1. Background
On March 12, 2012, the US Environmental Protection Agency (EPA) published the final rule on the "Arkansas; Regional Haze State Implementation Plan; Interstate Transport State Implementation Plan To Address Pollution Affecting Visibility and Regional Haze." This rule partially approved and partially disapproved Arkansas's identification of Best Available Retrofit Technology (BART) eligible sources and subject-to-BART sources; requirements for BART, Chapter 15 of the Air Pollution Control and Ecology Commission (APCEC) Regulation 19, the Long Term Strategy (LTS), and the Reasonable Progress Goal (RPG).  As a result of the disapproval of the aforementioned BART elements of the Arkansas Regional Haze State Implementation Plan (RHSIP), ADEQ has chosen to revise the RHSIP. To accomplish the revision and to comply with the statutory five-factor analysis requirements, ADEQ is requesting the subject-to-BART sources to update their five-factor analysis and perform the visibility improvement modeling.  
2. Purpose and Objective
The purpose of this addendum is to update the June 7, 2006 draft Best Available Retrofit Technology (BART) Modeling Protocol. The components in the aforementioned protocol to be changed are the CALPUFF modeling components, the background concentrations in CALPOST. In addition, CALMET is being changed from not having observations (NO OBS = 1) to the hybrid which incorporates observations (NO OBS = 0). This change from NO OBS = 1 to NO OBS = 0 will allow the use of the 8[th] highest change in deciview (dv) rather than the 1[st] highest.  Another change to the draft protocol is the use of the revised Interagency Monitoring of Protected Visual Environments (IMPROVE) equation which is discussed in another section. Please note EPA advised ADEQ to incorporate these changes for use in the visibility impact modeling of potential controls on the subject-to-BART sources in Arkansas.
The objective of this document is to present the dispersion modeling guidance for estimating the degree of visibility improvement from potential Best Available Retrofit Technology (BART) control technology options and describe dispersion modeling and analysis methods for quantifying the degree of visibility improvement from potential BART control scenarios/strategies. However, it does not explain how the visibility results are factored into the BART determination process (i.e., five-step process). 
3. Subject-to-BART Sources and Class I Areas Affected
BART determination modeling performed by ADEQ indicated there were six Arkansas facilities with subject-to-BART units whose emissions caused or contributed to visibility impairment at four Class I areas.  However, EPA disapproved ADEQ's BART exemption finding of Georgia-Pacific Paper's 6A and 9A boilers and found these units to be subject-to-BART.  Table 1 lists the facilities, subject-to-BART units and pollutants that were not approved. Figure 1 shows the location of the facilities relative to the 300 km radius of the following Class I areas: Upper Buffalo and Caney Creek Wilderness Areas, AR; Hercules-Glades and Mingo Wilderness Areas, MO; Sipsey Wilderness Area, AL. A short description of the facilities with subject-to-BART units and the Class I areas affected follows.
Table 1 Facilities with subject-to-BART units and the pollutants in the state of Arkansas
Facility Name
Unit ID - Description
Pollutants
American Electric Power  -  Flint Creek Plant
SN-01 - Boiler
SO2, NOx,
AR Electric Cooperative  -  Bailey Generating Station
SN-01  -  Boiler 
SO2, NOx, PM
AR Electric Cooperative  -   John L. McClellan Generating Station
SN-01 - Boiler
SO2, NOx, PM


Entergy  -  Lake Catherine
SN-02  -  Unit 4 Boiler Natural Gas Firing
NOx,

SN-02  -  Unit 4 Boiler Oil Firing
SO2, NOx, PM


Entergy  -  White Bluff
SN-01  -  Unit 1 Bituminous and Sub-bituminous Coal Firing
SO2, NOx

SN-02  -  Unit 2 Bituminous and Sub-bituminous Coal Firing
SO2, NOx
Entergy  -  White Bluff
SN-05  -  Auxiliary Boiler

Domtar  -  Ashdown
SN-03  -  #1 Power Boiler
SO2, NOx

SN-05 - #2 Power Boiler
SO2, NOx, PM
Georgia-Pacific Paper  -  Crossett
6A Boiler
SO2, NOx, PM

9A Boiler
SO2, NOx, PM
                                       
Figure 1 Map showing the location of Arkansas's subject-to-BART sources relative to the 300 km buffers around Caney Creek and Upper Buffalo Wilderness Areas, AR, Mingo and Hercules-Glades Wilderness Areas, MO and Sipsey Wilderness Area, AL

American Electric Power  -  Flint Creek Power Plant (AFIN 04-00107) is located in Gentry, Benton County, Arkansas and is currently permitted to operate under ADEQ Operating Air Permit Number 0276-AOP-R5. It produces power using a 6324 million BTU per hour, dry bottom, wall fired boiler (SN-01) to produce sufficient steam to operate the turbine generator at the 558 megawatt gross electrical output capability of the unit. The boiler burns primarily low sulfur western coal, but can also combust fuel oil and tire derived fuels. Fuel oil firing is only allowed during startup and shutdown of the boiler, startup and shutdown of the pulverizer mills, for flame stabilization when the coal is frozen, fuel oil tank maintenance, to prevent boiler tube failure in extreme cold weather, and when the unit is offline for maintenance. Fly ash resulting from the coal combustion process is collected by two hot side electrostatic precipitators. BART determination modeling indicated SN-01 impacted Caney Creek (CACR) and Upper Buffalo Wilderness (UPBU) areas, AR and Hercules-Glades (HEGL) Wilderness area, MO (Figure 1). 

Arkansas Electric Cooperative Corporation  -  Carl E. Bailey Generating Station (AFIN 74-00024) is located in Augusta, Woodruff County, Arkansas and is currently permitted to operate under ADEQ Operating Air Permit Number 0154-AOP-R3. It produces power using a 1350 million BTU per hour Riley Stoker boiler (SN-01) to drive a 122 megawatt generator. The primary fuel is natural gas but is also permitted to combust fuel oil. The facility is permitted to use any grade fuel oil with a sulfur content equal to or below 2.3 per cent. As shown in Figure 1, this facility is within the 300 km radius of Mingo (MING) and Hercules-Glades (HEGL) Wilderness areas, MO, and Upper Buffalo (UPBU) and Caney Creek (CACR) Wilderness areas, AR. Preliminary modeling of this unit showed emissions impacted visibility at all four of the Wilderness areas. 

Arkansas Electric Cooperative Corporation - John L.  McClellan Generating Station (AFIN 52-00055) is located in Camden, Ouachita County, Arkansas and is currently permitted to operate under ADEQ Operating Air Permit Number 0181-AOP-R4. The plant produces power using a 1436 million BTU per hour Riley Stoker boiler (SN-01) to drive a 134 megawatt generator. The primary fuel is natural gas but is also permitted to combust fuel oil. The facility is permitted to use any grade fuel oil with a sulfur content equal to or below 2.8 per cent. Emissions from this source were shown to impact Upper Buffalo (UPBU) and Caney Creek (CACR) Wilderness areas' visibility (Figure 1).

Entergy  -  Lake Catherine (AFIN 30-00011) is located in Malvern, Hot Spring County, Arkansas and is currently permitted to operate under ADEQ Operating Air Permit Number 171-AOP-R5. Lake Catherine is a four unit electric generating station which generates electric energy for sale. Electricity for sale is produced by using natural gas as their primary and No.6 fuel oil as the secondary fuel in both boilers. The boilers produce steam to drive turbines which turn the electric generators. The subject-to-BART source is Unit 4 (SN-03) which is a Combustion Engineering tilting tangential fired 5,850 million BTU per hour boiler powering a 552 MW generator. The primary fuel is natural gas with No.6 fuel oil as the secondary fuel. BART determination modeling indicated emissions from this unit impacts the visibility at Hercules-Glades (HEGL), MO and Upper Buffalo (UPBU) and Caney Creek (CACR), AR (Figure 1). 

Entergy  -  White Bluff (AFIN 35-00110) is located in Redfield, Jefferson County, Arkansas and is currently permitted to operate under ADEQ Operating Air Permit Number 0263-AOP-R6. Units Nos. 1 (SN-01) and 2 (SN-02)are identical Combustion Engineering tilting tangential 8950 million BTU per hour coal fired boilers with a maximum power rating of 850 MW each. The boilers use sub-bituminous or bituminous coal as the primary fuel and No.2 fuel oil as a start-up fuel. Particulate matter is controlled by an electrostatic precipitator on each boiler. The Auxiliary Boiler (SN-05) is a 183 million BTU per hour boiler burning No. 2 fuel oil as its only fuel type. The purpose of the Auxiliary Boiler is to provide steam for the start-up of the two primary boilers, SN-01 and SN-02. Results from the BART determination modeling indicated emissions from Units 1 and 2 and the auxiliary boiler impacted visibility at Hercules-Glades (HEGL), MO and Upper Buffalo (UPBU) and Caney Creek (CACR), AR (Figure 1).

Domtar  -  Ashdown (AFIN 41-00002) is located in Ashdown, Little River County, Arkansas and is currently permitted to operate under ADEQ Operating Air Permit Number 0287-AOP-R10. Domtar is a paper mill facility and has two Power Boilers, No. Power Boiler (SN-03) and No. 2 Power Boiler (SN-05), that are subject-to-BART. The No.1 Power Boiler was installed in 1968 as part of the original construction of the Ashdown Mill. It has a heat input rating of 580 million BTU per hour and an average steam generating rate of 120,000 pounds of steam per hour (lb/hr) at 850 psig. It combusts primarily bark, but it is also permitted to burn wood chips, wood waste, recycled sanitary products composed of cellulose and polypropylene, pelletized paper fuel (PPF), tire derived fuel (TDF), municipal yard waste, No. 6 fuel oil, reprocessed fuel oil, used oil generated on site, and natural gas. Natural gas is only used to supplement other fuels during high steam demand periods. The No.1 Power Boiler is equipped with a traveling grate and a combustion air system. To meet applicable Boiler MACT PM emissions standard of 0.07 lb/MMBTU Domtar Industries installed a wet electrostatic precipitator (WESP) during the spring of 2007. The No. 2 Power Boiler started operation in February 1976. It has a heat input rating of 820 million BTU per hour and an average steam generating rate of approximately 600,000 lb/hr. It combust primarily bituminous coal (over 80 percent of the heat input is supplied by coal), but it is also permitted to burn bark, bark and wood chips used to absorb oil spills, wood waste, petroleum coke, recycled sanitary products based on cellulose and polypropylene, PPF, TDF, municipal waste, No. 6 fuel oil, reprocessed fuel oil, used oil generated on site, natural gas, and non-condensable gases (NCGs). The NCGs are produced in the pulp and evaporator areas. It consist of nitrogen, total reduced sulfur (TRS) compounds, methanol, SO2, and minor quantities of other compounds such as methyl ethyl ketone (MEK). Under normal conditions, natural gas is not combusted. The No.2 Power Boiler is equipped with a traveling grate, combustion air system including overfire air, multi-clones, and two parallel venture scrubbers. The SO2 loading to the boiler is significant since the boiler burns coal and NCGs. Therefore, the scrubbing fluid includes water and a source of alkali, such as sodium hydroxide (NaOH) and/or pulp mill extraction stage filtrate. BART determination modeling indicated emissions from the two Power Boilers impacted visibility at Upper Buffalo (UPBU) and Caney Creek (CACR), AR. 

Georgia-Pacific Paper (AFIN 02-00013) is located in Crossett, Ashley County, Arkansas and is currently permitted under ADEQ Operating Air Permit Number 0597-AOP-R14. Georgia-Pacific is a Kraft paper mill that has two subject-to-BART sources, 6A (SN-19) and 9A (SN-22) boilers. The 6A Boiler is a 357 million BTU per hour boiler. The boiler burns natural gas and specification grade oil. Specification grade oil consists of new oil, used oil, and pitch from the production of tall oil. The boiler was installed in 1962. There are no emissions controls associated with the boiler. The 9A Boiler is a 720 million BTU per hour combination fuel boiler that is used to generate steam for general use throughout the facility. The boiler may serve as a backup combustion unit when the incinerator (SN-83) is offline. The boiler was installed in 1973. The combination of fuels permitted for the boiler are tire derived fuel (TDF), agriculture derived fuel (ADF), refuse derived fuel (RDF), non-condensable gases (NCGs), wood waste, specification grade oil, natural gas, and sludge. The 9A Boiler is equipped with a wet Venturi scrubber to control sulfur compound emissions. The scrubber was installed in 1980. ADEQ determined 6A boiler was pre-BART and emissions from 9A boiler did not cause or contribute to visibility impairment at Caney Creek Wilderness area, AR. However, in the final rule on the AR RHR SIP, EPA found the 6A boiler to be BART eligible. EPA also found both the 6A and 9A boilers to be subject-to-BART and a full BART analysis is required (FR 14606).  However, Georgia-Pacific voluntarily reduced 9A boiler's SO2 emission rate by 70.7%. Modeling performed by them indicates the current emission rate impacts Caney Creek below 0.5 dv.  Based on a call with EPA, Region 6, the current permit limit for the 9A boiler exempts this facility from performing a five-factor analysis.

5. Visibility Impact Modeling (Appendix Y § IV.D.5)
All technically feasible options must be modeled for visibility improvement over baseline and all modeling files must be included in the analysis.  The BART Guidelines state: `In identifying "all" options, you must identify the most stringent option and a reasonable set of options for analysis that reflects a comprehensive list of available technologies." It is not necessary to model all permutations of available control levels that exist for a given technology  -  the modeling is complete if it includes the maximum level of control each technology is capable of achieving.

An approvable approach to visibility modeling is to use EPA approved CALPUFF version 5.8.  The current version of CALMET approved for regulatory action is CALMET version 5.8. The NO OBS = 0 option is the recommended CALMET setting in order to utilize surface and upper air station observations.  An existing CALMET dataset (with NO OBS = 0) utilized by contractors and EPA for visibility modeling in another state is available for modeling sources in Arkansas.  This meteorological data was prepared using CALMET 5.53a which is an earlier regulatory version, compatible with CALPUFF 5.8. There were some deviations from the EPA guidance when this meteorological data was prepared. However, given the time and resource constraints, this dataset is available for your use and EPA believes it to be acceptable for the purpose of evaluating visibility benefits due to the use of controls as part of a full five-factor BART analysis for subject-to-BART sources in Arkansas. The use of these versions will require remodeling of the baseline. As a result of the change from NO OBS = 1 to NO OBS = 0, the 8th highest change in deciview may be used in both the baseline and post-control results. Additionally, for the CALPOST run, method 8 rather than method 6 is the preferred method.  Table 2 lists the CALPUFF modeling components to be used. The protocol for the development of the hybrid CALMET is located in Appendix 1. Appendix 2 contains the CALMET input control parameters. 

Table 2 CALPUFF modeling components
Processor
Version
Level
TERREL
    3.3
030402
CTGCOMP
    2.21
030402
CTGPROC
    2.63
050128
MAKEGEO
    2.2
030402
CALMET
    5.53a
040716
CALPUFF
    5.8
070623
POSTUTIL
    1.3
030402
CALPOST
    6.221
080724


As stated above, the use of the different versions of the CALPUFF modeling components will require remodeling of the baseline. The baseline years of 2001, 2002, and 2003 remain the same. The 24-hour highest emissions used in the BART determination modeling should be used. If the emissions were not from a Continuous Emissions Monitor (CEM), the source must include in their analysis documentation and justification for the emissions.  In the remodeling of the baseline, the source stack parameters used in the BART determination modeling should be used. For the post-control runs be sure to include a description of the source, LCC coordinates of each stack, and stack parameters in metrics reflective of each control 

For both the baseline remodeling and the post-control modeling, ADEQ is requiring the subject-to-BART sources to use the CALPUFF and POSTUTIL control input parameters found in the June 7, 2006 draft Best Available Retrofit Technology (BART) Modeling Protocol. The CALPOST control input parameters for visibility background is to remain the same except for the background concentrations.  In the subject-to-BART determination modeling, ADEQ used the CALPOST control input for Method 6 (MVISBK=6). This Method is based on the original IMPROVE equation. However, ADEQ is recommending the use of Method 8 (MVISBK=8) which is based on the revised IMPROVE equation.  The use of Method 8 will change the background concentrations of each Class I area modeled.  Table 3 lists the background concentrations to be used in the remodeling of baseline and the post-control modeling. 

Table 3 Annual average natural background concentration (ug/m[3]) (FLAG 2010)

                       CALPOST control input parameters

                                     BKSO4
                                     BKNO3
                                     BKOC
                                     BKEC
                                    BKSOIL
                                     BKCM
                                    BKSALT
Class I Area
                               Ammonium sulfate
                               Ammonium nitrate
                                Organic carbon
                               Elemental carbon
                                     Soil
                                  Coarse mass
                                   Sea salt
Caney Creek
                                     0.23
                                     0.10
                                     1.80
                                     0.02
                                     0.50
                                     3.00
                                     0.03
Upper Buffalo
                                     0.23
                                     0.10
                                     1.80
                                     0.02
                                     0.50
                                     3.00
                                     0.03
Hercules-Glades
                                     0.23
                                     0.10
                                     1.80
                                     0.02
                                     0.50
                                     3.00
                                     0.02
Mingo
                                     0.23
                                     0.10
                                     1.83
                                     0.02
                                     0.51
                                     3.05
                                     0.04

EPA in their Guidance on the Use of Models and Other Analyses for Demonstrating Attainment of Air Quality Goals for Ozone, PM2.5, and Regional Haze states that the use of either the IMPROVE or the revised IMPROVE equation is acceptable provided that the same algorithm is utilized for both the base and future extinction calculations.  According to EPA, Region 6 the revised IMPROVE extinction equation "is appropriate and the more preferred method for analyses. . ."

All BART-eligible or subject-to-BART units at a facility and pollutants are to be modeled together. Although EPA found the PM emissions reasonable, PM must be included in the modeling.  The approved PM BART emission limit should be modeled in the baseline and the control case.  We note that in some cases the addition of SO2 controls may increase PM emissions and possibly may require the reevaluation of the PM emission limit.  Post-control emission rates are to be calculated based on the evaluation of the control effectiveness of the available and technically feasible control options.

Compare the pre-control and post-control scenarios. You may include tables and charts comparing the number of days pre- and post-control impact > 0.5 dv and the number of days pre- and post-control impact >= 1.0 dv and the percent improvement for each control option modeled.

6. IMPROVE Equations
The original IMPROVE equation converts particulate matter (PM) species concentrations to light extinction (bext) as follows:
      bext = 3 * f(RH) * [sulfate] + 3* f(RH) * [nitrate] + 4 * [organic carbon] + 10 * [elemental carbon] + 1 * [fine soil] + 0.6 * [coarse mass] + 10 

Where, f(RH) is a water growth factor for sulfate and nitrate; its value depends on relative humidity (RH), ranging from 1 at low humidity to 18 at 98% humidity.  Brackets ([]) represent the concentrations of the PM species measured in micrograms per cubic meter (ug/m[3]).  The constants are the individual component's extinction efficiency. The 10 that is added accounts for Rayleigh scattering, which is due to the interaction of light with molecules of air itself with no pollutants and is measured in inverse mega meters (Mm[-1]). 

In 2007, the IMPROVE workgroup published a more robust algorithm for calculating background visibility. The revised IMPROVE light extinction equation is expressed as follows:

      bext = 2.2 * fs(RH) * [small sulfate] + 4.8 * fL(RH) [large sulfate] + 2.4 * fs(RH) * [small nitrate] + 5.1 * fL (RH) *[large nitrate] + 2.8 * [small organic mass] + 6.1 * [large organic mass] + 10 * [elemental carbon] + 1 * [fine soil] + 1.7 * fss(RH) * [sea salt] + 0.6 * [coarse mass] + Rayleigh scattering (site-specific) + 0.33 * [NO2(ppb)] 

Sulfate, nitrate, and organic mass are each split into two fractions representing small and large distributions of those species.  Though not explicitly shown in the equation, the organic mass concentration used in this new algorithm is 1.8 times the organic carbon mass concentration, changed from 1.4 times carbon mass concentration as used for input for the original IMPROVE algorithm. Sea salt and light absorption by nitrogen dioxide (NO2) which is measured in parts per billion (ppb) have been added.  Distinct water growth curves for small sulfates and nitrates, large sulfates and nitrates, and sea salt have also been added. Site-specific Rayleigh scattering is calculated for the elevation and annual average temperature of each of the IMPROVE monitoring sites compared to the original equation that assumed extinction due to Rayleigh scattering was 10 Mm[-1].
The apportionment of the total concentration of sulfate compounds into the concentrations of the small and large size fractions is accomplished using the following equations.
      [Large sulfate] = [Total sulfate]/20 * [Total sulfate], for [Total sulfate] < 20 ug/m[3]
      [Large sulfate] = [Total sulfate], for [Total sulfate] >= 20 ug/m[3]
      [Small sulfate] = [Total sulfate]  -  [Large sulfate]

The same equations are used to apportion total nitrate and total organic mass concentrations into the small and large size fractions.

After comparison of IMPROVE monitoring data with nephelometer direct measurements of visibility extinction, the IMPROVE program found the original IMPROVE equation over-predicts low extinction conditions and under-predicts for high extinction. These biases have direct relevance for estimates for the 20% worst visibility days that are used to assess visibility impact. The new algorithm shows broader scatter overall, but less bias in matching visibility measurement under high and low visibility conditions.  The main factor in reducing bias is the split between small and large particles. The revised IMPROVE equation has less bias, is more refined, accounts for more pollutants, incorporates more recent data, and is based on considerations of relevance for the calculations needed for assessing progress under the Regional Haze Rule.  
EPA guidance states that the use of either the original IMPROVE or the revised IMPROVE equation is acceptable provided that the same algorithm is used for both the base and future extinction calculations. Because Arkansas subject-to-BART sources are required to remodel the baseline and EPA believes the revised equation is the more preferred method, ADEQ is recommending the use of the revised equation in both the baseline and post-control modeling.
7. CSAPR and Arkansas
On May 30, 2012, EPA finalized a rule that allows the trading programs in the Cross-State Air Pollution Rule (CSAPR) to serve as an alternative to determining source-by-source Best Available Retrofit Technology (BART). This rule provides that states in the CSAPR region can substitute participation in CSAPR for source-specific BART for sulfur dioxide and/or nitrogen oxides emissions from power plants. This is commonly referred to as CSAPR being "better than BART." EPA also determined "that a state in the Transport Rule region whose EGUs are subject to the requirements of the Transport Rule trading program only for ozone season NOx is allowed to rely on our determination that the Transport Rule makes greater reasonable progress than source-specific BART for NOx."  This determination applies to Arkansas and thus Arkansas's subject-to-BART EGUs will not be required to perform a five-factor analysis of NOx emissions. 














                          APPENDIX 1: CALMET Protocol



\s














                    Appendix 2:  CALMET Control Parameters



\s
