EPA Region 3 Comments to EPA's "DRAFT Guidance for Ozone and Fine Particulate Matter Permit
Modeling", dated 10 February 2020, Comments Prepared March 2020

Comment 01: Section II.l Significant Emissions Rates for O3 and PM2.5

Note there may be some PM2.5 nonattainment areas (designated as serious or above) where
ammonia emissions are considered PM2.5 precursors. A source with ammonia emissions located
close to one of these nonattainment areas may need to assess its impacts on any nearby
nonattainment areas where ammonia has been defined as a PM2.5 precursor.

Comment 02: II.3 Significant Impact Levels for O3 and PM2.5

Region 3 reminds OAQPS staff that AERMOD does not currently calculate the annual PM2.5
concentration in the correct format of the NAAQS; the annual standard is calculated from
seasonal averages (40 CFR Part 50, Appendix N, Section 4.4 (a)1) and is not a straight annual
weighted value as determined in AERMOD. This oversight may call into question model values
that are very close to the SIL, NAAQS or PSD increment values.

Comment 03: II.5.2 PM2.5 PSD Increments Compliance

We should consider cautioning applicants with demonstrations that have (or will have) increment
expanding sources. Similar to (annual) NO2 increment expansion, addressing PM2.5 precursor
reductions using a screening technique (such as MERPS) may over-estimate increment
expansion given the conservative nature of our current screening techniques.

Comment 04: III. PSD Compliance Demonstrations for the O3 and PM2.5 NAAQS: Source
Impact Analysis

Figure II-2 provides a flow path for modeling sources based on location outside a designated
nonattainment area. Sources located inside nonattainment areas are to follow Nonattainment
NSR rules requiring emission offsets (and no modeling). For ozone purposes, there are areas in
the northeast Ozone Transport Region2 or OTR that while designated as attainment or

1	httpsi//www.govinfo.gov/content/pkg/CFR-2015-title4Q-vol2/pdf/CFR-2015-title4 0-vol2-part50.pdf

2	See §7511c. Control of interstate ozone air pollution: the OTR is defined in section (a) as "... comprised of the
States of Connecticut, Delaware, Maine, Maryland, Massachusetts, New Hampshire, New Jersey, New York,
Pennsylvania, Rhode Island, Vermont, and the Consolidated Metropolitan Statistical Area that includes the District
of Columbia..."

Page 1 of 9


-------
EPA Region 3 Comments to EPA's "DRAFT Guidance for Ozone and Fine Particulate Matter Permit
Modeling", dated 10 February 2020, Comments Prepared March 2020

unclassifiable are still required to secure emission offsets as if the area was designated as a
"moderate" nonattainment area. The application of this guidance may also be subject to section
(d) Best available air quality monitoring and modeling, which reads:

"[F]or purposes of this section, not later than 6 months after November 15, 1990, the
Administrator shall promulgate criteria for purposes of determining the contribution of
sources in one area to concentrations of ozone in another area which is a nonattainment
area for ozone. Such criteria shall require that the best available air quality monitoring
and modeling techniques be usedfor purposes of making such determinations. "

Comment 05: III.4.1 Conceptual Model

Applicants developing modeling protocols describing ozone or PM-2.5 trends and speciation
data should consider consulting state and local air monitoring reports and periodic ambient
monitor network assessment plans3 for additional information on local and regional trends along
with any potential transport issues. Monitor trends should be examined for statistical
significance along with any correlation with documented local and/or regional emission trends.
For PM-2.5, regional haze SIPs and Regional Planning Organization or RPO documents could be
consulted for Class I area speciation trends4. Future regional/local emission control programs
could also be cited as a "weight of evidence" showing source emission impacts could be offset
by future decreases in local and regional emissions in response to local and regional control
programs (for SIPs et cetera).

Just a comment on low-level jets, using single-site ASOS measurements in a dispersion
modeling analysis is not going to be able to resolve these features. These features will only be
captured using met tower/SODAR combinations or fine-scale WRF (prognostic meteorological)
simulations.

Comment 06: III.4.2 Tier 1 Assessment Approach

EPA should caution applicants to ensure that the developed MERPs calculations reflect the
current regional emissions mix in which the source is being located. Using information from
photochemical modeling studies that are "out of date" may be inappropriate since the source may

3	As required under 40 CFR 58.10(e)

4	See MANE-VU Report: httpsi//otcair,org/MANEVU/Upload/Publication/Reports/MANE-
VU Speciation and Trajectory Analyses - Final.pdf

Page 2 of 9


-------
EPA Region 3 Comments to EPA's "DRAFT Guidance for Ozone and Fine Particulate Matter Permit
Modeling", dated 10 February 2020, Comments Prepared March 2020

now be located in a different ozone and secondary PM-2.5 formation environment than
represented in the photochemical grid model. Examples of this may include areas with recent
emission increases due to wide-scale natural gas development (regional increases in NOx
emissions), portions of the northeast, Mid-Atlantic, Southeast and Midwest that are experiencing
significant shifts in electric generation from coal-fired power plants to combined-cycle natural
gas plants or regions where significant wide-ranging ozone/PM-2.5 emission control programs
such as the NOx SIP Call/CAIR/CASPR/Cross State rule have been recently implemented.

Comment 07: III.4.3 Tier 2 Assessment Approach

Similar to our previous comment, any use of a photochemical or similar models for single source
impact analysis should ensure that the emission inventory is reflective of area in which the
source is to be located.

Should the photochemical model include emissions that are current, model "base case" or
projected future case (what inventories were used for the MERPs projections)? Keep in mind
projected (future) emission inventories used in photochemical grid models will project certain
emission sectors into the future (including "new" power plants in certain areas to handle
expected growth in electricity consumption). Trends wise, one can see, at least in the east,
significant changes in PM-2.5 speciation over the last decade mainly in the sulfate component.
Recent emissions trends in SO2 emissions have changed the characteristics of when peak PM-2.5
concentrations are occurring (summer values have become less controlling; see Comment 20).

We should consider adding some emissions threshold where the use of CTMs would be more
appropriate than using a tier 1 approach. For example, a source with combined NOx and VOC
emissions of over 15,000 tpy (if one were to exist) would probably be better handled using a
CTM than a source with under 500 tpy of the same pollutants.

Using a CTM or tier 2 approach is time and resource intensive. It would be difficult to conduct
such an analysis given the typical 180-day to 360-day review periods normally allotted for
PSD/NSR applications. Region 3's experience using CTMs in SIPs indicates state and local
agencies rarely submit these within the CAA allotted 18-month submittal period. We should at
least acknowledge the substantial amount of time needed to develop an analysis of this
complexity.

Page 3 of 9


-------
EPA Region 3 Comments to EPA's "DRAFT Guidance for Ozone and Fine Particulate Matter Permit
Modeling", dated 10 February 2020, Comments Prepared March 2020

Comment 08: III.5.2 SIL Comparison for PM2.5

If one exceeds the SIL for sources with significant levels of PM-2.5 precursor emissions, how
does one determine the Significant Impact Area (SIA) for the cumulative analysis? Usually only
(AERMOD) receptors that exceed the SIL are used to determine the SIA for a cumulative
analysis. In this case, precursor emissions handled through the tier 1 methodology have no real
spatial component. Does one use the AERMOD determined SIA or somehow expand it using
the tier 1 (or tier 2) assessment (for cases 3 and 4; how do you determine a SIA for precursor
emissions only)? We could not find any further discussion regarding delineating the SIA in
section IV and IV. 1 of this guidance.

Comment 09: IV. PSD Compliance Demonstrations for the O3 and PM2.5NAAQS:
Cumulative Impact Analysis

There are many areas in which ozone and PM-2.5 monitoring is very sparse. This is especially
true for PM-2.5 chemical speciation sites. In some instances, the closest background monitor
may be hundreds of kilometers away from a proposed source and not exactly representative of
the area in which the new source will be located. The guidance should recognize this possibility
and if prudent offer possible solutions for this predicament (use of CTM to establish background
or confirm gradient?). We should also expect continued reductions in SLAM monitoring
network in response to declining federal, state and local resources.

Comment 10: IV.l Modeling Inventory

Consider referencing the National Emission Inventory or NEI. This database contains (actual)
state and local reported emissions for multiple source categories along with stack information for
most point sources. The inventory is produced every three years and is available online5. We
may want to note that state and local offices track yearly source emissions as part of their Title V
fee collection programs and could be an additional source of emissions information.

Additional emission information for larger sources is available from EPA's Clean Air Markets or
CAMD website6. This includes hourly information that may be useful for determining more

5 httpsi//www.epa.gov/air-emissions-inventories/national-emissions-inventory-nei

6 httpsi//ampd,epa.gov/ampd/

Page 4 of 9


-------
EPA Region 3 Comments to EPA's "DRAFT Guidance for Ozone and Fine Particulate Matter Permit
Modeling", dated 10 February 2020, Comments Prepared March 2020

representative model emission rates. More detailed hourly emissions can be downloaded using
EPA's Field Audit Checklist Tool (FACT)7 for some large sources (under 40 CFR Part 75)8.

Comment 11: IV.2 Monitored Background

Ozone and PM-2.5 monitoring is somewhat different that other criteria pollutants. Hourly values
are not always available from either type of monitor. Ozone monitors in some areas of the
country are not operated year-round and PM-2.5 monitors are often filter-based representing a
daily average; some PM-2.5 monitor sites do not collect on a daily basis (as briefly discussed in
section IV.3).

Comment 12: IV.3 Comparison to the NAAQS

The characterization of the annual PM-2.5 standard is incorrect. From page 48 of the proposed
draft guidance:

"[T]he PM2.5 design value for the annual averaging period is based on the 3-year
average of the annual average PM2.5 concentrations..."

The annual PM-2.5 design value is determined from a monitor's daily and quarterly values in
accordance with three (3) sets of equations outlined in Section 4.4 (a) of Appendix N to Part 50.
We noted (the possibility of) AERMOD's inconsistency with the form of the annual PM-2.5
NAAQS in our second comment.

We should note that the tier 1 assessment of PM-2.5 precursor impacts do not possess a seasonal
or temporal component to them (as they pertain to the level 1 and level 2 comparisons discussed
on page 52). Assessing seasonal or daily secondary precursor impacts are therefore not possible
as they would be for the direct component determined by AERMOD (or some other approved
dispersion model).

Tier 2 CTM (for PM-2.5) will also include a primary component (as discussed in Appendix A).
This situation is discussed in sections of our November 29, 2018 Modeling Guidance for
Demonstrating Air Quality Goals for Ozone, PM2.5 and Regional Haze. Section 4.6 (Local Area

7	httpsi//www.epa.gov/airmarkets/field-audit-checklist-tool-fact

8	httpsi//www.epa.gov/airmarkets/plain-english-guide-part-75-rule

Page 5 of 9


-------
EPA Region 3 Comments to EPA's "DRAFT Guidance for Ozone and Fine Particulate Matter Permit
Modeling", dated 10 February 2020, Comments Prepared March 2020

Analysis) of this guidance describes the proper procedure for assessing impacts of direct PM-2.5
using a Gaussian dispersion model by removing the direct PM-2.5 impacts from the
photochemical model so that these impacts aren't double counted in the analysis. This point
should also be added to this guidance.

Comment 13: V.l Overview of the PSD Increment System

We appreciate the discussion included in this section and believe it will help the regulated
community better understand PSD increment modeling. It might be helpful to provide an
updated version of section II.F BASELINE DATE AND BASELINE AREA CONCEPTS -
EXAMPLES from the EPA's 1990 New Source Review Workshop Manual as an appendix to this
guidance.

Comment 14: V.l.3 PSD Increment Expansion

This section should include a discussion of the use of negative emission rates for PSD increment
modeling for chemically active species such as NO2. It is generally not recommended to use the
Ambient Ratio Method or ARM for annual NO2 increment expansion due to its conservative
estimation of source emission impacts (consumption of NO2 via simple ozone chemistry); one
would be using an overestimation of NO2 impacts in increment expansion when modeling
negative emission rates with ARM. This may apply to MERPs adjustments to the secondary
component of PM-2.5 from precursor (SO2 and NOx) emission reductions. This point may need
to be expanded across several parts of section V.

Comment 15: V.2 PSD PM2.5 Increments

As noted in our previous comment, increment expansion via reductions of PM-2.5 precursor
emissions may need to be tempered along the same lines as NO2 increment expansion due to
potential conservative model assumptions for chemically active species.

As we noted in Comment 02, AERMOD does (may) not determine the annual PM-2.5
concentration in the proper format of the NAAQS. This should be corrected or at least
communicated to the modeling community (if it is so).

Page 6 of 9


-------
EPA Region 3 Comments to EPA's "DRAFT Guidance for Ozone and Fine Particulate Matter Permit
Modeling", dated 10 February 2020, Comments Prepared March 2020

As noted in Comment 12, tier 2 modeling using a CTM for cases 3 & 4 will include both primary
and secondary PM-2.5. Combining impacts with AERMOD will therefore create a "double
counting" situation for the primary PM-2.5 component.

Comment 16: V.3.2 PM2.5 Increments: Cumulative Analysis

Should emissions from applicants that have revised a previously approved permit application that
triggered the minor source baseline date be included in a PSD PM-2.5 cumulative analysis?
Region 3 has had some instances where an applicant has revised its project after it was approved
due to changes in planned operations. If the previous plant design was included in the PSD
increment analysis this would create a situation where there could be double counting of the
source. Similarly, we have seen cumulative PSD increment analyses include sources that
triggered the minor baseline date but were never built creating "phantom" increment
consumption by sources that never existed and whose applications have (long) since expired.

Comment 17: V.3.2.2 Assessing Secondary PM2.5 Impacts

In addition to the control programs noted in footnote 33, sources should also consider impacts
from SO2 reductions related to any nearby SO2 SIP actions.

Comment 18: Appendix A: 1. PM2.5 Monitoring Networks

It's not entirely clear if the monitoring network description in this section represents the current
status of the PM-2.5 monitoring network or some past representation. Consider adding the year
that this network description applies.

Comment 19: Appendix A: 1.3. PM2.5 Chemical Speciation Monitoring

We should probably note that there is a discontinuity between PM-2.5 components measured by
the chemical speciation network and what the photochemical models (such and CAMx) track
(thus the Sulfate, Adjusted Nitrate, Derived Water, Inferred Carbon Hybrid material balance
approach or SANDWICH9 methodology developed by EPA for PM-2.5 SIP modeling
demonstrations).

9 https://www3.epa.gov/ttnamtil/files/2006conference/frank.pdf

Page 7 of 9


-------
EPA Region 3 Comments to EPA's "DRAFT Guidance for Ozone and Fine Particulate Matter Permit
Modeling", dated 10 February 2020, Comments Prepared March 2020

Comment 20: Appendix A: 3. Seasonal and Daily Patterns of PM2.5

Some of the seasonal PM-2.5 trends included in this section are probably out of date. In the east,
SO2 emissions have been significantly reduced (mainly from coal-fired boilers in the EGU
sector) such that summer-time sulfate levels are much lower than in the past10. This has resulted
in a shift in when peak 24-hour PM-2.5 values are occurring; from summertime to wintertime.

Comment 21: Appendix B: 2.1. Emissions

Hourly emissions for select sources are available using the EPA Field Audit Checklist Tool or
FACT, which available for download at https://www.epa.eov/airmarkets/field-audit-checklist-
tool-fact.

Comment 22: Appendix B: 2.5. Source groups

Users should be cautioned about using the SRCGROUP option in AERMOD when modeling for
NO2 significance using the ARM (or other simple chemical transformation) option. Attempting
to model multiple operating scenarios by dividing them into different source groups will not
prevent the ARM chemistry from impacting what should be separately modeled operating
scenarios.

Comment 23: Appendix B: 3.1. Surface characteristics and representativeness

Section 3.1.1 of EPA's AERMOD Implementation Guide also discusses meteorological data
representativeness.

Comment 24: Appendix C: Assessment of O3

I do think looking at the source sensitivities in MERPS is an interesting approach that should be
communicated to the modeling community. That being said, we feel that using this example is
somewhat problematic. The Tennessee Valley Authority (TVA) Gleason Combustion Turbine

10 See Atmos Environ (1994). 2017 Dec 2; 175: 25-32 ( https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6134864/)

Page 8 of 9


-------
EPA Region 3 Comments to EPA's "DRAFT Guidance for Ozone and Fine Particulate Matter Permit
Modeling", dated 10 February 2020, Comments Prepared March 2020

Plant11 appears to be a 500 MW simple cycle electric generating station. These types of simple
cycle units typically are demand response units (at least in Region 3). We would be concerned
that this plant would mostly be running during periods of high electric demand during the
summer months when ambient temperatures are high and subsequently when ozone levels would
be elevated. I'd be cautious about including this as an example based on Region 3's experiences
with its OTR ozone nonattainment areas that have identified sources which have elevated
emissions during High Electric Demand days12 as something to be discouraged.

Comment 25: Appendix D

This data is over ten years old. The seasonality assumptions may no longer be applicable in
areas where regional control programs have taken effect (see footnote 10 to comment 20).

11	httpsi//www. tva.gov/Energv/Our-Power-System/Natural-Gas/Gleason-Combustion-Turbine-Plant

12	httpsi//www.energy.gov/sites/prod/files/2014/05/fl5/tap webinar 20080717 diem.pdf

Page 9 of 9


-------