Overview
The modeling analysis determined the extent and significant contributors to predicted 1-hour sulfur dioxide (SO2) exceedances in the Muscatine area.  Model predictions regarding the culpability of facilities in the area to the predicted exceedances were used to develop SO2 emissions control measures for the affected facilities.  The control measures show modeled attainment of the 1-hour SO2 National Ambient Air Quality Standard (NAAQS).  This section summarizes the modeling analysis conducted for the attainment demonstration.  
Model Selection and Options 
Air Quality Model Selection:  The dispersion model used for this analysis was the American Meteorological Society/Environmental Protection Agency Regulatory Model (AERMOD).  The most current version of AERMOD available during each phase of the analysis was used.  Preliminary analyses were not reevaluated with the newest version of AERMOD. The final control strategy analysis utilized version 14134.    
All analyses were conducted with EPA's regulatory default options.  The modeling analysis was conducted using the latest EPA guidance available for evaluating SO2 impacts.  This guidance includes the August 23, 2010 1-hour SO2 clarification memo; the applicable portions of the March 1, 2011 1-hour NO2 clarification memo; and the applicable portions of the December 2013 SO2 Technical Assistance Document (TAD).
Receptor Grid/Spacing/Terrain Elevations:  The receptor grid was centered on the Musser Park monitor, and extended out to the edges of the nonattainment area.  The grid utilized 50-meter receptor spacing along property boundaries and extending out to 0.5 kilometers from the Musser Park monitor, 100-meter spacing out to 1.5 kilometers, 250-meter spacing out to 3 kilometers, 500-meter spacing out to 5 kilometers, and 1000-meter spacing beyond 5 km.  The receptor grid encompassed the entire nonattainment area.  
Those portions of the fence lines of the facilities being evaluated that fall outside of the nonattainment area were omitted from the analysis.  Finer grid spacing of 50 meters was used to resolve modeled impacts around other nearby individual facilities included in the analyses, but only within the confines of the nonattainment area.  Receptors were excluded from areas within the property boundaries of each facility in the analysis.
The most recent version of AERMAP (11103) was used to import terrain and source elevations from the National Elevation Dataset (NED).  
Downwash: All building downwash analyses were conducted using the most recent version (04274) of EPA's Building Profile Input Program with Plume Rise Enhancements (BPIP-Prime).
Meteorological Data:    A detailed representivity analysis to support the use of Davenport meteorological data was previously approved by EPA for use in the PM2.5 Muscatine SIP analysis.  Modeling for the SO2 Muscatine SIP submittal was conducted using the surface station and upper air data from Davenport, and used consecutive years from 2008-2012.  This represents the most recent, readily available 5-year period at the time of the initial analysis per section 8.3.1.2 of 40 CFR Part 51 Appendix W.  
The most current version of AERMET available during each phase of the analysis was used.  The data used in the preliminary analyses were not updated with the newest version of AERMET.  The final control strategy analysis utilized data processed with AERMET version 14134.  
Modeled Sources 
Source Inventory:  A modeling analysis was conducted in support of the Technical Support Document (TSD) developed by the Department in April, 2013 to evaluate proposed SO2 non-attainment boundaries in Muscatine County.  That analysis demonstrated that industrial sources along the Mississippi River have a role in causing or contributing to monitored exceedances at the Musser Park SO2 monitor.  Based on this analysis, all major sources of SO2 emissions within the nonattainment area were included in the SIP analysis.  These sources include Grain Processing Corporation (GPC), Muscatine Power and Water (MPW), and Monsanto.  
The MidAmerican Energy  -  Louisa Generating Station (LGS) facility is located immediately south of the nonattainment area.  This source was shown to be insignificant at the Musser Park monitor during predicted exceedances, but it is possible that it could cause a significant concentration gradient in the vicinity of the southern portion of the nonattainment area.  For this reason, LGS was also included in the modeling analysis.  There are two other major sources of SO2 emissions in the vicinity of the Muscatine SO2 nonattainment area, Gerdau Ameristeel (Gerdau) and SSAB, but they were both screened out of the analysis due to their low SO2 emissions and distance from the nonattainment area.
The relative locations of the sources in the modeling inventory are shown in Figure 1. All emission units were modeled using their actual stack parameters and site layout.  There were no stacks above formula GEP height.  There are stacks greater than 65 meters at GPC, MPW and Louisa.  The tallest stack at Monsanto is 49 meters.  Each stack that is taller than 65 meters is adjacent to tall buildings making the formula height taller than the actual stack height (Table 1).  Therefore each of these stacks were modeled at their actual stack heights.
Figure 1. Facility Locations

Table 1. Tall Stack GEP Height Determination.
Facility
Emission Point
Actual Height
(meters)
GEP Formula Height
(meters)
GPC
EP001.0
66.75
82.95
MPW
EP70
67.06
81.32

EP80
68.58
81.32

EP90
91.44
131.34
LGS
EP1
185.93
194.66
Intermittent Sources:  Per EPA's March 1, 2011 clarification memo all emission units that operate intermittently were excluded from the analysis.  Emergency engines and fire pumps that operate intermittently were excluded.  Additionally, emission units that are limited to burning a specific fuel intermittently were modeled at emission rates that represent the fuel that is burned during normal operations.  The two auxiliary boilers (EP2 and EP3) at Louisa are limited to burning fuel oil for no more than 48 hours per year.  These two sources were modeled at emission rates associated with burning natural gas.
Background Value
Temporally varying background concentrations by hour and season from the Davenport SO2 monitor were used to account for contributions to the predicted impacts from background SO2 sources.  The background values account for emissions from natural sources, major and minor sources not included in the analysis, and unidentified sources.  The existing statewide SO2 monitoring network was evaluated for an appropriate background (Table 2).
Table 2. Comparison of Existing SO2 Monitoring Sites.
                               Monitor Location
                    2011-2013 Monitor Design Concentration
                                     (ppb)
                       2013 County-wide SO2 Emissions[A]
                                    (tons)
                           Proximity to SO2 Sources
                                 Cedar Rapids
                            23 (average of 2 sites)
                                     6,419
                               Adjacent / 1+ km
                                    Clinton
                                      38
                                     3,642
                                   Adjacent
                                   Davenport
                                      15
                                     4,771
                                     1+ km
                                  Des Moines
                                       1
                                      163
                                     5+ km
                                  Lake Sugema
                                       3
                                      ~0
                                   >10 km
                                   Muscatine
                                      217
                                    20,086
                                   Adjacent
                              Default (Tier 1)[B]
                                     10.5
                                      ---
                                      ---
[A]	Includes major and minor point sources.
[B]	Average of 2011-2013 design values for Cedar Rapids, Davenport, Des Moines, and Lake Sugema.

All major sources in Muscatine County were explicitly included in the modeling analysis except for Gerdau and SSAB.  Both of these sources are approximately 8-9 km away from the non-attainment boundary.  These two sources combined accounted for 254 of the 20,086 tons of SO2 emitted in Muscatine County in 2013.  The next closest major sources not included in the modeling analysis are Linwood and Lafarge, which are located in Scott County, approximately 20 km away from the non-attainment area.  These two distant sources accounted for 1,539 of the 4,771 tons of SO2 emitted in Scott County in 2013.  In addition, the most recently available emissions data from all minor sources in Muscatine County totaled approximately 2 tons/yr.  The ideal background would represent the contribution from these major and minor sources as well as any unknown or natural sources, while at the same time not include the contribution from the sources being explicitly modeled.
The Muscatine monitors are impacted heavily by sources that are being explicitly included in the modeling analysis.  As such those monitors were eliminated as an option to represent the background concentrations in the area.  Of the remaining monitor locations, two are situated adjacent to industrialized areas (Cedar Rapids and Clinton), and could overestimate the concentrations caused by distant major sources.  The Des Moines and Lake Sugema monitors are impacted by less SO2 emissions than need to be represented by the background for the nonattainment area.  Therefore, the Davenport monitor was evaluated to determine if it would be a representative site.
The Davenport monitor is the nearest location other than those in Muscatine.  The Davenport monitor is near a moderately industrialized area, but is not situated adjacent to those sources of emissions.  It is approximately 1 km from the nearest industry and 11 km from Linwood and Lafarge.  It is in a county with a moderate amount of SO2 emissions.  For these reasons it could account for the sources screened out of the control strategy, major sources in Scott County that may impact the Muscatine area (such as Lafarge and Linwood), minor sources in Muscatine County (approximately 2 tons), and more distant sources that could impact the Muscatine area (as measured at Lake Sugema).  In addition, using this monitor is consistent with the meteorological data used for the analysis.
To account for seasonal and diurnal variations in the background levels, the background concentration was based on the average diurnal and seasonal concentration pattern observed at the Davenport monitor during the years 2011-2013.  As referenced in Appendix A of the Guidance for 1-hour SO2 Nonattainment Area SIP Submissions, "an appropriate methodology for calculating temporally varying background monitored concentrations by hour and season" is outlined in the SO2 NAAQS Designations Modeling TAD and in the March 2011 EPA guidance for 1-hour NO2 Modeling.
For the years 2011 through 2013, the 99th percentile monitor concentration was calculated for each season and hour of day and averages of these values across the three years (Figure 2).
Figure 2: Davenport Monitor Concentrations by Season and Hour of Day


Only one hour in the winter approaches the overall 99[th] percentile design value for the Davenport monitor and the default (Tier 1) background value of 10.5 ppb is higher than all but one of the seasonal/diurnal concentrations.  This shows that the use of the default (Tier 1) value for all hours and seasons would have been very conservative.  The method of using temporally varying background monitor concentrations by hour and season from the Davenport monitor is still conservative as it is calculated from the 99[th] percentile versus an unbiased estimate from average values, which would yield a much lower background value.
Modeling Analysis Methodology and Results
The modeling analysis was conducted in multiple iterations as part of two distinct phases.  The first phase of the analysis was a screening analysis to determine the sources that needed to be included in the analysis.  The second phase of the analysis was used to develop the control strategy and included all significant sources identified for inclusion in the analysis from Phase 1. 
Phase 1  -  Preliminary Analysis:  This phase was accomplished by modeling actual emissions from each facility and then determining what percentage of predicted NAAQS exceedances each facility significantly contributed to.  This information was further subdivided by classifying the significant contributions as either primary or secondary contributors.  If the facility's significant contribution to the predicted NAAQS exceedance was greater than or equal to half of the total concentration (minus background) it was considered the primary contributor.  If the facility's contribution was less than half of the total concentration, but still more than the Significant Impact Level (SIL) it was considered a secondary contributor.
Each iteration of this phase implemented new information received from the facilities during the course of the analysis.  The last of the iterations of Phase 1 represented the best picture of predicted concentrations caused by actual emissions from each facility.  The results of this analysis are summarized in Figure 3 below.

Figure 3. Culpability Analysis Results

GPC was identified as the primary contributor to all predicted NAAQS exceedances within the nonattainment area.  MPW, Louisa, and Monsanto also all had significant contributions at multiple predicted exceedances.  Both SSAB and Gerdau were evaluated at their allowable emission rates, and were insignificant at each predicted NAAQS exceedance.  As such, these two sources were not included in the second phase of the analysis.
Phase 2  -  Control Strategy Development:  Sources identified in Phase 1 were modeled at their maximum permitted allowable emission rates.  Using the process summarized below, more restrictive maximum permitted emission rates were developed where necessary to ensure modeled attainment.
GPC was provided with a model input file that included its emission units as well as the exceedance receptors to which it contributed.  GPC reviewed the input data for accuracy and then mitigated all modeled exceedances caused by their facility alone.  Since GPC was the primary contributor at every predicted exceedance, subsequent iterations of this phase were completed after GPC's initial control strategy was submitted.
The remaining facilities were then added to the analysis with their maximum permitted allowable emission rates and the cumulative impacts were determined across the entire nonattainment area.  Any remaining predicted exceedances were then discussed with the affected facilities and additional control measures were developed.
Monsanto proposed to decrease the emission rate for Boiler 8 at their facility to mitigate a small number of exceedances just north of their property.
MPW proposed multiple operational scenarios for their three main boilers (Units 7, 8, and 9).  The model results varied depending on which combination of boilers was running.  The modeling results for each scenario (with background included) are summarized in Table 3.
Table 3.  Comparison of Cumulative SO2 Modeling Results to NAAQS for MPW Operating Scenarios   
                                 MPW Scenario
                            Cumulative Model Result
                                  (ug/m[3])
                               1-hour SO2 NAAQS
                                  (ug/m[3])
                                      All
                                    181.35
                                      196
                                    U9 Off
                                    181.91
                                       
                                    U8 Off
                                    182.05
                                       
                                    U7 Off
                                    181.99
                                       
                                    U7 Only
                                    180.94
                                       
                                    U8 Only
                                    181.80
                                       
                                    U9 Only
                                    180.41
                                       

These results indicate that all currently proposed operational scenarios at MPW combined with all mitigation strategies at each facility will result in attainment of the 1-hour SO2 NAAQS.
