposed PSD Regulations 1
1.2 De Minimis' Concept 2
2. Purpose of the Analysis 5
2.1 Determination of Air Quality Impacts of
Sources that have Received Permits to Date 5
2.2 Determination of Emission Levels Associated
with Certain Air Quality Impacts 6
2.3 Number of Sources to be Affected by De
Minimis Concept 7
2.4 Urbanwide Air Quality Impact Due to De
Minimis Changes in Emissions 7
2.5 Impact With Respect to Class I Areas 8
3. Methodology 9
3.1 Selection of Sources to be Evaluated 9
3.2 Engineering Analysis 12
3.3 Model Selection 14
3.4 Prediction of Air Quality Impacts by Use of
Selected Models 16
3.5 Relationship Between Specific Emission Levels
and Air Quality Concentrations 18
3.6 Urban Area Impact of Major Sources Making
De Minimis Changes 20
3.7 Class I Area Protection 21
4. Results 22
4.1 Distribution of Emission Levels 22
4.2 Distribution of Actual Stack Heights 32
4.3 Distribution of Effective Stack Heights 32
4.4 Distribution of Concentrations 39
4.5 Emissions Associated With Specific Air
Quality Levels 45
4.6 Urban Area Air Quality Impact Due to De
Minimis Changes in Emissions 54
4.7 Class I Area Protection 54
i i i
-------
4.8 Number of Sources Affected by the Proposed
De Minimis Emission Levels
58
IV
-------
FIGURES
Number Page
1 Summary Form for September 5, 1979, Proposed PSD
Regulations 10
2 Distribution of Particulate Matter Emissions for
All Sources Used in the Analysis 23
3 Distribution of Nitrogen Oxide Emissions for All
Sources Used in the Analysis 24
4 Distribution of Sulfur Oxide Emissions for All
Sources Used in the Analysis 25
5 Distribution of Carbon Monoxide Emissions for All
Sources Used in the Analysis 26
6 Distribution of Particulate Matter Emissions for
Sources that Emit Less Than 100 tons/year 27
7 Distribution of Nitrogen Oxide Emissions for
Sources that Emit Less Than 100 tons/year 28
8 Distribution of Sulfur Oxide Emissions for Sources
that Emit Less Than 100 tons/year 29
9 Distribution of Carbon Monoxide Emissions for
Sources that Emit Less Than 100 tons/year 30
10 Distribution of Actual Stack Heights 33
11 Distribution of Effective Stack Heights - Volume 10
Method 34
12 Distribution of Effective Stack Heights - Volume 10
Adjusted Method 35
13 Distribution of Effective Stack Heights - D
Stability and 5m/s Windspeed 36
14 Distribution of Effective Stack Heights - Volume 10
Adjusted for Sources that Emit Greater Than 100
tons/year 37
-------
Number
15 Distribution of Effective Stack Heights - Volume 10
Adjusted for Sources that Emit Less Than 100
tons/year 38'
16 Distribution of TSP Concentration Using Adjusted
Volume 10 for Sources that Emit Less Than 100
tons/year 40
17 Distribution of S02 Concentration Using Adjusted
Volume 10 for Sources that Emit Less Than 100
tons/year 41
18 Distribution of N02 Concentrations Using Adjusted
Volume 10 for Sources that Emit Less Than 100
tons/year 42
19 Distribution of CO Concentrations Using Adjusted
Volume 10 for Sources that Emit Less Than 100
tons/year 43
20 Distribution of Concentration/Emission for TSP 46
21 Distribution of Concentration/Emission for S02 47
22 Distribution of Concentration/Emission for N02 48
23 Distribution of Concentration/Emission for CO 49
24 Distribution of Concentration/Emission for Pb 50
25 Distribution of Concentration/Emissions for CO,
S02, and TSP Combined 51
26 Depiction of Plume Height in Complex Terrain, as in
the Valley Model; h is the Height of the Plume
at Final Rise Above Ground for the Unstable and
Neutral Cases and Above Stack Base for the Stable
Cases 60
27 Number of Current Modifications Subject to PSD
Versus De Minimis Emission rates 65
28 Number of Total Modifications at Major Sources
Versus De Minimis Emission Rates 66
-------
TABLES
Number Page
1 De Minimis Emission Levels 3
2 Ambient Air Quality De Minimis Levels 4
3 Modified Sources used for PSD De Minimis Analysis 13
4 Selected Percentile Rankings for Criteria Pollutant
Emissions from all Sources used in the Modeling
Analysis 31
5 Selected Percentile Rankings for Criteria Pollutant
Emissions from Sources which Emit Less than 100
tons/yr 31
6 Effective Stack Heights Calculated by the PTMAX
Model ' 39
7 Pollutant Concentrations 44
8 Emission Rates Associated with Selected Air
Quality Levels 52
9 Emission Rates Associated with Selected Air
Quality Levels Using Several Techniques 53
10 Point Source Input Data for Regional Air Quality
Analysis 55
11 Results of Urbanwide Area Modeling of Selected De
Minimis Levels 54
12 Estimated Air Quality Impacts 56
13 Estimated Distance at Which 1 ug/m3 Maximum 24-h
Concentration is Predicted to Occur for a 40-ton/
yr Change with Worst Case Meteorology 57
14 Estimated Distance at Which 1 ug/m3 Maximum 24-h
Concentration is Predicted to Occur Under Any
Meteorological Condition 57
15 Estimated Distance at Which 1 ug/m3 Maximum 24-h
Concentration is Predicted to Occur for a 40-
ton/yr Change in Elevated Terrain Using Valley
Model 59
vii
-------
Number
16 Effects of Proposed Regulations on Modifications
Reviewed under Current Regulations 61
-------
ACKNOWLEDGMENT
This report was prepared for the U.S. Environmental Protec-
tion Agency, Control Programs Development Division, Research
Triangle Park, North Carolina, by PEDCo Environmental, Inc.,
Cincinnati, Ohio.
The project was directed by Mr. William Kemner and managed
by Mr. David bunbar. Principal authors were Mr. David Dunbar,
Ms. Barbara Blagun, and Dr. Jeff Smith.
Mr. James Weigold was the Project Officer for U.S. EPA, and
his guidance and cooperation were greatly appreciated. The
authors thank Messrs. Gary McCutchen and Warren Peters of EPA
for their cooperation and assistance in completing this effort.
-------
-------
SECTION 1
INTRODUCTION
In 1974, EPA promulgated regulations to prevent emissions
of sulfur dioxide (S02) and particulate matter (PM) from sig-
nificantly deteriorating air quality in areas where the air
quality concentrations were lower than the applicable National
Ambient Air Quality Standard (NAAQS). These regulations pre-
vented the construction of certain categories of new or modified
sources unless a permit had been issued which indicated that the
proposed source would apply Best Available Control Technology
(BACT) for S02 and PM and that the emissions of S02 and PM would
not cause significant deterioration of air quality.
On August 7, 1977, the President signed the Clean Air Act
Amendments of 1977 into law. These amendments established a new
set" of requirements for the prevention of significant deteriora-
tion (PSD). These new requirements follow the basic outline of
the 1974 regulations but are more elaborate and in some cases
more stringent. In response to the 1977 Amendments, EPA promul-
gated regulations on June 19, 1978, to amend the 1974 regulations
to make them consistent with requirements of the Clean Air Act.
1.1 PROPOSED PSD REGULATIONS
In response to the June 19, 1978, PSD regulations, many
industrial and environmental groups petitioned the United States
Court of Appeals for the District of Columbia circuit to review
the substantive provisions of the June 19, 1978, PSD regulations.
On June 18, 1979, the court issued a decision that upheld some of
those provisions and overturned others. (Alabama Power Company
v. Costle, 13 ERG 1225). In its opinion, the court summarized
its rulings and indicated that it would provide a complete com-
prehensive opinion at a later date. Based on the June 1979 court
decision, EPA proposed on September 5, 1979, to revise the
June 19, 1978, regulations to comply with the court's decision.
The final court decision was issued on December 14, 1979.
One of the major elements of the proposed regulations
(September 5, 1979) is the revision of the definition of "major
modification." Under the June 19 regulations a modification is
"major" if its potential emission increases would equal or exceed
the applicable 100/250 ton threshold. The court rejected this
approach. It held that a change in a major source is subject to
review if it results in a net increase in the source's potential
to emit after all contemporaneous emission reductions at the
source are considered.
-------
1.2 DE MINIMIS CONCEPT
In Alabama Power, the court provided that EPA may exempt
from review those situations determined to be de minimis. Speci-
fically, the court stated:
The Agency does possess authority, inherent in
the statutory scheme, to overlook circumstances
that in context fairly may be considered de
minimis.
The court spoke to EPA's capability to exempt modifications with
small net increases and to permit proposed sources (new or modi-
fied) to avoid BACT review and the ambient monitoring require-
ments through the application of de minimis thresholds for those
pollutants emitted from a source that would otherwise be subject
to review.
The September 5 regulations proposed to exercise the above
authority by establishing a pollutant-specific exemption system
that excludes or limits review of proposed sources having emis-
sion levels or air quality impacts below certain values. Table 1
contains for each pollutant regulated under the Act an emission
cutoff that would be considered to cause an insignificant or de
minimis air quality impact. The values proposed in Table 1 have
two principal uses; the first would be to show that a net in-
crease from a modification would be de minimis for all pollutants
for which the source is major, and the second would be to limit
the pollutants (from the source already subject to review) to
which BACT must be applied or for which an air quality analysis
must be submitted.
Table 2 proposes certain air quality concentrations which
are used to limit the air quality review of certain pollutants
that would have emission levels greater than those in Table 1.
In order to apply Table 2, a source must use a preliminary
screening technique to determine if its air. quality impact would
exceed the acceptable de minimis levels. The screening technique
is set forth in Guidelines for Air Quality Maintenance Planning
and Analysis, Volume 10, (Revised): Procedures for Evaluating
Air Quality Impact of New Stationary Sources (October 1977 ).1 If
a source's impact is expected to exceed the de minimis levels,
based on the Volume 10 approach, a source may elect to use a more
sophisticated modeling analysis. Upon showing that the antici-
pated air quality impact would be less than significant (i.e.,
less than the air quality concentrations in Table 2), a major new
or modified source would not be required to conduct a detailed
ambient impact assessment with respect to PSD for that pollutant.
Therefore, such a source would not be required to perform an ana-
lysis of its impact on the increments and NAAQS. Additionally,
the source would not be required to analyze its effect on soils,
vegetation, and visibility, nor would it be required to conduct
any ambient air quality monitoring for those pollutants with
predicted concentrations less than those in Table 2.
-------
TABLE 1. DE MINIMIS EMISSION LEVELS
Pollutant
Carbon monoxide
Nitrogen dioxide
Total suspended particulates
Sulfur dioxide
Ozone (VOC)
Lead
Mercury
Beryllium
Asbestos
Fluorides
Sulfuric acid mist
Vinyl chloride
Total reduced sulfur
Hydrogen sulfide
Methyl mercaptan
Dimethyl sulfide
Dimethyl disulfide
Reduced sulfur compounds
Hydrogen sulfide
Carbon disulfide
Carbonyl sulfide
Tons/yr
100
10
10
10
10
1
0.2
0.004
1
0.02
1
1
1
1
1
1
1
10
10
-------
TABLE 2. AMBIENT AIR QUALITY DE MINIMIS LEVELS
Pollutant'
Carbon monoxide
Nitrogen dioxide
Total suspended particulates
Sulfur dioxide
Lead
Mercury
Beryllium
Asbestos
Fluorides
Sulfuric acid mist
Vinyl chloride
Total reduced sulfur
Hydrogen sulfide
Methyl mercaptan
Dimethyl sulfide
Dimethyl disulfide
Reduced sulfur compounds
Hydrogen sulfide
Carbon disulfide
Carbonyl sulfide
Level, pg/m3
500, 8-h avg
1, annual avg
5, 24-h avg
5, 24-h avg
0.03, 3-mo avg
0.10, 24-h avg
0.005, 24-h avg
1, l~h avg -
0.01, 24-h avg
1, 24-h avg
1, max value
1, 1-h avg
0.5, 1-h avg
0.5, 1-h avg
2, 1-h avg
1, l~h avg
200, 1-h avg
100, 1-h avg
No de minimi's air quality level has been proposed for ozone;
any net increase of 10 tons/yr of VOC subject to PSD would re-
quire an ambient impact analysis, including the gathering of air
quality data.
The proposed regulations indicated that the values in Tables
1 and 2 were not supported by extensive analysis. Therefore, the
following analysis was undertaken to provide further guidance and
insight into the selection of the de minimis levels to be used in
the final promulgation of the PSD regulations.
-------
SECTION 2
PURPOSE OF THE ANALYSIS
The purpose of this analysis was to provide technical sup-
port, guidance, and insight for the selection of de minimis
levels to be used in the promulgation of the final PSD regula-
tions.
2.1 DETERMINATION OF AIR QUALITY IMPACTS OF SOURCES THAT HAVE
RECEIVED PERMITS TO DATE
Since the de minimis concept is generally based on the air
quality impact of a source, it is important to determine what, the
air quality impacts might be for sources that would be affected
by the de minimis concept. A source, based on its emissions, may
be of regulatory concern, but it may have an insignificant impact
on air quality and thus have little or no need for preconstruc-
tion review as it relates to air quality management programs.
Therefore, a major purpose of this analysis is to determine the
air quality impacts of sources subject to PSD review under the
current (June 19, 1978) regulations. Many of the sources that
are subject to the current regulations (because of their source
configuration, type of emissions, dispersion characteristics, and
the particular areas where they plan to locate) may have air
quality impacts that would be insignificant in terms of the PSD
increments or the NAAQS.
Since over 600 PSD permits have been issued to date for a
variety of new and modified sources, it seemed appropriate to
review the air quality impacts of these sources if they were
provided and to calculate the air quality impacts if they were
not provided. A review of these 600 permits would yield a real-
istic estimate of the range or distribution of air quality im-
pacts that would be associated with sources affected by the de
minimis concept proposed on September 5, 1979.
Analysis of this distribution could be used to analyze
selected air quality concentrations for determining the impact of
the de minimis concept, in terms of both the environmental and
adminTstrative impacts.
-------
2.2 DETERMINATION OF EMISSION LEVELS ASSOCIATED WITH CERTAIN
AIR QUALITY IMPACTS
In the September 5, 1979, PSD regulations, certain de
minimis emission levels associated with specific air quality con-
centrations were proposed for determining whether a source with
small net increases in pollutant-specific emissions would be
subject to PSD review. At the time of the proposal, it was
stated that an analysis of the proposed emission levels and air
quality de minimis concentrations would be provided later. A key
aspect of this analysis is the relationship of the air quality
impact associated with the emission levels proposed for each
pollutant. Because this relationship is basically empirical,
realistic data are needed to determine this relationship. The
best available data set is the more than 600 PSD permits that had
been issued from April 1, 1978, to November 1, 1979. The emis-
sions and air quality data from these permitted sources were used
to determine the relationship of predicted air quality concentra-
tion (x) to mass emission rate (Q). The use of these available
data made it possible to characterize the empirical relationship
(X/Q) f°r SO2, nitrogen oxides (NO ), carbon monoxide (CO), total
suspended particulates (TSP), and^Lead (Pb). An empirical rela-
tionship could not be developed for hydrocarbons (HC) or volatile
organic compounds (VOC) and ozone (O3) because of a lack of a
recommended dispersion model. Five types of ozone prediction
methods are currently available. These models vary from simple
algebraic relationships to sophisticated numerical models. In
general, the simple methods tend to ignore or treat superficially
many atmospheric processes that affect the formation of ozone.
On the other hand, the sophisticated model treats these processes
in such detail that considerable species-specific emissions data
are needed as input. Additionally, all these models are more for
a regional than for a specific individual source application.
It should be noted that this analysis centered on the cri-
teria pollutants since very little, if any, data exist regarding
the air quality/emission relationships for noncriteria pollu-
tants. Only limited emissions data on noncriteria pollutants
existed in the permit files, and no data were available on the
associated air quality impact of these noncriteria pollutants.
As a result, the de minimis levels for the noncriteria pollutants
may have to be set on the basis of a percentage of the applicable
emission standard for these pollutants; however, the relation-
ships developed for some of the criteria pollutants could be used
to obtain a relative indication of the associated air quality/
emissions relationship that might exist for the noncriteria
pollutants.
Since the first test for exemption under the de minimis con-
cept is an emission level, a major purpose of this" analysis was
to determine the emission rates associated with specified air
quality concentrations. Given the percentage of sources that
would have a specific air quality impact at a given emission
rate, one can determine the average emission rate (based on a
-------
of sources) that would cause a significant air quality
impact, as defined by the de minimis levels.
2.3 NUMBER OF SOURCES TO BE AFFECTED BY DE MINIMIS CONCEPT
Another purpose of this analysis was to determine the number
of sources that would be affected under the proposed de minimis
levels and the number that might be affected under other selected
de minimis levels. Although the court did limit EPA's discretion
in formulating the de minimis levels by stating that a cost-ef-
fective rationale wouldnotbe appropriate, the extent of the
impact of the .proposed versus any final de minimis levels is
important in determining the number of reviews that would be
required as a result of the final promulgation. Since some
modifications not subject to the current PSD regulations would be
subject to the proposed regulations, the number of additional re-
views that would be required was needed in order to determine the
possible workload that would be incurred by certain .specified de
minimis levels.
If the number of additional reviews is substantial, this
could create serious problems in terms of manpower available for
reviews and time required to obtain a permit. Both problems
could affect the quality of the review, the overall costs of
obtaining a PSD permit, and delays in construction.
2.4 URBANWIDE AIR QUALITY IMPACT DUE TO DE MINIMIS CHANGES IN
EMISSIONS
Another purpose of this analysis was to determine the over-
all air quality impact for an area if all the major sources
within the area emitting over 100 or 250 tons/yr proposed to
modify.
While one source may modify its facility and not cause a
significant air quality impact, a number of sources making such a
change could cause a significant impact. If the sources were
located near to each other, the cumulative air quality impact
could consume a significant amount of the increment. Since the
extent of the impact is directly proportional to the number of
sources and their relative proximity to each other, it is impor-
tant to determine the potential air quality impact from a number
of existing sources making de minimis changes in emissions. A
set of existing source data was used to determine the impact of
modifying by a de minimis amount so that the estimate obtained
would represent that which would be expected to occur for a given
set of sources. By using actual sources with specific locations,
one can obtain a reasonable and realistic assessment of the
overall urbanwide air quality impact of the de minimis concept.
-------
2.5 IMPACT WITH RESPECT TO CLASS I AREA
The proposed regulations place certain limitations on the
use of the air quality de minimis levels (Table 2). The exemp-
tion from PSD review because of de minimis levels does not apply
to major construction that would be located in a nonattainment
area or that would adversely impact a Class I area. Therefore,
the final purpose of this analysis was to determine the maximum
distance from a Class I area where a source making a de minimis
change in emissions 'would be expected to have a 1 ug/m3 impact
(defined as significant impact on a Class I area) averaged over a
24-h period. This determination would provide insight into the
relative distance from the Class I area beyond which de minimis
changes would not have a significant impact upon the area.
-------
SECTION 3
METHODOLOGY
The following methodology was used in the analysis of the de
minimis concept proposed by EPA on September 5, 1979.
3.1 SELECTION OF SOURCES TO BE EVALUATED
Tlie data base for this analysis is the approximately 600 PSD
permits issued between April 1, 1978, and November 1, 1979. This
data base represented information on a variety of new and modi-
fied sources in several different source categories and was
obtained by reviewing the permit files in EPA Regional Offices
III-X. For each permit, several pieces of information were
summarized on the survey form shown in Figure 1.
Since the analysis of the de minimis concept is centered on
the emissions and the associated air quality impact for each
source^ the data on both were reviewed for each permit if avail-
able. The sources were separated into three categories:
IV- Those for which dispersion modeling -had been conducted
and the results presented
2 .(• Those for which modeling had not been conducted, but
;; which had sufficient emissions and stack data to con-
?,"-- duct dispersion modeling
3y: Those for which no modeling had been conducted and for
•f:, which insufficient data were available for additional
"-.; modeling.
Those ;in the first two categories were further reviewed to deter-
mine whether they were new or modified sources and whether they
emitted; more than 100 tons/yr. This further categorization was
necessary to evaluate the sources that would be most affected by
the dg;.minimis levels.
S§.nce new and modified sources emitting more than 100 or 250
tons/yrLare subject to PSD review, the de minimis levels are only
used to determine the pollutants for which BACT review is re-
quired^. However/ the modifications emitting less than 100 tons/
yr at major sources are affected by the de minimis levels in that
these ;levels determine whether the modifications must obtain a
PSD permit. Therefore, the major focus of this analysis was on
tnose modifications of less than 100 tons/yr. The permit data
-------
SEPTEMBER 5, 1979 PSD ASSESSMENT
Source Type/Size:_
Name/Mail Add.:
Page of_
_EPA RO PEI NO.
_C oun ty
at UTM
Located In: ATT/NA area of AQCR No._
Determination is: CONDITIONAL/FINAL/PENDING for NEU/MODIFIED/RECONSTRUCTED/REPLACEMENT Source.
Key Dates: Application-Reed. , Completed •, Determination-Proposed , Final,
_and impacts ATT/NA area
"A"-IMISSIONS - ton/yr
"B"-PERHIT CONDITIOOflJ .
Affected Facil ities
[Name and Number)
Type*
Max/PC**
capaci ty
At capacity
Uncontr.
Contr.
Allowable under:
BACT
SIP
NSPS or
NESHAPS
LAER or
other
specify
Oper.
hours
Fuel
type/
amount
Materials
type/
amount
"C"- FUGITIVE SOURCES
(Name and Number)
Type*
Dust-t
Uncontr.
Dn/yr
Contr.
Emissions
Uncontr.
-ton/yr
Contr.
"D"- AIR QUALITY IMPACT 1 "E"- MONITORING NETWORK
Type*
^ax. Cone.
ug/m
km from
source
Start Date
Type*
No. Monitors
* Specify pollutant - use PM, SOX> NOX» HC, CO, Pb for NAAQS; Hg, Be. As, VC1 for NESHAPS; SA, TRS, RCS, Fl for NSPS.
*•* If source operates at other than maximum capacity due to permit conditions - circle "PC", insert details at "B"
and complete; "A" based on "permit capacity.
Figure 1. Summary form for September 5, 1979, proposed PSD regulations.
-------
Page of
"F"- CONTROL EQUIPMENT
CYCLONES:, ' Type ;. Tubes: No. , D in.. L = ft; AR- = in. H20; Inlet Vel. = ft/s;
(Source No. ) Volume = SCFM: Eff. = %; Other:_
Type
ESP
(Source No. ) Linear vel. =
FABRIC FILTER: Type
_; Total plate area =
ft/s; Volume
Cleaning Mode:
ft2; SCA =
_SCFM; Eff. %; Other_
Fabric:
.2
ftVlOOO ACFM; L/W ratio
AP =
Jn.
(Source No. ) Filter vel. (A/C) = ft/s; Volume =
; L/G =
SCFM: Eff. %; Other
SCRUBBER:
Type_
(Source No. ) Volume =
AFTERBURNER: Temp. =
OTHER TYPE CONTROLS/REMARKS/ETC.
gal/1000 ACFM; AP = in. H20; Liquid used_
SCFM: Eff. = %•> Other
3F; Residence Time = s; L/D: Volume =
SCFM; Eff.
"G"- CONTROL COSTS
"H"-. EMISSIONS CHANGE SINCE 8/7/'77-tQn/vr
"I"- STACK DATA
Source
No.
System Cost-$xlO
Cap:
Oper:
Source
Type*
Increase
Decrease
Net
Source
No.
No. of
stacks
Diam.
-ft.
Ht.
-ft.
Temp.
-°F
Exit vel .
-fps
NOTES, REMARKS, ETC.
"J"- FUEL USE
Type_
HV
% A;
Annual Use Rate:.
_BTU/lb-ga.l-CF
"K"- AIR QUALITY MODELING
Pre-screening
Detailed/Type
_Screening_
Figure 1 (continued)
-------
did not indicate whether the sources were modifications to major
sources (i.e., more than 100 or 250 tons/yr) because it did not
matter whether the existing source was major or not under the
current regulations. It was therefore assumed that all modifica-
tions that had previously received permits were modifications to
existing major sources and the de minimis levels would determine
their applicability with respect to PSD review.
3.2 ENGINEERING ANALYSIS
An engineering analysis of the selected modified sources to
be used in evaluation of specific de minimis levels was performed
to ensure that they were typical of those likely to be modified
in the future. Table 3 lists the number of sources in each of 32
industrial categories that either had provided modeling results
or had sufficient data available to conduct additional modeling
and the number of sources considered to be typical. As shown in
the table, the sample population of modified sources consisted of
57 sources, which represented 20 of the 32 major sources cate-
gories that had been issued PSD permits from April 1, 1978, to
November 1, 1979. For each of the 57 sources, the following
characteristics were reviewed to determine whether a source was
typical: processes, capacity, emissions, control devices, stack
parameters, dispersion modeling results (if available), operating
practices, and fuels and feedstocks. The review revealed that
eight sources were atypical; that is, the modification that had
been permitted was not one typical of the type of modification
that would be expected for the source category in general, or the
feed stock or fuel used by the modification was not typical of
the fuel or feed stock generally expected to be used by the
source category. For each of the eight sources, an attempt was
made to substitute permit data from new sources that had been
issued PSD permits from April 1, 1978, to November 1, 1979, and
that would be more typical of the sources within that particular
source category.
In general, substitutions were made so that typical unit
operations for a particular source category would be reflected.
This is why substitutions were made in the sulfur recovery,
secondary metal, chemical process, hydrofluoric (HF) acid, and
fuel conversion source categories. In the sulfur recovery cate-
gory, the modified source was a gas-sweetening process at a
refinery rather than a sulfur recovery operation. Therefore, a
new Glaus plant was substituted for this modified source for this
analysis. In the secondary metal category, secondary aluminium
and lead plants were substituted for a modified grinding opera-
tion. The HF modification was deleted, because it consisted of
adding an alkylation unit rather than a unit operation typical of
acid production. No substitution was made for the HF source as
no other permits had been issued for that source category. The
chemical process and fuel conversion modifications were deleted
because they were merely adding a boiler rather than really
modifying the process, and a typical new multiple-point-source
12
-------
if
TABLE 3. MODIFIED SOURCES USED FOR PSD DE MINIMIS ANALYSIS
Industry category
1. Fossil-fuel-fired steam generator
2. Coal cleaning
3. Kraft pulp mills
4. Portland cement
5. Primary zinc smelting
6. Iron and steel
7. Primary aluminum smelting
8. Primary copper smelting
9. Municipal incinerator
10. HF
11. H2S04
12. Petroleum refinery
13. Lime plant
14. Nitric acid
15. Phosphate rock
16. Coke ovens
17. Sulfur recovery
18. Carbon black
19. Primary lead smelting
20. Secondary metal
21. Chemical process
22. Industrial boiler
23. Petroleum storage
24. Taconite ore
25. Glass fiber
26. Charcoal production
27. Fuel conversion
28. Sintering
29. Asphalt plant
30. Rock crushing
31. Natural gas compression
32. Oil and gas extraction
Total
Number of
sources
1
1
5
2
-
4
-
-
2
1
-
12
2
-
-
-
1
2
-
2
1
4
-
-
4
-
1
-
6
3
2
1
57
Number of
typical sources
0
1
5
2
-
4
-
-
1
0
-
12
2
-
-
-
0
2
-
1
0
2
-
-
4
-
0
-
6
3
2
1
48
13
-------
chemical plant and a new coal gasification operation were sub-
stituted in their place.
The other principal reason for substitutions of new plants
for modified plants was the use of atypical fuels or feed stocks.
A new coal-fired steam generator was substituted for a modified
bagasse/oil-fired boiler. Similarly, in the industrial boiler
category, a new coal-fired unit was substituted for a modified
wood waste boiler. In addition, a new municipal incinerator was
deleted because it was primarily a liquid waste incinerator and
no substitution was made because no other municipal incinerator
had been issued a permit during the time period analyzed.
3.3 MODEL SELECTION
In Alabama Power, the court.recognized that modeling tech-
niques would be the principal device relied upon for the projec-
tion of the air quality impact from a regulated source. There-
fore, model selection is an important step in the process of
determining de minimis exemptions. Although the modeling tech-
niques set forth in the Guidelines for Air Quality Maintenance
Planning and Analysis, Volume 10 (Revised): Procedures for
Evaluating Air Quality Impact of New Stationary Sources1formed
the basis of the de minimis analysis (since it was referenced in
the proposed regulations), additional modeling techniques were
used to provide a check on the results from the Volume 10 ap-
proach.
The preliminary model selection consisted of an initial
evaluation of the three levels of air quality analyses (specified
in Volume 10)—namely, simple screening, basic modeling, and
refined modeling.
The simple screening technique utilizes some of the Gaussian
dispersion equations outlined in the Workbook of Atmospheric Dis-
persion Estimates2, which assumes flat terrain and no aerodynamic
downwash. '
The second level of modeling requires the use of simple
computer programs—either a series of calculations set forth in
Volume 10 performed on a pocket or desktop calculator or the
basic EPA (PTMAX) program, which is available through the UNAMAP
series. The models selected for use from UNAMAP are programed
versions of the Gaussian dispersion equations. PTMAX is an in-
teractive program for analyzing the maximum short-term concentra-
tions from a single-point source as a function of stability and
windspeed.
In this model, the final plume height or effective stack
height (i.e., height of the plume centerline when it becomes
essentially level) is used for each computation and is estimated
by the Briggs equation. The model assumes flat terrain, unlim-
ited mixing heights, and no aerodynamic downwash or background
14
-------
coijGeiilafations. PTMAX calculates a maximum 1-hour (1-h) concen-
tj^atTion for a given inert nonreactive pollutant under "worst
ease" meteorological conditions, which can then be transformed to
a 3-h, 8-h, or 24-h value by multiplying the 1-h concentration by
0.9, 0.7, or 0.4,1 respectively. By use of PTMAX, many stabili-
ty-windspeed combinations can be evaluated rather quickly and the
"worst case" conditions determined. This method, however, is
still considered detailed screening, since it uses only limited
meteorological input.
Phase three or refined modeling, which is beyond the scope
of Volume 10, is defined in the Air Quality Modeling Guidelines
(April 1978).3 The preferred model simulates atmospheric trans-
port and dispersion in the area of interest; it considers both
the availability of source and meteorological data and the local
topography and plant configuration. As stated in the Guidelines,
however, "there is no single model capable of properly addressing
all conceivable situations."3 A single-source Gaussian disper-
sion model (CRSTER) is recommended when no significant terrain or
meteorological complexities are encountered. For multi-source
situations, the Gaussian-Plume Multiple-Source Air Quality Algo-
rithm (RAM) is suggested in the Air Quality Modeling Guideline.3
This model has both an urban and rural version, but only the
urban version was used in this analysis. The choice of locations
for receptor sites significantly affects the evaluation of source
impact. When the CRSTER is used, an appropriate receptor field
must be designated by the user; RAM, on the other hand, has a
significant point receptor option by which the program itself
selects the receptor sites.
Of the three levels of modeling available, the first was
eliminated because the de minimis analysis demands more than
simple screening techniques. The second was evaluated and was
considered to be acceptable because it provides more realistic
estimates than the first. Additionally, it is the method refer-
enced for obtaining air quality impact exemptions in the
September 5, 1979, proposed regulations. Use of the PTMAX com-
puter model instead of a pocket or desk calculator (procedure in
Volume 10) eliminates possible calculation errors and reduces the
time required for obtaining concentration estimations. The
refined models require a more extensive data base—for example,
complete sets of surface and upper air meteorological data,
detailed topographical data, and a general idea of population
density for urban/rural determinations. Although the phase three
technique provides the most refined estimate, it also requires
considerably more detailed information and resources. Because of
time, economic, and data constraints, it had limited use in this
analysis; however, it should be pointed out that the more refined
modeling results were used in the analysis in those cases where
they were provided in the permit files.
A cross section of approximately 50 sources was selected for
comparison of different concentration estimates by use of a total
set of meteorological parameters (stability class, windspeed,
15
-------
temperature, etc.) for one entire year with concentration aver-
aged every 24 hours vs. a transformed 1-h concentration calcu-
lated with only one stability class, windspeed, and ambient tem-
perature per run. Because of the time required and topographical
information needed for selecting a proper receptor grid for
CRSTER, EPA's urban RAM model with the significant point receptor
option was selected for the comparison. These runs produced 24-h
maximum values, which could then be compared to the PTMAX 1-h
maximum values, transformed to 24-h averaged values by the pro-
cedures in Volume 10. The results of this analysis indicated
that there was a relatively good correlation between the urban
RAM and PTMAX results and that for all pollutants except S02 the
concentrations predicted by PTMAX were slightly lower than those
obtained from RAM. For S02, some of the PTMAX results were
lower, while others were slightly higher. Therefore, for the
purposes of this analysis the PTMAX results were considered
comparable to those that could be obtained by use of a more
refined modeling technique.
3.4 PREDICTION OF AIR QUALITY IMPACTS BY USE OF SELECTED MODELS
Air quality impacts of typical sources that had received a
PSD" permit were predicted with the PTMAX air quality simulation
models described in Section 3.8. As previously stated, other
more refined models were used but only to cross-check the results
obtained with PTMAX. Input data consisted of actual plant con-
figurations from the PSD permits issued from April 1, 1978, to
November 1, 1979. The selection of this sample population for
the analysis provided an indication of air quality impacts of
sources expected to receive permits over the next several years,
assuming similar industrial trends.
First, source air quality impacts were assessed by use of
the second phase of the Volume 10 screening techniques for a
sample population consisting of actual data from permits issued
in eight EPA Regional Offices from April 1, 1978, to November 1,
1979. Specific parameters from the PSD permit files were input
into PTMAX, which provided an analysis of the 1-h maximum concen-
tration from a point source in flat terrain as a function of
stability and windspeed. The PTMAX program was run for all
windspeed-stability combinations, and the highest estimated con-
centration was designated as the maximum 1-h concentration. This
value was then converted to an estimated maximum concentration
for an 8-h averaging time for CO (1-h average x 0.7) and 24-h for
TSP and S02 (1-h average x 0.4) by using the respective lh/8h and
lh/24h ratios in Volume 10, which are based on general experience
with elevated point sources. The ratio of the 1-h average to the
annual average for NO (0.14) was calculated from refined model-
ing data compiled from modeling results from several PSD permits
and previous EPA modeling studies.4
The Volume 10 results were adjusted, and the next highest
valid data point from the PTMAX results was selected to elimi-
nate any estimated concentrations that the -program designated
16
-------
should be used with caution because either (1) the height of the
plume was sufficient to require extreme caution in the interpre-
tation of the computation since the particular windspeed-stabili-
ty combination may no longer exist at that height, or (2) the
distance to the point of maximum concentration was so great that
the same stability was not likely to persist long enough for the
plume to travel that far.
The PTMAX model calculates the final plume rise by the
Briggs plume rise equation. Calculations for plume rise may not
necessarily be valid within the first few hundred meters horizon-
tally from the stack, since the plume rise from a stack may occur
some distance downwind at high windspeeds. Therefore, for this
analysis, concentrations occurring at less than 200 m from the
source were eliminated, and the next highest concentration beyond
200 m was selected.
In addition, a method (other than those outlined in Volume
10) was examined to provide a relative comparison of the pre-
dicted concentrations. The PTMAX model runs were evaluated at
neutral case conditions, Class D stability, and a windspeed of 5
m/s.
Meteorological input for the PTMAX model consisted of one
windspeed, stability class, and ambient temperature per run. To
cross-check the effect of a total set of meteorological param-
eters, a more sophisticated model was run with one year of repre-
sentative meteorological data for a sample of 50 sources. The
EPA urban RAM model was selected because of its significant point
receptor option. Land use and population density estimates were
calculated to determine the applicability of the RAM urban model
for particular sources in specific locations. The source's
geographic location was classified as urban if land use types6
Ilf I2, Clf R2, and R3 comprised >^50 percent of the surrounding
area or if the population density of the area was >750 people/
km2.
An alternate analysis was developed for the assessment of Pb
because of the inherent problems with deposition and the lack of
Pb sources in the PSD permit files. Because deposition is not
one of the factors considered in the PTMAX model, valid air
quality impacts could not be determined for Pb by this technique.
The EPA modeling results for the Standards Support and Environ-
mental Impact Statement (SSEIS) on Lead5 were used as a supple-
mentary data base for de minimis level evaluation of Pb. A
climatological dispersion model with the ability to accommodate
the deposition phenomenon was used in that report to evaluate the
air quality impacts of various stationary Pb sources. Because
plant configuration is an important factor in deposition, extra-
polation of the SSEIS modeling results to actual situations could
lead to erroneous estimates of Pb concentrations, especially in
those cases where the source configuration would differ signifi-
cantly from the model plants used in the SSEIS analysis. How-
ever, since these model plants are representative of typical
17
-------
plants that may be constructed, the results obtained from this
analysis can be used to determine the average impact from a
variety of Pb sources.
3.5 RELATIONSHIP BETWEEN SPECIFIC EMISSION LEVELS AND AIR
QUALITY CONCENTRATIONS
A relationship between concentration and emission rate was
developed to evaluate selected de minimis cutoffs proposed by
EPA; specific air quality concentrations for criteria pollutants
were selected; and corresponding emission levels were calculated
by several different techniques, as described below. Since the
air quality increments represent a given percentage of the NAAQS,
it was determined that a similar percent approach would be used
to determine the level of air quality that would be considered
insignificant for the purpose of avoiding a detailed PSD review
when existing sources are modified. Therefore, emission rates
were calculated for various air quality levels that represented
specific percentages of the NAAQS.
One way of relating emissions to air quality was to develop
a ratio of the concentration (x) to the emission rate (Q) for
each individual source used in the analysis. These ratios (re-
ferred to in Volume 10) were generated as useful tools to enable
one to readily calculate an emission rate for a source or a group
of sources, given a specific air quality value, or to calculate
an air quality concentration, given a specific emission rate.
Both pollutant-specific and combined pollutant x/Q distributions
were generated for the various modeling techniques described in
Sections 3.3 and 3.4. The x/Q ratio represents an incremental,
normalized, air quality impact. It is normalized in the sense
that the air quality concentration for each source is divided by
the specific emission rate that produced the particular air
quality impact, and thus permits the air quality concentrations
to be compared on a common basis. This ratio represents the
expected incremental change in ambient concentration (pg/m3) due
to a unit change in emissions of 1 ton/yr (0.02877 grams per
second); thus the x/Q ratio indicates a source's degradation of
ground-level air quality for any emitted inert, nonreactive, and
nonsettling pollutant. The x/Q value can be multiplied by a re-
spective Q value to determine the air. quality impact of that
particular emission rate, or its inverse can be multiplied by a
particular air quality value to determine the Q that would con-
tribute that concentration to the environment. These x/Q values
consider several source parameters in the relationship between
concentration and emissi-on rate since each x/Q has a different
combination of emission rate, effective stack height (H'), and
meteorological parameters (stability and windspeed) factored into
its determination.
The magnitude of the x/Q value corresponds to the dispersive
nature of the source in question. Since the de minimis emission
levels will be used by a variety of sources, it was determined
18
-------
that ,j^n average or typical situation should be used in developing
the relationship between emissions and air quality to obtain a
realistic and representative estimate of the emission level that
would produce a given air quality impact. Because a number of
different sources were used in this analysis and a distribution
of the ratios of air quality to emissions was developed, it was
determined that the 50th percentile values representing the
average of the distributions would be used. In the selection
of a 50th percentile x/Q value from the distributions, the aver-
age dispersive characteristics of the sample (both in terms of
the plant's emission characteristics and certain meteorological
conditions) are factored into the analysis. Therefore, the
values generated by this analysis tend to reflect representative,
realistic conditions.
An alternate technique was developed for comparative pur-
poses. The EPA PTDIS Gaussian dispersion model from the UNAMAP
series, which estimates short-term concentrations directly down-
wind of a point source, was selected to predict a concentration-
mass emission rate relationship. An option in the model allows
the input of an effective stack height instead of the separate
physical stack heights, stack exit velocities, stack gas tempera-
tures, and stack diameters (as required by the PTMAX model). By
using the mean effective stack height of the sample population as
the input value, all the mean values of the stack parameters are
factored into the model as they relate to a specific effective
stack height instead of being considered as separate nonassoci-
ated average values. For example, the mean from the distribution
of each parameter could yield a physical stack height of 18 m, a
stack velocity of 20 m/s, a stack temperature of 400°K and a
stack diameter of 1.5 m. If these average values are used, the
calculated H' would be 30 m. However, if the mean of the H' dis-
tribution was used, the H' would be 40 m. Using the average of
the various stack parameters to construct a composite stack could
produce an unrealistic stack configuration which in turn could
produce an unrealistic H'. Therefore, the mean of all the cal-
culated H''s was used instead of an H' calculated using a com-
posite of several parameters.
The mean H' is used to calculate the concentrations at
various downwind distances selected by the user. The receptor
grid for the model is narrowed in subsequent runs until the
maximum concentration associated with the H' and the stability
class-windspeed combination is located. Since the concentration
and the mass emission rate are directly correlated in a Gaussian
model, the mass emission rate was input as unity to simplify
calculations. The mass emission rates can be calculated by
dividing the selected de_ minimis air quality levels by the con-
centrations estimated by the PTDIS model. For example, if the
PTDIS-calculated concentration is 0.386 |jg/m3 at Q = 1 ton/yr,
then a de minimis concentration of 7.4 |jg/m3 would have an asso-
ciated Q of 7.4/0.386 or 19 tons/yr.
19
-------
The 50th percentile of the H' distribution that was calcu-
lated by using adjusted Volume 10 PTMAX results described in
Section 3.4 was input into the PTDIS model along with the stabil-
ity class-windspeed combinations used in the PTMAX model and an
average mixing height of 1000 m. Mixing heights were varied from
700 to 1200 m to check the effect that mixing height might have
on the predicted concentration, but no change in predicted con-
centration was noted since H' was significantly less than these
mixing height values. Each combination of stability and wind-
speed was run until a maximum concentration and a corresponding
"worst case" condition were identified. This pollutant concen-
tration associated with each ton of emissions can then be used to
calculate the emissions associated with an air quality value, as
described above. Taking the average effective stack height
associated with the worst case conditions (Volume 10 approach)
and using this value to calculate concentrations with the PTDIS
model (which' also uses the worst case conditions) results in
concentrations and ultimate mass emission rates that are extreme-
ly conservative. Thus a second approach was developed, which
repeats this procedure but uses the mean H' from the H' distribu-
tion of the neutral condition values calculated by PTMAX (Section
3.4). This approach used PTDIS with stability D and a windspeed
of 5 m/s, since this was the combination for which the average H*
was" calculated. These results (Section 4.5) are more realistic
since the original analysis and the PTDIS runs were conducted
under the same neutral meteorological condition.
In the same manner, an H' of 30 m, D stability, and 2 m/s
wind speed were input into PTDIS to duplicate conditions similar
to those originally used by the EPA in the initial de minimis
analysis.
3.6 URBAN AREA IMPACT OF MAJOR SOURCES MAKING DE MINIMIS CHANGES
The effect of a number of sources all making de minimis
changes in a localized area was estimated to determinetheir
composite impact in terms of air quality. The urban version of
EPA's RAM (a Gaussian plume, multiple-source, air quality model)
was used to estimate incremental increases in ground-level con-
centrations due to de minimis level increases. The effect of
those increases was analyzed by inputting data for 37 S02 point-
sources located in a medium-size Midwestern city into the model,
along with 1972 meteorological data for that metropolitan area.
This city was selected because the data on the sources in this
area were readily available and the source configuration which
existed for this city was typical of that which might exist for a
number of urban areas across the United States. The mass emis-
sion rates for all 37 sources were set at the proposed de minimis
levels, and a honeycomb receptor grid option was selected. The
model generated 45 receptors, positioned around the sources, and
estimated the concentrations at these receptors that would result
from all 37 sources making de minimis changes. The worst case
20
-------
day .4^7- m/s winds with 98.8% persistence, 272°K ambient tempera-
ture, D stability, and average mixing height of <500 m) was
selected for this analysis.
3.7 CLASS I AREA PROTECTION
Modeling was conducted to estimate the impact of de minimis
emission rate increases on Class I areas. The EPA PTDIS model
was run at varying stability-windspeed combinations to estimate
the Class I increment consumption associated with several pro-
posed and alternate de minimis emission rate increases. The mean
H' arrived at by the Volume 10 approach, H' = 30 m, was input
into the model along with a 1000 m mixing height. In addition,
the 10th and 90th percentile values (13 and 122 m, respectively)
from the distribution of effective stack heights (calculated by
the Volume 10 approach) were also used. Several stability class-
windspeed combinations were used in calculating the predicted
concentrations with the PTDIS model. Two approaches were used in
calculating the maximum distance from a Class I area where a
source making various de minimis emission changes would have a
predicted maximum 24-h impact of 1 pg/m3. These two approaches
were used to ensure that the worst case conditions in terms of
the" maximum concentration and downwind distance were calculated.
Since many of the Class I areas are located in areas with ele-
vated terrain, the Valley model7 was used to determine if terrain
features would increase the distance from a Class I area where a
source making a de minimis change might locate and still have a
1 |jg/m3 maximum 24-h impact on the Class I area.
21
-------
SECTION 4
RESULTS
4.1 DISTRIBUTION OF EMISSION LEVELS
In order to obtain some indication of the range of emission
changes from the sources for which the analysis of the de minimis
levels was conducted, cumulative frequency distributions by
pollutant were developed. Figures 2 through 9 show cumulative
frequency distributions of the PM, NO , S9 , and CO emission
changes for sources to which either PTMSX dispersion models were
applied^F-or for which previous modeling results were available.
Figures 2 through 5 include all sources; Figures 6 through 9
include only sources having emission changes of less than 100
tons/yr. ;
The cumulative frequency distribution data were plotted on
log probability paper. In most cases, these plots approximated
straight lines (i.e., the data appeared to be lognormally dis-
tributed). Because linear regression analysis of several plots
yielded a correlation coefficient ranging from 0.97 to 0.99,
which supports the lognormal hypothesis, the log-normal distribu-
tion was assumed and the distribution parameters for each data
set were estimated.
In general, the lognormal distributions provided a good fit
to the emissions data in Figures 2 through 5. The data distribu-
tions in Figures 6 through 9 were generally lognormal in the
lower range, but not in the higher range. Note that the plotted
data curves toward an upper limit value. 'This is because the
less than 100 tons/yr emission changes are a subset of the total
amount of emission changes available for analysis, and when a
normal or lognormal distribution is truncated, it becomes asymp-
totic to the level at which the distribution was truncated. The
largest range of emissions (0.05 to 2000 tons/yr) is for SO ; CO
emissions range from 0.1 to 66.0 tons/yr; PM emissions rangeirom
1 to 600 tons/yr; and NO emissions range from 10 to 2500 tons/
yr. x
Table 4 lists the '10th, 50th, and 90th percentile values
from the cumulative frequency distribution for NO , SO , PM, and
CO emission changes from all sources; Table 5 lisrs these values
for sources that have emission changes of less than 100 tons/yr.
The 10th, 50th, and 90th percentiles for sources that have emis-
sion changes less than 100 tons/yr are quite similar for all
four pollutants. Although the values are about 1, 15, and 75
tons/yr, respectively, when all sources are considered, the 50th
22
-------
1000
100
(/)
c
o
f\
u->
o
I—I
oo
10
2 5 10 15 20 30 40 50 60 70 80 85 90 95 98
PERCENTAGE
Figure 2. Distribution of participate matter emissions for all
sources used in the analysis.
23
-------
10,000
1,000
I/I
c
o
oo
o
oo
oo
100
10
5 10 15 20 30 40 50 60 70 80 85 90 95 98
PERCENTAGE
Figure 3. Distribution of nitrogen oxide emissions for all sources
used in the analysis.
24
-------
10,000
1,000
in
E
O
oo
•z.
o
oo
oo
100 —
10
2 5 10 15 20 30 40 50 60 70 80 85 90 95 98
PERCENTAGE
Figure 4. Distribution of sulfur oxide emissions for all sources
; used in the analysis.
25
-------
1000
s-
\
CO
co
o
t—I
co
co
i—i
U4
1
2 5 10 15 20 30 ,40 50 60 70 80 85 90 95 98
PERCENTAGE
Figure 5. Distribution of carbon monoxide emissions for all sources
used in the analysis.
26
-------
01
c:
o
co
z:
o
co
co
LU
0.1
2
5 10 15 20 30 .40 50 60 70 80 85 90 95 98
PERCENTAGE
Figure 6. Distribution of particulate matter emissions for sources
that emit less than 100 tons/yr.
27
-------
.o
•M
o
I—1
CO
0.1
2
5 10 15 20 30 40 50 60 70 80 85 90 95 98
PERCENTAGE
Figure 7. Distribution of nitrogen oxide emissions for
sources that emit less than 100 tons/yr.
28
-------
100
10
CO
E
-O
+->
**
00
•z.
O
1—I
oo
0.1
2 5 10 15 20 30 . 40 50 60 70 80 85 90 95 98
PERCENTAGE
Figure 8. Distribution of sulfur oxide emissions for sources
that emit less than 100 tons/yr.
29
-------
100
10
oo
z:
o
oo
oo
1.0
0.1
2 5 10 15 20 30 40 50 60 70 80 85 90 95 98
PERCENTAGE
Figure 9. Distribution of carbon monoxide emissions for sources
that emit less than 100 tons/yr.
30
-------
TABLE 4. SELECTED PERCENTILE RANKINGS FOR CRITERIA POLLUTANT
EMISSIONS FROM ALL SOURCES USED IN THE MODELING ANALYSIS
Pol lutant
N0x
.S0x
/ PM
CO
Tons/yr
10th
percentile
18
9
1
0.6
50th
percentile
180
200
11.5
30
90th
percentile
1500
3000
80
1000
TABLE 5. SELECTED PERCENTILE RANKINGS FOR CRITERIA POLLUTANT
EMISSIONS FROM SOURCES THAT EMIT LESS THAN 100 tons/yr
Pollutant
N0x
S02
PM
CO
Tonsjyr
10th
percentile
1
0.45 .
1.5
0.29
50th
percentile
29
12
20
10
90th
percentile
80
85
70
55
percentile value ranges from 11.5 tons/yr for PM to 200 tons/yr
for SO .
X.
The 50th percentile value from the cumulative frequency dis-
tribution indicates that half the sources in the distribution
would be greater than a certain value and half would be less. In
the case of the sources making changes of less than 100 tons/yr,
the 50th percentile value would indicate that half the sources in
the sample would make emission changes of 15 tons/yr or less and
half would make changes greater than 15 tons/yr. Likewise the
90th percentile value would indicate that 90 percent of the
sources would be making changes of 75 tons/yr or less and 10 per-
cent would be making changes of greater than 75 tons/yr.
Because major sources that make significant changes in
their emissions (i.e., greater than 100/250 tons/yr) are clearly
subject to PSD review by virtue of the Clean Air Act and the
associated regulations, the major impact of the de minimis levels
31
-------
will be on those major sources that make smaller, less signif-
icant changes in their emissions. In other words, depending upon
where the de minimis levels are set, some sources would make
emission changes that would be below these levels and therefore
not be subject to PSD review. In the design of the de minimis
analysis, a major emphasis is placed on those major sources
making emission changes of less than 100 tons/yr. Therefore,
only those sources that fall into this category were used in the
analysis. If the full range of modified sources had been used,
the assessment as pointed out by the data above would skew the
results towards larger sources that would be basically unaffected
by the de minimis levels in terms of whether they would be sub-
ject to PSD or not. This skewing of the impact toward the larger
sources results from emissions varying significantly at the upper
end of the distribution.
4.2 DISTRIBUTION OF ACTUAL STACK HEIGHTS
Figure 10 illustrates the distribution of actual physical
stack heights used in the modeling studies. It should be noted
that some sources had more than a dozen stacks. As illustrated
in the figure, only 2 percent of all stacks were under 5 m and
only 5 percent were more than 60 m. Half of the stacks ranged
from 10 to 35 m, and 28 percent from 5 to 10 m. Thus, a large
percentage of the stacks for the sources in the analysis were
relatively short, but typical of the kind of sources that would*
be making changes of less than 100 tons/yr as a result of a
modification to the source.
4.3 DISTRIBUTION OF EFFECTIVE STACK HEIGHTS
The PTMAX model uses the Briggs equation to calculate the
final plume rise. The H' is the height of the plume centerline
when it becomes essentially level. Rarely will H' correspond to
the physical stack height. Calculations for plume rise are not
necessarily valid within the first few hundred meters of the
stack since the plume rise from a stack occurs over some distance
downwind. Therefore, for the analysis, concentrations occurring
less than 200 m from the source were eliminated from the data set
(Section 3.4).
Figures 11, 12, and 13 show cumulative frequency distribu-
tions of H' calculated by the Volume 10, Phase II method; by the
adjusted Volume 10, Phase II method (i.e., precautionary values
omitted and next highest value substituted); and at neutral
conditions, D stability and a windspeed of 5 m/s, respectively.
In Figures 14 and 15, the H' data set used for the adjusted PTMAX
runs was subdivided into a distribution of sources with pollutant
emissions of more than 100 tons/yr and less than 100 tons/yr; the
results of these distributions are summarized in Table 6. The
mean H' for sources with Q >100 tons/yr is 75 m; adding this data
set to the sample with Q <100 tons/yr only raised the Q <100
32
-------
1000
100
O)
10
1
2 5 10 15 20 30 40 50 60 70 80 85 90 95 98
PERCENTAGE
Figure 10. Distribution of actual stack heights.
33
-------
1000
100
l/J
s-
_
OJ
10
1
2 5 10 1520 30-4050 60 70 8085 90 95 98
PERCENTAGE
Figure 11. Distribution of effective stack heights—Volume 10
method.
34
-------
1000
100
i-
o>
--M
o>
Ul
10
2 5 10 15 20 30, 40 50 60 70 80 85 90 95 98
PERCENTAGE
Figure 12. Distribution of effective stack heights—Volume 10
adjusted method.
35
-------
1000
100
t/1
i-
QJ
CD
10
5 10 15 20 30 40 50 60 70 80 85 90 95 98
PERCENTAGE
Figure 13. Distribution of effective stack heights--D
stability and 5 m/s windspeed.
36
-------
1000
\ I II 11
100
V)
s~
OJ
-p
"(U
10
1
2
10 15 20 30 . 40 50 60 70 80 85 90 95 98
PERCENTAGE
Figure 14. Distribution of effective stack heights—Volume 10
adjusted for sources that emit greater than 100
tons/yr.
37
-------
1000
100
S-
OJ
CD
I—I
UJ
10
5 10 15 20 30 40 50 60 70 80 85 90 95 98
PERCENTAGE
Figure 15. Distribution of effective stack heights—Volume 10
adjusted for sources that emit less than 100
tons/yr.
38
-------
TABLE 6. EFFECTIVE STACK HEIGHTS CALCULATED BY THE PTMAX MODEL
(meters)
Percenti le
10th
50th
90th
Minimum
stack height
Maximum
stack height
Vol. 10
method
all sources3
10
30
112
4
185
Adjusted Vol. 10 method"
All sources
13
32
142
6
28
Q <100
tons/yr
13
30
122
6
176
Q >100
tons/yr
14
75
183
4
286
Neutral
conditions
all sources
12
38
115
5
286
Precautionary values included in data set.
Adjusted to eliminate any precautionary values (Section 3.3).
Stability D, 5 m/s windspeed.
tons/yr mean H' by 2 m because of the comparatively large number
of modified sources applying for PSD permits with emission
changes of <100 tons/yr. When the sources with Q >100 tons/yr
were included, the 90th percentile value increased by 20 m, but
the 10th percentile value remained the same.
As in the case of the emission changes, although the larger
sources with emission changes of 100 tons/yr or more were fewer
in actual number, because they vary widely at the upper end of
the distribution, they tend to skew the distribution and ulti-
mately the results toward the larger sources if they are included
in the analysis.
4.4 DISTRIBUTION OF CONCENTRATIONS
To evaluate de minimis air quality concentrations, pol-
lutant-specific concentration distributions were generated for
all sources considered in the analysis. Figures 16, 17, 18, and
19 illustrate the distributions for 24-h concentrations of PM and
S02 / 8-h concentrations of CO, and annual averages of N02, re-
spectively. These concentrations were estimated by using the
PTMAX model (with precautionary values eliminated) and any addi-
tional modeling results available in the PSD permit files. Table
39
-------
1000
100
CD
f,
-z.
O
1—i
t-
Qi
t—
•z.
UJ
o
-z.
o
o
10
1.
0.1
2
Figure 16.
5 10 20 30 40 50 60 70 80 90 95 98
PERCENTAGE
Distribution of TSP concentration using adjusted Volume 10
for sources that emit less than 100 tons/yr.
40
-------
100
10
CD
-------
10.
1.0
E
CD
o
o
0.1
0.01
5 10 15 20 30 40 50 60 70 80 85 90 95 98
PERCENTAGE
Figure 18. Distribution of N02 concentrations using adjusted Volume
10 for sources that emit less than 100 tons/yr.
42
-------
1.0
E
' PL
«=C
o:
LU
o
o
o
0.01
0.001
2
I I I
5 10 15 20 30 40 50 60 70 80 85 90 . 95 98
PERCENTAGE
Figure 19. Distribution of CO concentrations using adjusted Volume
10 for sources which emit less than 100 tons/yr.
43
-------
7 summarizes these concentrations and provides the minimum and
maximum as well as mean values expected for each pollutant.
Because of the limited available data on Pb in the permit files,
no additional modeling was performed for this analysis.
TABLE 7. POLLUTANT CONCENTRATIONS
(|jg/m3)
Percentile
10th
50th
90th
Minimum
concentration
Maximum
concentration
24-h concentration
TSP
0.5
7
60
0.01
622
S02
0.3
4.
50
0.12
129
8-h
concentration
CO
0.009
0.05
0.2
0.005
0.286
Annual average
concentration
N0x
0.09
0.85
7.0
0.03
10.3
The cumulative frequency distributions of concentrations can
be used along with the distribution of emissions to obtain some
indication of the number or percentage of sources within the
sample population that would emit greater than a given amount or
that would have greater than a specified air quality impact.
Although these distributions are limited to the sources within
the given sample of sources for which the de minimis analysis was
performed, they can provide a relative assessment as to the
number of sources that would be affected if a specified emission
or air quality level were designated as being de minimis. It
should be pointed out, however, that not all sources that would
be affected by the de minimis levels are included in these dis-
tributions, so they would not be indicative of the total number
that would be affected. A more complete assessment of all those
that are currently subject to PSD or that would be subject to
PSD, depending on the de minimis levels that might be selected,
is provided in Section 4.8. The main purpose of the distribu-
tions was to ensure that a full range of emission levels or air
quality concentrations was considered in the de minimis analysis
and that the analysis was based on these typical sources expected
to make small emission changes over the course of a year.
44
-------
3S-. 5 EMISSIONS ASSOCIATED WITH SPECIFIC AIR QUALITY LEVELS
Pollutant-specific ratios x/Q of pollutant concentrations
(x) and their corresponding mass emission rates (Q) were gener-
ated for TSP, SO2, NO , and CO by the use of the x values from
modeling results contained in PSD permit files and the x values
estimated by Volume 10, Phase II, PTWAX modeling with precaution-
ary values adjusted following the procedure listed in Section
3.4. To avoid a skewed distribution, sources with mass emission
rates greater than 100 tons/yr were also eliminated from the x/Q
data set because they would not be representative of emission
changes that would be affected by the de minimis exemption. Even
after these precautionary values were adjustable, a sample size
of greater than 25 data points was available for all criteria
pollutants concerned, i.e., TSP, SO2, CO, and NO . Because of
deposition problems with Pb, the x/Q ratio was generated by a
separate methodology (Section 3.4).
Figures 20 through 23 represent the pollutant-specific x/Q
plots generated from adjusted Volume 10, PTMAX data, and modeling
results from permit files for sources with emission changes <100
tons/yr. The mean 50th percentile value for each pollutant was
divided into the respective values being considered for de
minimis limits to determine the associated mass emission rate.
For example, the 50th percentile x/Q for TSP is 0.43 (Figure 20).
Dividing 7.8 pg/m3 (3% of the NAAQS) by 0.43 yields an associated
emission rate of approximately 18 tons/yr. For TSP and S02,
various percentages of the primary standard (2 through 5%) and
percentages of the Class II increment (10 and 20%) were evalu-
ated. The same percentage range of the annual standard was
evaluated for NO . In the case of CO, one percent of the stan-
dard yielded a corresponding emission rate of greater than 100
tons/yr; therefore, further evaluation of other percentages was
not necessary. Table 8 lists the Q values estimated by this
analysis for TSP, S02, CO, and NO by use of the 50th percentile
mean x/Q values. These valuesxare pollutant-averaging time
specific—24-h for TSP and S02, 8-h for CO, and an annual average
for NO .
X
The x/Q for Pb was constructed by use of EPA modeling re-
sults,5 and a x/Q distribution (Figure 24) was generated, which
yielded a 50th percentile value of 0.1. This value can be used
in the same manner as the x/Q's for TSP, S02, CO, and NO . An
air quality value equal to 5 percent of the standard ^0.075
jjg/m3) would therefore be associated with a Q of 0.75 ton/yr.
Since no air quality standard exists for HC and since NO
and VOC are interrelated in the formation of ozone, results from"
the NO analysis were used for VOC (O3) de minimis determination.
X "'
In addition to the pollutant-specific x/Q distribution, a
distribution including the x/Q values for all pollutants was
constructed (Figure 25) so that the x/Q concept could be used to
obtain emission rates from air quality impact estimates associ-
ated with noncriteria pollutants. The combined-pollutant x/Q
- 45
-------
10
1.0
(/I
•ZL
o
(/I
(/I
o
t—(
2 o.i
0.01
5 10 15 20 30 40 50 60 70 80 85 90 95 98
PERCENTAGE
Figure 20. Distribution of concentration/emission for TSP.
46
-------
10
1.0
en
co
•z^
o
CO
to
LU
•a:
ce:
o
o
0.1
0.01
2 5 10 15 20 30 40 50 60 70 80 85 90 95 98
PERCENTAGE
Figure 21. Distribution of concentration/emission for S02.
47
-------
100
10
en i
UJ
•z.
o
o
o
o
0.1
5 10 15 20 30 40 50 60 70 80 85 90 95 98
PERCENTAGE
Figure 22. Distribution of concentration/emission for N02.
48
-------
10.2.
1.0
en
oo
^
o
CO
oo
2 o.i
o
o
0.0]
5 10 15 20 30 40 50 60 70 80 85 90 95 98
PERCENTAGE
Figure 23. Distribution of concentration/emission for CO.
49
-------
Ol
rv
CO
o
I—I
CO
CO
LU
O
10.
1.0
0.1
0.01
0.001
2 5 10 20 30 40 50 60 70 80 90 95 98
PERCENTAGE
Figure 24. Distribution of concentration/emission for Pb.
50
-------
100
10
en
oo
z:
o
oo
oo
UJ
z:
o
i—i
«=c
UJ
o
o
o
0.1
0.01
5 10 20 30 40 50 60 70 80 90 95 98
PERCENTAGE
Figure 25. Distribution of concentration/emission for CO,
S02, and TSP combined.
51
-------
TABLE 8. EMISSION RATES ASSOCIATED WITH SELECTED
AIR QUALITY LEVEL
Pol lutant
TSP, 24-h avg
S02) 24-h avg
CO, 8-h avg
NO , annual
avg
50th percent! le
X/Q,
(jg/m3
tons/yr
0.43-
0.385
0.23
0.05
Percent of
standard(S) or
increment (I),
%
10 I
2 S
20 I
3 S
4 S
5 S
2 S
10 I
3 S
4 S
20 I
5 S
1 S
2 S
5 S
X,
pg/m3
3.7
5.2
7.4
7.8
10.4
13.0
7.3
0.1
10.9
14.6
18.2
18.3
100
2.0
5.0
Q,
tons/yr
8.6
12.1
17.2
18.1
24.2'
30.2
18.9
23.6
28.3
37.9
47.3
47.5
435
40
100
distribution was evaluated for the noncriteria pollutants instead
of a pollutant-specific x/Q distribution similar to those gener-
ated for TSP, S02 , CO, and NO because of the limited available
modeling data for the noncriteria pollutants in the permit files.
The x/Q for all pollutants was deemed to be appropriate for this
purpose, as it uses the ratio of emissions to air quality for
each source and, in this way, factors out any unique characteris-
tics of a given pollutant. The mean 50th percentile value for
the combined x/Q w^s 0.325.
An alternate method for relating mass emission rate and
concentration was developed for comparative purposes (Section
3.5). The EPA PTDIS model, a flat terrain Gaussian dispersion
model in the UNAMAP series that estimates short-term concentra-
tions directly downwind of a .point source, was selected to pre-
dict the concentration-mass-emission-rate relationship. This
model allows the user either to input a value for H' or to enter
the various stack parameters, such as physical stack height,
stack exit velocity, stack gas temperature, and stack diameter,
from which the model calculates H'. By use of the mean H' of the
sample population as the input value, all the mean values of the
key stack parameters listed above can be considered as they
relate to a specific H' instead of separate nonassociated average
values.
52
-------
** " The mass emission rates associated with selected de minimis
air quality levels can be calculated by dividing the selected de
minimis air quality values by the concentrations estimated by the
PTDIS model. The 50th percentile of the H' distribution calcu-
lated by the PTMAX model (by use of the adjusted Volume 10 Phase
II approach as described in Section 3.4), was input the PTDIS
model along with stability-windspeed combinations used in the
PTMAX model. Each combination of stability and windspeed was run
until a maximum concentration and a corresponding "worst case"
condition were identified. This value is the pollutant concen-
tration associated with each ton of emission emitted and can be
used to calculate the emissions associated with a given air
quality value. The results of this analysis are presented in
column Q2 of Table 9.
TABLE 9. EMISSION RATES ASSOCIATED WITH SELECTED AIR QUALITY
LEVELS ARRIVED AT BY SEVERAL MODELING TECHNIQUES
Pollutant
TSP
S02
CO
N0x
X,
|jg/m3
10.4
14.6
100
2
Tons/yr
Qia
24.2
37.9
435
40
Q2b
3.7
5.1
35.2
5
Qsc
25.7
36.1
247.5
35
Q
-------
this case a more refined modeling technique was used. The re-
sults of these calculations are presented in column Q4 of Table
9.
4.6 URBAN AREA AIR QUALITY IMPACT DUE TO DE MINIMIS CHANGES IN
EMISSIONS
To obtain an indication of the regional impact of major
sources making de minimis changes, the urban version of the RAM
model was run for 37 actual stationary sources of SO2 in a mid-
western metropolitan area. The locations of these sources, along
with their particular source characteristics, were input into the
model (Table 10). The source strength for all sources was re-
duced to unity so that each source would equally impact the
spacing of the honeycomb grid established by the model for the
area. Once the base case had been established, it was assumed
that each source would be modified by a specific amount and the
incremental change in air quality concentration would be calcu-
lated for the 45 receptors spread across the metropolitan area.
The results of this modeling effort are shown in Table 11.
The maximum change in the 24-h concentration from all sources
making a 50-ton/yr change would be 1.5 pg/m3. A 25-ton/yr change
would be approximately 0.75 pg/m3. Thus, on an urbanwide basis
the overall air quality change associated with 37 sources making
a de minimis change of 50 tons/yr would be slightly above the
significant levels set forth in the June 19, 1978, PSD regula-
tions (43 FR 26398).
TABLE 11. RESULTS OF URBANWIDE AREA MODELING OF SELECTED DE MINIMIS LEVELS
Max aggregrated x
from sig. pt.
sources jjg/m3
Max aggregrated x
from all sources
|jg/m3
10 tons/yr
0.21
0.30
25 tons/yr
0.53
0.75
40 tons/yr
0.85
1.2
50 tons/yr
1.1
1.5
4.7 CLASS I AREA PROTECTION
Under the Clean Air Act, clean areas of the Nation could be
designated under one of three classes, which permit varying de-
grees of air quality deterioration. Allowable increments of air
pollution were established for each class at a level that was
considered significant for that area. Because the Class I incre-
ment permitted the least air quality deterioration, a modeling
54
-------
TABLE 10. POINT-SOURCE INPUT DATA FOR REGIONAL AIR QUALITY ANALYSIS
POINT SOURCE LISTING
Cn
SOUKCt
1
2
3
4
5
. — fc _
7
8
9
10
11
1 ^
1 e
13
14
- --15 —
16
17
1 6
19
20
•- 1
22
23
24 ... —
25
26
27
28
29
30— — -
31
32
33
34
35
.... - j(,
37
tA3T
COORD
(USER
739.00
739.00
73^.00
739.00
744.10
744.10 —
733.60
742.00
736,70-
736.70
NUHTH 302K./SEC) PAKTIG/.StC) STACK
COOKO EMISSIONS EMISSIONS HTCM)
UNITS)
4401 .70
4401 .70
4401.70
4401.70
4399.80
4399.80'
4410.30
4405.00
4403.-50
4403,50
735.40 4394.50
737.00 4394.70 -
737.00
743.00
730.20
738.20
739.80
739.00
741.00
740.20
•741)-. £0 --
740.20
741.40
-741 .40 —
732.50
732.50
732.50
732.50
739.00
73<*. 1 0" "
739.10
73V. 10
739.10
740.00
740.00
742.40
7
-------
study was conducted to determine if the de minimis levels would
adversely impact a Class I area.
The EPA PTDIS program was run in relatively flat terrain for
varying stability-windspeed combinations to determine the ex-
pected allowable incremental changes in air quality at varying
emission rates and to estimate concentrations associated with
these emission rates at various distances from the source (Sec-
tion 3.7). Table 12 lists the distance from the source at which
a 1 ug/m3 maximum 24-h air quality impact would occur for varying
emission rates with a 30-m effective stack height. Table 13
lists the distance from the source at which a 1-ug/m3 maximum
24-h air quality impact is estimated to occur for a 40-ton/yr
change in emissions with effective stack heights of 13, 30, and
122 m. This assessment was performed with meteorological condi-
tions that provided the highest estimated concentration for each
of the above effective stack heights (i.e., worst case). In
addition to the use of the worst case stability and windspeed, the
estimated concentrations and associated downwind distances were
reviewed for all the stability-windspeed combinations and the
maximum distance where a 1 ug/m3 maximum 24-h concentration is
expected to occur was selected. The results are presented in
Table 14.
TABLE 12. ESTIMATED AIR QUALITY IMPACTS3
Mass
emission
rate,
tons/yr
250
100
50
25
Distance
at
which
1 ug/m3
impact
occurs
=30
£10
— 5
= 3
Concentration, pg/m3
At 3 kmb
10
4
2
1
At 5 kmb
4
2
1
0.5
At 10 kmb
2.5
1
0.5
0.3
At 30 kmb
1
0.4
0.2
0.1
"Worst case" meteorological conditions--B stability class (unstable), wind-
speed 0.5 m/s (light winds).
3Mode1 results transformed to 24-h averaging time by Volume 10 techniques;
distance and concentrations valid for any inert nonreactive pollutant for a
24~h averaging time.
56
-------
TABLE 13. ESTIMATED DISTANCE AT WHICH 1 pg/m3 MAXIMUM
24-h CONCENTRATION IS PREDICTED TO OCCUR FOR A
40-ton/yr CHANGE WITH WORST CASE METEOROLOGY
H'
13
30
122
Stabi 1 ity
D
B
A
Windspeed, m/s
1
0.5
0.5
Distance, km
20
5
3
TABLE 14. ESTIMATED DISTANCE AT WHICH 1 [jg/m3 MAXIMUM
24-h CONCENTRATION IS PREDICTED TO OCCUR UNDER
ANY METEOROLOGICAL CONDITION
H'
13
30
122
Stability
D
D
B
Windspeed, m/s
1
1
0.5
Distance, km
20
20
5
On the basis of these results, a Class I area -in level
terrain should not be impacted adversely by a de minimis change
of 40 tons/yr unless the proposed modification would be within 20
km of a Class I area. It should be noted that the meteorological
condition (D stability and windspeed of 1 m/s) associated with a
40-ton/yr emission change (from an effective stack height of 13
m) that is estimated to have a 1 pg/m3 maximum 24-hour impact 20
km away may have a relatively low frequency of occurrence on an
annua.1. basis nationwide. A limited analysis of nine locations
geographically spread across the county was undertaken to deter-
mine the frequency of occurrence for D stability and a windspeed
of 1 m/s. The results of this analysis indicated that the aver-
age frequency of occurrence for D stability and a windspeed for
1 m/s was approximately 2 percent. Although the frequency of oc-
currence will vary from area to area (1 to 3 percent for the
areas analyzed), there is a indication that the frequency of oc-
currence for D stability with a windspeed of 1 m/s will generally
be quite low. Therefore, a review of some of the other more typ-
ical or representative meteorological conditions was undertaken.
This review indicated that the distances where a 1 [jg/m3 maximum
24-h concentration could be expected to occur ranged from 2 to 8
km. Therefore, in many cases, sources making a 40 ton/yr change
and locating more than 10 km from a Class I area would not have
an impact of greater than 1 pg/m3 maximum 24-h concentration.
1 Since many Class I areas will be located in areas with
elevated terrain, the Valley model7 was used to determine the
effect that elevated terrain might have on the maximum distance
where a 1-pg/m3 maximum 24-h concentration is predicted to occur
57
-------
as a result of a source making a 40-ton/yr change. The Valley
model was run for effective stack heights of 13, 30, and 122 m
and three stability and windspeed combinations (B, 0.5 m/s; D, 1
m/s; and F, 2.5 m/s). The stability classes and windspeeds were
selected to be consistent with the conditions used in the PTDIS
modeling, which provided the maximum concentration and downwind
distance, and the procedures set forth in Volume 10.1 Volume 10
indicates that F stability and a windspeed of 2.5 m/s should be
used to estimate the impact at receptors in elevated terrain to
determine if terrain is likely to be intercepted. The basic
approach was to situate a receptor grid at various distances on
elevated terrain downwind from the source. Terrain heights were
chosen to coincide with expected maximum impact locations and
expected l-[jg/m3 maximum 24-h concentration locations. The
results from the Valley model are presented in Table 15. Two
items should be noted regarding these results. The first is that
for unstable/neutral atmospheric stability conditions (B and D
stability), the plume in the Valley model is assumed to maintain
a constant height above the terrain. "The plume parallels the
terrain feature by increasing and decreasing its effective height
relative to the stack base; this is, in effect, a flat-plane
situation [as shown in the upper sketch of Figure 26] . These
conditions may therefore lead to an underestimation of concentra-
tion in complex terrain"7 and therefore more stable conditions
should be used (i.e., F stability) as indicated in lower sketch
in Figure 26. Since B and D stability conditions in the Valley
model are more representative of flat terrain situations, it
would seem that the results should be directly comparable to
those from the PTDIS model, which indicated that these conditions
would provide a maximum distance of 20 km. However, the results
from the Valley model indicate that the maximum distance would be
from 5 to 10 km. The major reason for this apparent discrepancy
is that the Valley model uses a sector averaging approach to
estimate concentration. "The bivariate Gaussian formulation is
converted to the cross-section averaging form for a 22.5° sector.
Such a conversion results in a uniform concentration across the
wind sector at a given distance and height."7 In some cases the
averaging over the sector will produce results that are slightly
less than those obtained for a given distance when PTDIS is used.
Therefore, based on the results obtained from PTDIS and
Valley, 10 km represents a realistic approximation of the dis-
tance beyond which a 40-ton/yr emission change would not
significantly impact a Class I area.
4.8 NUMBER OF SOURCES AFFECTED BY THE PROPOSED DE MINIMIS EMIS-
SION LEVELS
Changing the current definition of modification to the pro-
posed definition would have far-reaching effects on the applica-
bility of the PSD regulations, as shown in Table 16. There are
151 modifications for which PSD permits have been issued and for
which data were gathered as part of a survey of the PSD permits
58
-------
TABLE 15. ESTIMATED DISTANCE AT WHICH 1 (jg/m3 MAXIMUM 24-h
CONCENTRATION IS PREDICTED TO OCCUR FOR A 40-ton/yr
CHANGE IN ELEVATED TERRAIN USING THE VALLEY MODEL
H'
30
13
122
Stability
F
B
F
D
F
Windspeed, m/s
2.5
.5
2.5
1
2.5
Distance, km
10
5a
10
10a
10
For those conditions in the Valley model, the plume is assumed to maintain a
constant height above terrain, which in effect is a flat-plane situation.
issued from April 1, 1978, to November 1, 1979. These modifica-
tions were obviously subject to current regulations. Of the 151,
79 had controlled emissions above the cutoff of 100 or 250 tons/
yr without any emission reductions elsewhere within the source,
so these 79 would also be subject to the proposed regulations; 52
had controlled emissions below the cutoff of 100 or 250 tons/yr,
and had one or more pollutants for which controlled emissions
exceeded the de minimis levels without any offsets indicated; the
other 20 had no pollutant for which controlled emissions exceeded
the de minimis levels. If all 52 were major for the pollutant
exceeding the "de minimis" levels', these modifications would be
subject to the proposed regulations, but i'f some of the 52
sources were not major, then some would not be subject. There-
fore, 52 is the outside estimate of the number of modifications
below 100 or 250 tons/yr that would be subject to review. The
actual number may be somewhat less, depending on the major source
status of the existing source. The proposal, however, would
clearly exclude the modifications (20) that resulted in emis-
sions less than the de minimis levels, regardless of whether or
not the sources were major.
The following conclusions were drawn from the analysis of
the proposed and current definitions of modification.
1. Of the modifications subject to the current regula-
tions, 14 percent would not be subject under the pro-
posed regulations.
2. Of the modifications subject to current regulations,
52 percent would be subject under the proposed
regulations.
3. Of the modifications subject to the proposed regula-
tions, 34 percent may or may not continue to be
subject, depending on whether or not the sources were
major before the modification and whether the sources
could offset the increases by using the netting provi-
sion.
59
-------
UNSTABLE AND
NEUTRAL CATEGORIES
II
STABLE
CATEGORIES
FRACTION
OF PLUME
REMAINING
IN SECTOR
Figure 26. Depiction of plume height in complex terrain, as
in the Valley Model; h is the height of the
plume at final rise abSve ground for the unstable
and neutral cases and above stack base for the
stable cases.7
60
-------
TABLE 16. EFFECTS OF PROPOSED REGULATIONS ON MODIFICATIONS
REVIEWED UNDER CURRENT REGULATIONS
' U.S.
EPA
region
III
IV
v
VI
VII
VIII
IX
X
Total
Number
of
modifi-
cations
13
38
10
49
- 9
12
18
2
151
Number with
>100 or 250
tons/yr con-
trolled
5
15
6
32
4
4
11
2
79
Number with <100 or 250
tons/yr
Above
de minimis
levels
6
17
3
14
4
4
5
0
52
Below
de minimis
levels
2
6
2
3
1
4
2
0
20
Based on the data, a higher percentage of modified than new
sources would be subject under the proposed regulations, but the
actual numbers of modified and new sources appear to be less than
under the current regulations. This conclusion could be erro-
neous, however, especially for modified sources, since only
modifications with more than 100 or 250 tons/yr were subject to
PSD review and the number of modifications subject to the pro-
posed regulations but not to the current regulations could not be
obtained from the permits. Major modified sources that would
increase emissions above the de minimis limits, but that would
have increases of less than 100 or 250 tons/yr were not subject
to the current regulations, but they would be subject to the
proposed regulations; currently these are only subject to the
State's new source review procedures. In fact, many States do
not consider modified sources of 10 to 20 tons/yr a major source
of emissions, so these sources are reviewed only to ensure that
they meet the State's emission limits unless there is evidence
that air quality problems may exist as a result of the modifica-
tion.
States do not summarize and thus do not routinely report to
EPA the amount of emissions from minor sources, but these data
would be in the State permit file. Determination of how many
sources a year would have emissions more than de minimis but less
61
-------
than 100 or 250 tons/yr would require a review of all of the
States' permit files. Since a detailed review of all State
permit files could obviously not be undertaken, selected States
were contacted to obtain a representative sample of the number,of
modifications that could be subject to PSD review under the
proposed regulations. Data from Connecticut, Vermont, New York,
Massachusetts,8 Ohio,9 North Carolina,10 and Florida11 on the
estimated number of modifications above the de minimis levels and
below the cutoff of 100 to 250 tons/yr were put into four cate-
gories based on the population of the states surveyed: greater
than 15 million, 5 to 15 million, 1 to 5 million, and less than 1
million people. These data, which represented the modifications
that would receive permits in any 1 year, were used for a rough
estimate of the number of modifications that might fall into this
category for the entire United States.
The estimate of the total number of modifications that are
not subject to the current regulations but that would be subject
to the proposed regulations was obtained by multiplying the
estimated number of permits to be issued for a given population
range by the number of States having a population in that range.
This amounted to approximately 5000 modifications per year
(values obtained ranged from 3400 to 6600). No estimate was
made, however, of how many of these modifications would occur at
existing major sources. In order to obtain some estimate of how
many of these modifications may occur at major existing sources,
a review of the NEDS file was undertaken. According to the in-
formation in NEDS, there were approximately 56,000 point sources
in the NEDS system as of January 1979. Of these, approximately
12,000 were major sources (i.e., with emissions of any criteria
pollutant greater than 100 tons/yr). Based on an estimate that
there will be 5000 modifications per year, this would mean that
approximately 10 percent of the existing 56,000 stationary
sources would be modified in any given year. This estimate seems
realistic, based on some limited data from the State of Louisana
that indicated that approximately 100 TSP or S02 sources received
State new source review permits during 1978. If the same per-
centage of modifications per year for all sources in NEDS holds
true for those emitting greater than 100 tons/yr, then approxi-
mately 1200 of the 12,000 sources with emissions greater than
100 tons/yr would be expected to modify their source every year.
Therefore, the estimates obtained from .the state agencies would
seem to represent the total modifications that would be expected
per year. Thus, the number expected only from those with exist-
ing emissions greater than 100 tons/yr would be approximately
1200. Based on the proposed definition, it is estimated that
approximately 1200 additional modifications per year would be
subject to PSD over and above those that are now currently sub-
ject to review and would continue to be subject to review based
on the proposed de minimis levels.
Because the above estimate was developed as a result of
communication with state agency personnel rather than a direct
62
-------
review of the files, no estimate was obtained on the distribution
of these modifications based on their total emissions. There-
fore, it is difficult to obtain an estimate of the number of
additional modifications that would be affected under alternative
de minimis levels.
Although the de minimis levels are pollutant-specific, very
few sources emit just one pollutant. Therefore, a pollutant-
specific analysis does not necessarily provide an indication of
the absolute number -of sources that would be subject to review
given certain de minimis levels. In other words, even though a
source would no longer be subject to review for TSP because it
had emission changes of less than 10 tons/yr, it would still be
subject to review if it had S02 emission changes of 20 tons/yr.
To obtain some estimate of the 'total number of currently per-
mitted modifications that would be subject given specific re-
vised de minimis emission levels, all 151 modifications included
in the survey of PSD permits were evaluated and categorized ac-
cording to the greatest amount of emission changes from any of
the criteria pollutants that would be emitted from the source.
For example, if a source had emission changes of 10 tons/yr of
PM, 25 tons/yr of S02 , 30 tons/yr of N0x and 110 tons/yr of VOC,
it was categorized as having emission changes of greater than 100
tons/yr. Therefore, unless the de minimis levels were raised to
above 100 tons/yr for VOC, it would still be subject to PSD even
if the de minimis levels for all other pollutants it emitted as a
result of the change were raised to 35 tons/yr. The results of
this analysis are shown in Figure 27. Only five criteria pollu-
tants were considered in this analysis since none of the modifi-
cations reported emission estimates for lead.and very few pro-
vided estimates for noncriteria pollutants.
If the same general emissions distribution of modifications
for which PSD permits have been issued to date holds true for
those modifications that were not previously subject to review,
then one can obtain some estimate of the impact of selected de
minimis levels for all modifications (those currently subject
plus those not currently subject to PSD) that would be subject to
PSD review as a result of the proposed regulations. Figure 28
combines both these data sets on modifications by using the
distribution for the ones that have received permits to date.
The estimate of the total modifications that could be sub-
ject based on various de minimis levels as shown in Figure 28
assumes that all proposed de minimis levels would be the same for
all pollutants. If different'de minimis levels are suggested for
each pollutant, a specific analysis of the 151 modifications that
received permits would be needed for each combination of de
minimis levels considered. To determine the difference in selec-
ting various de minimis levels by pollutant versus selecting one
common level, the following de minimis level combination was
evaluated: TSP at 25 tons/yr, S02, NO , and HC at 40 tons/yr,
and CO at 100 tons/yr. As a result, approximately 74 percent of
the modifications would be subject, compared with 68 percent if
63
-------
40 tons/yr were used for all pollutants, as indicated in Figure
27. Given that the modifications that are currently not subject
to PSD have the same general distribution of sources and emis-
sions as the modifications that are currently subject to review,
approximately 890 of the 1200 additional modifications would be
subject to the proposed regulations given the above de minimis
levels of TSP, SO2, NO , and CO of 25, 40, 40, 40, and 100 tons/
yr, respectively. (If 40 tons/yr for all pollutants were con-
sidered de minimis this would be 816 of the 1200.)
64
-------
NUMBER OF SOURCES LESS THAN DE MINIMIS LEVEL
Figure 27. Number of current modifications subject to PSD versus
de minimi's emission rates.
65
-------
100 —
80
60
40
20
100
200
300 400 500 600 700 800
NUMBER OF SOURCES LESS THAN DE MINIMIS LEVEL
1000 1200
Figure 28. Number of total modifications at major sources versus de minimi's emission rates.
-------
REFERENCES
1. Guidelines for Air Quality Maintenance Planning and Ana-
lysis, Volume 10 (Revised): Procedures for Evaluating Air
Quality Impact of New Stationary Sources. EPA 450/4-77-001,
October 1977.
2. Turner, D. B. Workbook of Atmospheric Dispersion Esti-
mates. PHS Publication No. 999-AP-26 (NTIS PB 191482). En-
vironmental Protection Agency, Research Triangle Park, North
Carolina. 1970.
3. Guideline on Air Quality Models. EPA 450/2-78-027, April
1978.
4. Schewe, G. Dispersion Model Analysis of the Air Quality
Impact of Criteria Pollutant Emissions from Fossil-Fueled
Industrial Boilers, Unpublished. Source Receptor Analysis
Branch to Industrial Studies Branch. December 6, 1979.
5. Memo from Joseph A. Tikvart to John O1 Conner. Final Re-
port on Dispersion Modeling Results for the SSEIS on Lead.
June 15, 1977.
6. Aver, A. H. Jr. Correlation of Land Use and Cover with
Meteorological Anomales. Journal of Applied Meteorology,
17, May 1978, 636-643.
7. Burt, E. W. Valley Model Users Guide. EPA 450/2-77-018,
September 1977.
8. Personal communication with P. Fairchild, Northeastern
States Commission on Air Quality Management, February 25,
1980.
9. Personal communication with H, Johnson, Ohio Environmental
Protection Agency, February 22, 1980.
10. Personal communication with M. Sowell, Division of Environ-
mental Management, State of North Carolina, February 19,
1980.
11. Personal communication with J. Preece, Department of En-
vironmental Regulation, State of Florida, February 21,
' 1980.
67
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-450/2-80-072
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Impact of Proposed and Alternative De Minimi's Levels
for Criteria Pollutants
5. REPORT DATE
June 80
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
David Dunbar, Barbara Blegun, Dr. Jeff Smith
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
10. PROGRAM ELEMENT NO.
PEDCo Environmental Inc.
Durham, N.C. 27701
11. CONTRACT/GRANT NO.
68-02-3173
12. SPONSORING AGENCY NAME AND ADDRESS
U.S. EPA
Office and Air Quality Planning and Standards
Research Triangle Park, N.C. 27711
13. TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
The report estimates the impact of the prevention of significant deterioration
(PSD) regulations under various applicability size cutoffs for criteria pollutant
emissions. The analysis is based on the information obtained from completed PSD
permitting actions. The report summarizes existing permit data by showing
distributions of: 1) emissions levels of criteria pollutants; 2) actual and
effective stack heights; 3) maximum downwind air quality impact; and 4) emissions
associated with specific air quality impacts. The report also evaluates the
combined worst case air quality impacts from several sources making a de minimi's
change in emissions and the impact of changes over distance from pristine (Class I)
areas.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS C. COSATI Held/Group
PSD
de_ minimi's
Class I
Air Quality Modeling
19, SECURITY CLASS (Tins Report/
Unclassified
21. NO. OF PAGES
74
20. SECURITY CLASS {TSus page/
Unclassified
|22. PRICE
EPA Form 2220-1 (Rev. 4-77)
PREVIOUS EDITION IS OBSOLETE
-------
------- |