MEMORANDUM
To:
Margaret Sheppard, U.S. EPA
CC: 
Rachel Schwartz, Sally Hamlin, Rebecca von dem Hagen, Bella Maranion, Cindy Newberg, U.S. EPA
From:
Kasey Knoell, Becky Ferenchiak, Jenny Tanphanich, Ed Carr, Mark Wagner, ICF International
Date:
February 24, 2014
Re:
Assessment of the Potential Impact of Hydrocarbon Refrigerants on Ground Level Ozone Concentrations (EPA Contract Number EP-W-10-031 Task Order 305, Task 01)

In response to technical direction received from EPA, ICF has prepared an assessment of the potential impact on ground level ozone concentrations caused by release of hydrocarbon refrigerants across refrigeration and air conditioning end uses. Refrigerant emissions were modeled from the U.S. EPA's Vintaging Model (VM_IO file_V4.4_11.6.12) and their impacts on ground level ozone were modeled using the CMAQ model (version 4.7.1). Releases of isobutane and propylene were modeled in Los Angeles, Houston, and Atlanta under three emissions scenarios: 1) all refrigeration and air conditioning (ref/AC) end-uses, 2) ref/AC excluding motor vehicle air-conditioners (MVACs), and 3) ref/AC excluding MVACs and chillers. A final scenario was modeled, in which limitations on the properties of hydrocarbons (e.g., flammability) are considered to determine which end-uses were likely to transition to hydrocarbon refrigerants. The remainder of this memorandum summarizes the results of this assessment. A quantitative analysis of the uncertainty levels can also be performed by undertaking additional CMAQ simulations if requested by the EPA TOCOR.

Please contact Mark Wagner at 202-862-1155 with any questions or comments.

Potential Impacts of Hydrocarbon Refrigerants on Ground-Level Ozone Concentrations
Executive Summary
Hydrocarbons are increasingly being considered for use in a number of refrigeration and air-conditioning (ref/AC) end-uses.  Although hydrocarbons have a low global warming potential (GWP) and are thus desirable alternatives to commonly used hydrofluorocarbons (HFCs), there is concern that use of hydrocarbons could negatively impact local air quality due to their high maximum incremental reactivity (MIR) values.  
In order to evaluate their potential impact, three conservative hydrocarbon emission scenarios and one more realistic scenario were analyzed. The first three evaluate emissions of propylene and isobutane first from the entire refrigeration and air conditioning sector, second excluding motor vehicle air conditioning, and third excluding motor vehicle air conditioning and chillers.  A fourth scenario, evaluating emissions of a mix of propylene, isobutane, and propane, was developed based on SNAP hydrocarbon applications and UL Standards and represents a more realistic transition to hydrocarbons. The hydrocarbon emissions from these scenarios were estimated based on U.S. EPA's Vintaging Model, and their potential contributions to ozone concentrations were assessed using U.S. EPA's Community Multiscale Air Quality (CMAQ) model. 
CMAQ modeling was performed for April through the end of September, as these months presented the largest releases of hydrocarbon refrigerant as well as weather conditions favorable for ozone formation. The ozone concentrations were estimated for the Atlanta, Houston and Los Angeles regions, due to their distinctive geographic setting and chronic high levels of ground level ozone, and then scaled for national emission estimates. The results of the CMAQ modeling indicated that hydrocarbon refrigerants under the most conservative scenario could potentially increase ground level ozone by up to 9% compared to the national ozone standard, but less than a 0.2% increase under the most realistic scenario. 
The analysis performed is based on several assumptions and projections that cannot be known with certainty at this time.  These limitations are associated with both the unknown market penetration of alternatives and climate conditions, among other factors.  
Introduction
Ozone-depleting substances (ODS) such as CFC-12, R-502, and HCFC-22 were historically used across the refrigeration and air-conditioning sector. Since identifying the damaging nature of CFCs and HCFCs, EPA moved swiftly to transition to alternatives. The phaseout of CFCs in 1996 and continued phasedown of HCFCs, has led the U.S. EPA Significant New Alternatives Policy (SNAP) program to review and approve a number of potential substitutes to these ODS. In particular, hydrocarbon refrigerants, which have zero ODP and low GWP, have already been recently approved for smaller refrigeration applications (e.g., domestic refrigeration, small retail food refrigeration, and vending machines) and have the potential to be used more broadly across the refrigeration and air-conditioning sector. Compared to the HFCs and HCFCs currently in use, which have high GWPs, hydrocarbons have the potential to significantly reduce climate impacts from the refrigeration and air conditioning sector.
Although the use of hydrocarbons could potentially mitigate greenhouse gas (GHG) emissions, it could also potentially contribute to increased levels of ground level ozone.  Hydrocarbons have high MIR values, and may influence local air quality if emitted in sufficiently high quantities. Table 1 provides a summary of the propensity of these refrigerants to form tropospheric ozone. 
Table 1: Propensity of Refrigerants to Form Tropospheric Ozone
Compound
           North America Maximum Incremental Reactivity (MIR) Scale 
                              (g-O3/g-substance)
                              Europe POCP Scalea 
                               (Relative Units)
HCFC-22
                                    <0.1
                                      0.1
HFC-134a
                                    <0.1
                                      0.1
Propylene
                                     11.57
                                      112
Ethane[a]
                                     0.28
                                      12
Isobutane
                                     1.34
                                      31
Propane
                                     0.56
                                      18
Ethylene
                                     9.07
                                      100
      Source: IPCC 2005
      [a]The POCP value is the ozone creating potential of a compound relative to ethylene (ethene), expressed as an index where ethylene = 100. 
      [b]EPA uses the reactivity of ethane as the threshold for determining whether a compound has negligible reactivity. Compounds that are less reactive than, or equally reactive to, ethane under certain assumed conditions may be deemed negligibly reactive and therefore suitable for exemption from the definition of a VOC (EPA 2012).

It should be noted that EPA is currently considering exempting certain hydrocarbons in certain end-uses from the ban on venting of refrigerant under §608 of the Clean Air Act. In order to evaluate the potential impact of the release of hydrocarbons during disposal, ICF modeled conservative and realistic hydrocarbon emission scenarios from the refrigeration/AC sector and assessed their impacts on ground-level ozone, taking into consideration this revision to the end-of-life management requirement. 
The remainder of this report summarizes the methodology used and results in the following order:
         * Section 3: Hydrocarbon Emission Scenarios
         * Section 4: Impact on Ground-Level Ozone Concentrations
         * Section 5: Analysis Limitations 
         * Section 6: Summary of Findings
         * Section 7: References 
Hydrocarbon Emission Scenarios 
Under this analysis, isobutane and propylene were each modeled under three conservative emissions scenarios for the refrigeration and air conditioning sector. A fourth, more realistic emissions scenario was also modeled, which considered more likely transitions to a variety of hydrocarbon refrigerants (i.e., isobutane, propylene, and propane) by end-use in the sector. The four emission scenarios considered are as follows:  
   1) Scenario 1: the entire ref/AC sector: chillers, motor vehicle air conditioners (MVACs) (i.e., light-duty trucks and vehicles, trains, transit buses, school buses, and tour buses), residential and light commercial AC and heat pumps, retail food, cold storage warehouses, industrial process refrigeration, refrigerated transport, and household refrigeration and freezers; 
   2) Scenario 2: the ref/AC sector excluding MVACs; 
   3) Scenario 3: the ref/AC sector excluding MVACs and chillers; and 
   4) Scenario 4: the ref/AC sector for end-use applications in which a SNAP submission has been received for a hydrocarbon or in which a UL Standard addressing flammable refrigerant is in place. See Appendix B for more information on the end uses included in this scenario.
In order to estimate their potential impact on ground-level ozone concentration, the national scale hydrocarbon emissions from each scenario were estimated using EPA's Vintaging Model (VM). These national emissions were then proportionally scaled to each urban area by population by county. For Los Angeles, three counties were used representing about 4.4% of the U.S. population; for Houston, eight counties representing about 2.0% of the U.S. population were used; and for Atlanta, ten counties representing about 1.4% of the U.S. population were used. These county level emissions were then spatially allocated to each emission grid cell based on population density. The remainder of this section provides a summary of the methodology used to estimate the national inventory of hydrocarbon emissions, as well as a description of each scenario considered. 
Methodology for Emissions Estimates 
The year 2030 was chosen as the analysis year for all scenarios, in order to conservatively estimate the impact of hydrocarbons. The majority of non-hydrocarbon equipment will have been retired by the year, due to the lifetimes of equipment (i.e., 5 to 27 years). In addition, a 2030 nationwide emission inventory was available for other emission sources and was used to estimate the baseline ozone concentration (without hydrocarbon refrigerants).  It was assumed that the hydrocarbon refrigerants would enter the market in 2012, and reach 100% market penetration in 2030. Output from the VM is provided on an annual basis; for the purposes of this analysis, the annual data was weighted on a monthly basis, and assumed that 11.9% of annual emissions occurred during each month from April to August.  Emissions of refrigerant are reasonably anticipated to occur primarily during the warmer spring and summer months, when air conditioning equipment is increasingly used. Emissions due to servicing MVACs are also expected to be higher during the spring and summer, with up to 80% of emissions due to repairs occurring April through August (MACS 2009). Furthermore, refrigerant emissions from servicing motor vehicle air conditioning systems were estimated at 47% of annual refrigerant emissions from motor vehicles (EPA 2008). This assumption was also applied for all scenarios to all stationary AC end-uses based on the assumption that seasonal emissions would follow similar patterns. Emissions of refrigerant from refrigeration equipment were assumed to occur equally year-round. 
Smaller end-uses (i.e., end-uses with a charge size less than 200 kg) were assumed to have a 100% release of refrigerant upon disposal. Disposal release rates for larger end-uses (i.e., chillers, industrial process refrigeration systems, and large retail food systems) and servicing release rates for all end-uses were assumed to remain consistent with current VM assumptions, due to the assumed regulations and financial incentives for the end-user. 
Each state implements individual emission and VOC regulations that inhibit venting. For example, Regulation 7:27-16.1a for New Jersey (which is a nonattainment area) requires the use of reasonably available control technology for equipment. Venting would violate this regulation. Maryland implements a regulation that states that a person may not cause or permit the discharge of VOC from any installation constructed on or after November 15, 1992 in excess of 20 pounds (9.07 kilograms) per day unless the discharge is reduced by 85 percent or more overall. In addition to state regulations, industry standards also restrict venting. Under OSHA regulations, facilities are required to demonstrate that 25 percent of the lower flammability limit (LFL) would not be exceeded at any point during venting of hydrocarbon refrigerants. Chillers and other large charge size commercial refrigeration systems do not currently implement the proper engineering controls that would be needed to allow for safe venting. The use of flares, stacks, or rapid emission controls may likely be required to meet these industry standards. Equipment would have to be modified to incorporate these controls, or venting would not be allowed. Furthermore, assuming 100% is vented during servicing of equipment is unrealistic from an economic perspective.  Even with a low cost refrigerant, such as propane or isobutane, venting a charge size of over 1,000 grams still represents a significant cost and provides an incentive to collect refrigerant, which also is in keeping with good maintenance and technician practices.
The reader is referred to Appendix A for a summary of detailed assumptions used in this analysis. 
Conservative Emissions Scenarios
Hydrocarbons are currently being proposed for use in a wide range of ref/AC end-uses. To estimate the most conservative impact that releases of hydrocarbons could have on ground-level ozone, the first scenario considered under this analysis assumes release from the entire ref/AC sector. However, due to the characteristics associated with hydrocarbons (e.g., flammability), this may not be realistic. To assess more realistic, but still conservative scenarios, two additional scenarios were considered.  All three scenarios assume that the end-uses would be replaced with either propylene or isobutane. Propylene exhibits the highest MIR of hydrocarbons, and therefore represents the most conservative estimates. Propylene is also a proxy for the hydrocarbon blend R-443A, a substitute refrigerant under evaluation for use in residential air conditioning, because that blend contains more than 50% propylene. Isobutane has the highest MIR of the saturated hydrocarbons under evaluation and was used as a conservative proxy for isobutane, propane, and the hydrocarbon blend R-441A. R-441A contains more than 50% propane, as well as isobutane and other saturated hydrocarbons. 
Table 2 presents the monthly emissions in 2030 which were assumed under Scenarios 1, 2, and 3. Under each scenario, 100% of the emissions were assumed to occur from either isobutane or propylene. In other words, the impact on local ozone concentrations were evaluated for six total CMAQ modeling scenarios.  
Table 2: 2030 National Refrigerant Emissions under Scenarios 1, 2, and 3
Month 
                           Scenario1 Emissions (MT)
                                 (All Ref/AC)
                                       
                           Scenario 2 Emissions (MT)
                                   (No MVAC)
                                       
                           Scenario 3 Emissions (MT)
                             (No MVAC or Chillers)
                                       

                            Isobutane or Propylene
                            Isobutane or Propylene
                            Isobutane or Propylene
January
                                     8,065
                                     5,894
                                     5,793
February
                                     8,065
                                     5,894
                                     5,793
March
                                     8,065
                                     5,894
                                     5,793
April
                                    13,404
                                     9,356
                                     9,140
May
                                    13,404
                                     9,356
                                     9,140
June
                                    13,404
                                     9,356
                                     9,140
July
                                    13,404
                                     9,356
                                     9,140
August
                                    13,404
                                     9,356
                                     9,140
September
                                     8,065
                                     5,894
                                     5,793
October
                                     8,065
                                     5,894
                                     5,793
November
                                     8,065
                                     5,894
                                     5,793
December
                                     8,065
                                     5,894
                                     5,793
  Note: Emissions of isobutane and propylene are equivalent; shaded months indicate those used in the analysis.
 Figure 1 presents the percent contribution of isobutane and propylene emissions by end-use under Scenarios 1 through 3. As illustrated in these charts, residential AC end-uses presented the most significant source of emissions across all three scenarios. 
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
Figure 1. Percent Contribution of End-Use to Total Emissions under Scenarios 1-3
                                       
                                       
                                       
Note: Emissions under the residential and light commercial AC and heat pumps end-use (Res and Comm AC) consists of 43% from residential unitary AC, 4% from window AC, 2% from small commercial unitary AC, 0.2% from large commercial unitary AC, 0.2% from PTACs, and 0.1% from water and ground source HPs (relative to total emissions in Scenario 1). Emissions under the MVAC end-use consists of 15% from light-duty vehicles, 14% from light-duty trucks, 0.3% from school and tour buses, 0.1% from transit buses, and 0.1% from trains (relative to total emissions in Scenario 1).
Note: Emissions under the residential and light commercial AC and heat pumps end-use (Res and Comm AC) consists of 61% from residential unitary AC, 6% from window AC, 3% from small commercial unitary AC, 0.3% from large commercial unitary AC, 0.3% from PTACs, and 0.2% from water and ground source HPs (relative to total emissions in Scenario 2).
                                       
                                       
Note: Emissions under the residential and light commercial AC and heat pumps end-use (Res and Comm AC) consists of 63% from residential unitary AC, 6% from window AC, 3% from small commercial unitary AC, 0.4% from large commercial unitary AC, 0.3% from PTACs, and 0.2% from water and ground source HPs (relative to total emissions in Scenario 3).
                                       
                                       
                                         
Realistic Emissions Scenario
In addition to the three end-use scenarios, a fourth more realistic scenario (i.e., Scenario 4) was developed in which hydrocarbons were only modeled in refrigeration and air conditioning equipment for which a SNAP submission was received or for which there is a UL Standard in place regulating flammable refrigerants. Furthermore, the type of hydrocarbon (i.e., isobutane, propylene, or propane) or blend likely to be used in each end-use was also considered.  The reader is referred to Appendix B for additional information regarding the assumptions used under this scenario. Estimated emissions for the most realistic scenario are shown in Table 3.
  Table 3: 2030 National Refrigerant Emissions of under Scenario 4
Month
                           Scenario 4 Emissions (MT)
                        ("Most Realistic Scenario")

                                    Propane
                                   Isobutane
                                   Propylene
January
                                     1,266
                                      78
                                      183
February
                                     1,266
                                      78
                                      183
March
                                     1,266
                                      78
                                      183
April
                                     1,373
                                      84
                                      207
May
                                     1,373
                                      84
                                      207
June
                                     1,373
                                      84
                                      207
July
                                     1,373
                                      84
                                      207
August
                                     1,373
                                      84
                                      207
September
                                     1,266
                                      78
                                      183
October
                                     1,266
                                      78
                                      183
November
                                     1,266
                                      78
                                      183
December
                                     1,266
                                      78
                                      183
  Note: Shaded months indicate those used in the analysis
 Figure 2 presents the percent contributions of each refrigerant and each end-use to the total emissions under the Scenario 4. 
Figure 2. Percent Contribution to Total Emissions under Scenario 4


Note: Emissions under the residential and light commercial AC and heat pumps end-use (Res and Comm AC) consists of 67% from window AC and 3% from PTACs (relative to total emissions in the  Scenario 4).

Impact on Ground Level Ozone Concentrations 
Using EPA's CMAQ photochemical model version 4.7.1, a simulation of the potential ground-level ozone impacts under the hydrocarbon emission scenarios summarized in the previous section was performed. The remainder of this section describes the methodology and results of the CMAQ modeling, followed by a comparison of the results to the national standards of ambient air quality.
Methodology
The projected emissions of isobutane and propylene under Scenarios 1, 2, and 3, and the mix of isobutane, propylene, and propane under Scenario 4 described in Section 3 were input into the CMAQ to predict impacts on the local air quality.  Concentrations were only modeled from April through the end of September, as these months presented the largest releases of hydrocarbon refrigerant (see Table 2 and Table 3), as well as warmer air temperature and longer period of sunlight which increase the potential for ozone creation. 
 The maximum increase in 8-hr ozone concentrations were estimated for the Atlanta, Houston and Los Angeles regions. These three cities were chosen based on their distinctive geographic setting and continuing and chronic problem with high levels of ground level ozone. Local refrigerant emissions were scaled from the national emission estimates to estimate reasonable conservative impacts.
The simulation was performed using 2005 hourly meteorological data at 36-km horizontal resolution with 14 vertical layers to model the maximum increase in 8-hour ground level ozone concentrations. The model used the carbon bond 05 chemical mechanism which explicitly includes 156 chemical reactions (EPA, 2010).  The 2030 nationwide emission inventory (NHTSA, 2012), as previously mentioned, was readily available and used to estimate the baseline ozone concentration (i.e., without hydrocarbon refrigerant). 
Split factors for use in the carbon bond 05 chemical mechanism for isobutane, propylene, and propane were determined from the Emit Worksheet in the emitdb.xls workbook form. The split factors and related parameters used in processing the refrigerant emissions are presented in Table 4 below.
  Table 4. Mass Fractions of CB05 Species per gram of Emitted Compound (g/g)
Compound
                                Paraffin (PAR)
                           Molecular Weight (g/mole)
                                 Olefins (OLE)
                           Molecular Weight (g/mole)
                               Unreactive (UNR)
                           Molecular Weight (g/mole)
Isobutane
                                      1.0
                                     14.54
                                      0.0
                                       --
                                      0.0
                                       --
Propylene
                                      0.34
                                     14.03
                                     0.668
                                     28.06
                                      0.0
                                       --
Propane
                                      0.5
                                     14.70
                                      0.0
                                       --
                                      0.5
                                      14.7
Note: Paraffin compounds are saturated hydrocarbons. Paraffinic compounds consist only of hydrogen and carbon atoms; all bonds are single bonds, and the carbon atoms are not joined in cyclic structures, but instead form an open chain. Olefins are unsaturated chemical compounds containing at least one carbon-carbon double bond. 




CMAQ Results
Table 5 summarizes the maximum increase in 8-hour ground-level ozone from the four scenarios for the Los Angeles, Houston, and Atlanta regions. 
Table 5. Maximum Increase in the Daily 8-hour Ground Level Ozone from Hydrocarbon Refrigerants in 2030 for Three US cities 
                       Refrigerant Type and Use Scenario
                                    Cities

Los Angeles                   8-hr O3 Maximum Change                     (ppb)
Houston                          8-hr O3 Maximum Change                     (ppb)
Atlanta                             8-hr O3 Maximum Change                        (ppb)
Isobutane
 Scenario 1
                                     0.72
                                     0.15
                                     0.05
 Scenario 2
                                     0.50
                                     0.10
                                     0.03
 Scenario 3
                                     0.49
                                     0.10
                                     0.03
Propylene 
 Scenario 1
                                     6.61
                                     0.96
                                     0.56
 Scenario 2
                                     4.47
                                     0.67
                                     0.39
 Scenario 3
                                     4.37
                                     0.66
                                     0.38
Combination of Isobutane, Propylene and Propane (Scenario 4)
 Scenario 4
                                     0.15
                                     0.03
                                     0.01
ppb = parts per billion; O - 3 = ozone
In Los Angeles, isobutane under Scenario 1 resulted in a maximum increase in the 8-hr ozone concentration of 0.72 ppb. Propylene, as anticipated due to its high MIR, resulted in the most significant increases of ground-ozone concentrations; under Scenario 1, results for the Los Angeles region show that the maximum increase in the 8-hr ozone concentration is 6.61 ppb. Releases of isobutane and propylene under Scenarios 2 and 3 each show similar patterns to Scenario 1, but with smaller increases. Scenario 4 (i.e., the "most realistic" scenario) resulted in a peak 8-hr ozone increase of just 0.15 ppb for Los Angeles. These relatively small increases in ozone are about 75% associated with the use of propylene as a refrigerant and 21% from propane under this scenario (this is true of Scenario 4 under all locations). For all four scenarios, Los Angeles showed the largest ground-ozone concentration increases of all three cities.  Figure 3 shows the maximum increases and spatial extent of the 8-hour ozone concentration increase for isobutane Scenario 1, propylene Scenario 1, and Scenario 4 for Los Angeles. In these figures, yellow, orange, and red shading represent modeled changes in  ozone concentrations.
Figure 3.  Maximum 8-hour Ozone Increase for Los Angeles in 2030 
(Clock-wise from Left: Scenario 1 Isobutane, Scenario 1 Propylene, and Scenario 4) 
 
                                       

                                       
 
In Houston, isobutane under Scenario 1 resulted in a maximum increase in the 8-hr ozone concentration of 0.15 ppb. For the propylene Scenario 1, results for the Houston region show that the maximum increase in the 8-hr ozone concentration is 0.96 ppb. Releases of isobutane and propylene under Scenarios 2 and 3 each show similar patterns to Scenario 1, but with smaller increases. Scenario 4 (i.e., the "most realistic" scenario) resulted in a peak 8-hr ozone increase of just 0.03 ppb for Houston.  Figure 4 shows the maximum increases and spatial extent of the 8-hour ozone concentration increase for the isobutane Scenario 1, the propylene Scenario 1, and Scenario 4 for Houston.  In these figures, yellow, orange, and red shading represent modeled changes in ozone concentrations. 

Figure 4. Maximum 8-hour Ozone Increase for Houston in 2030 
(Clock-wise from Left: Scenario 1 Isobutane, Scenario 1 Propylene, and Scenario 4) 
                                       

In Atlanta, isobutane under Scenario 1 resulted in a maximum increase in the 8-hr ozone concentration of 0.05 ppb. For the propylene Scenario 1, results for the Atlanta region show that the maximum increase in the 8-hr ozone concentration is 0.56 ppb. Releases of isobutane and propylene under Scenarios 2 and 3 each show similar patterns to Scenario 1, but with smaller increases. Scenario 4 (i.e., the "most realistic" scenario) resulted in a peak 8-hr ozone increase of just 0.01 ppb for Atlanta.  Figure 5 shows the maximum increases and spatial extent of the 8-hour ozone concentration increase for the isobutane Scenario 1, the propylene Scenario 1, and Scenario 4 for Los Angeles.  In these figures, yellow, orange, and red shading represent modeled changes in ozone concentrations.

Figure 5. Maximum 8-hour Ozone Increase for Atlanta in 2030 
(Clock-wise from Left: Scenario 1 Isobutane, Scenario 1 Propylene, and Scenario 4) 
                                       
                                       
                                       
Comparison to National Ambient Air Quality Ozone Standard
The current national ambient air quality 8-hour average ozone standard (NAAQS) is 75 ppb. The maximum change in ozone concentrations caused by each refrigerant across all scenarios and locations, and comparison to the national standard, is summarized in Table 6. 
Table 6. Comparison of Maximum Increase in 8-hour Ozone Concentration Relative to the NAAQS
                          Refrigerant Type (Scenario)
                                    Maximum
                     8-hr O3 Increase (ppb)               
                            Percent Increase over 
            75 ppb 8-hr National Ambient Air Quality Ozone Standard
Isobutane (Scenario 1)
                                     0.72
                                      1%
Propylene (Scenario 1)
                                     6.61
                                      9%
Most Realistic Mix (Scenario 4)
                                     0.15
                                     0.2%
      ppb = parts per billion; O - 3 = ozone
With the exception of Los Angeles, all of the three regions are forecast to be in attainment of the current 8-hr 75 ppb ozone standard by 2030. Currently, the South Coast is demonstrating attainment by 2023 for the 1997 8-hr standard of 85 ppb. The deadline for demonstrating attainment of the 75 ppb standard is December 2015 and regions must show compliance by 2032. 
Limitations of this Analysis
The analysis described above is based on numerous assumptions and projections that cannot be known with certainty at this time.  The following discussion recognizes these uncertainties and indicates various factors that have potential to impact and change the results of this analysis. 
Market Modelling
         *       The market penetration assumptions used for this analysis assume hydrocarbons will be the sole replacement for HFCs across the refrigeration and air-conditioning sector.  However, the introduction of additional substitutes could potentially lower the penetration of hydrocarbons into the market, thus decreasing the emissions and ground level ozone impacts of hydrocarbons. 
         *       EPA's Vintaging Model provides emissions estimates up to 2050. However, projected estimates are based on assumptions developed using current data and expert opinion regarding the near future, and cannot be reasonably expected to reflect future emissions with 100% accuracy.  Changes that may occur in the distant future (i.e., post-2020), such as regulatory measures and market events, will likely impact future emissions.  
         *       Scenarios 1 through 3 in EPA's Vintaging Model does not account for reductions in emissions that may occur as a result of using hydrocarbons as a replacement for HFCs.  Factors such as refrigerant price, charge size restrictions, and advancements in leak-tightness of ref/AC systems are not adjusted for in the modeling.  Consequently, hydrocarbon emission projections are considered to be conservative.
         *       Although 2030 was chosen at the analysis year for this assessment because the hydrocarbon refrigerants were assumed to reach maximum market penetration by this time, some larger pieces of equipment with long lifetimes (i.e., lifetimes larger than 18 years) would not yet have reached end-of-life by 2030. Therefore, the effects of the disposal emissions from these pieces of equipment are not factored into this analysis. It should also be noted, however, that the remaining refrigerant charge in these large pieces of equipment are also more likely to be recovered given its economic value.
Air Quality Modelling 
   * The air quality modeling used in this study was at 36-km resolution, as used in the Corporate Average Fuel Economy Standards, Passenger Cars and Light Trucks, Model Years 2017 - 2025, Final Environmental Impact Statement (NHTSA, 2012). In Appendix E: Air Quality Modeling and Health Impacts Assessment for Proposed Corporate Average Fuel Economy Standards for Passenger Cars and Light Trucks: Model Years 2017 - 2025, the model performance evaluation is discussed stating that, "EPA evaluated CMAQ model performance for the 2005 base year as part of the air quality modeling analysis to support the rulemaking for light-duty vehicle greenhouse gas emissions standards for MYs 2012-2016 (EPA, 2010).  Overall, the performance for the 2005 modeling platform was similar to that for other national- and regional-scale applications, and the model was able to reproduce historical concentrations of ozone and PM2.5 with low bias and error results."  All meteorological and emission inventory files were taken directly from the NHTSA study. The only modification was to include the emission changes due to refrigerants, as discussed in Section 2. Higher grid resolution (i.e., smaller grid cell) may potentially show higher ozone concentrations due to the mixing of air at the coarser resolution, however most refrigerant emissions are widely dispersed over a grid cell, so the effect of this limitation on the overall modeling results is likely minimal. 
   * Further climate change impacts will result in region-specific changes that may impact ozone production rates. For example, increase (or decrease) in local cloud cover would lead to decrease (increase) local photolysis rates, resulting in reduced (increased) ozone production.
   * It is possible that larger impacts of VOC concentrations could have been noted if additional VOC-limited areas in the U.S. were modeled (e.g., Chicago). However, it is very unlikely that impacts for those areas would exceed that for Los Angeles. While the higher ozone concentrations observed in Los Angeles (8-hour ozone [2008] design value is 112 ppb; Chicago is 77 ppb) depends upon the local VOC-to-NOx ratio, it tends to be more sensitive to a given change in VOC. Therefore, while it is possible that a higher ozone value may have occurred in Chicago than Los Angeles on a given day, it is more likely that the Los Angeles area will have experienced a greater change in ozone due to increased VOCs over the course of the ozone season. 
   * The changes in standards for other pollutants such as air toxics or criteria pollutants could affect the ozone precursor emissions.  For example, EPA recently lowered the 1-hour NO2 standard to 100 ppb and could potentially lower this value again in the future to 80 ppb as initially proposed.  In order to meet this standard, significant NOx emission reductions may be required, thus impacting the potential production of ground level ozone. The ozone air quality standard may be lowered in the future (historically the standard has been lowered a number of times) and would require additional reductions of ozone precursor emissions, which could increase the importance of hydrocarbon refrigerants to the production rate for ozone. 
Summary of Findings
From the analysis summarized in the previous sections of this report, it is concluded that non-attainment resulting from hydrocarbon refrigerant emissions is not likely to be a major concern for local air quality. Hydrocarbon refrigerants could potentially increase ground level ozone by less than 1% under Scenario 4, but up to a 9% increase on a given day for the most reactive hydrocarbons (propylene) in the most extreme case.   However, in most cases this upper bound level of increase is not likely, as most ozone nonattainment areas are not VOC-limited (i.e., the formation of ozone in these areas  are not by limited by VOC emissions, but by other compounds such as nitrogen oxides [NOx]).  In fact, Scenario 4, the "most realistic" scenario for hydrocarbon refrigerant emission usage, showed a less than 0.2% increase in ground level ozone for the most extreme case.  Future changes that could alter these findings include: changes in cloud cover from global warming which could increase or decrease ground level ozone production; and future emission reduction measures that may be required to reduce ozone precursors which would could increase the relative importance of the hydrocarbon refrigerants on the production of ground level ozone.  
References
Intergovernmental Panel on Climate Change (IPCC). 2005. Safeguarding the Ozone Layer and the Global Climate System: Issues Related to Hydrofluorocarbons and Perfluorocarbons. Available at: http://www.ipcc.ch/publications_and_data/_safeguarding_the_ozone_layer.htm. 
MACS.  2009.  Email communication between Elvis Hoffpauir, MACS and Karen Thundiyil, EPA. 10 March.
National Highway Traffic Safety Administration (NHTSA), 2012. Final Environmental Impact Statement, Corporate Average Fuel Economy Standards, Passenger Cars and Light Trucks, Model Years 2017 - 2025, Appendix E: Air Quality Modeling and Health Impacts Assessment. NHTSA, July 2012.
United States Environmental Protection Agency (EPA). 2008. Vintaging Model, Version VM IO File v4.2 10-07-08.
United States Environmental Protection Agency (EPA). 2010. Operational Guidance Document CMAQ 4.7.1 Available at www.cmascenter.org.
United States Environmental Protection Agency (EPA). 2012. "Air Quality: Revision to Definition of Volatile Organic Compounds -- Exclusion of trans-1,3,3,3-tetrafluoropropene." Final Rule. 40 CFR Part 51. EPA-HQ-OAR-2010-0605; FRL-9679-2. June 22, 2012. Available online at: http://www.gpo.gov/fdsys/pkg/FR-2012-06-22/html/2012-15347.htm


 Appendix A
Table A-1. Scenarios 1, 2, and 3 Isobutane and Propylene Modeling Assumptions 
 SNAP End-Use
Vintaging Model End-Use
                              Size Classification
                               Annual Leak Rate
                            Disposal Release Rate^
 Motor Vehicle Air Conditioning (MVAC)
MVAC Light-duty Vehicle*
                                     Small
                                      18%
                                     100%
 
MVAC Light-duty Trucks*
                                     Small
                                      18%
                                     100%
 
School & Tour Buses*
                                     Small
                                      10%
                                     100%
 
Transit Buses*
                                     Small
                                      10%
                                     100%
 
Trains*
                                     Small
                                      2%
                                     100%
 Chillers
Centrifugal Chillers**
                                     Large
                                      2%
                                      10%
 
Positive Displacement Chillers**
                                     Large
                                 1.5% to 2.5%
                                      10%
 
Screw Chillers**
                                     Large
                                      1%
                                      10%
 
Scroll Chillers**
                                     Large
                                     0.5%
                                      10%
 
Reciprocating Chillers**
                                     Large
                                     1.5%
                                      10%
 Industrial Process Refrigeration Systems
Industrial Process Refrigeration
                                     Large
                                  3% to 7.5%
                                  10% to 20%
 Cold Storage Warehouses
Cold Storage
                                     Large
                                   8% to 16%
                                      10%
 Retail Food Refrigeration
Large Retail Food
                                     Large
                                  14% to 28%
                                      10%
 
Medium Retail Food
                                     Small
                                   8% to 15%
                                     100%
 
Small Retail Food
                                     Small
                                      1%
                                     100%
 
Ice Makers
                                     Small
                                      5%
                                     100%
 Refrigerated Transport
Refrigerated Transport
                                     Small
                                  15% to 20%
                                     100%
 Household Refrigerators and Freezers
Refrigerated Appliances
                                     Small
                                      1%
                                     100%
 Residential Dehumidifiers
Dehumidifiers
                                     Small
                                     0.50%
                                     100%
 Residential and Light Commercial Air Conditioning and Heat Pumps
Window Units
                                     Small
                                     0.65%
                                     100%
 
Residential Unitary AC
                                     Small
                                      11%
                                     100%
 
Small Commercial Unitary AC
                                     Small
                                      9%
                                     100%
 
Large Commercial Unitary AC
                                     Small
                                      8%
                                     100%
 
Water & Ground Source HPs
                                     Small
                                      4%
                                     100%
 
Packaged Terminal AC/Packaged Terminal Heat Pumps (PTAC/PTHP)
                                     Small
                                      4%
                                     100%
^The assumed disposal release rates are dependent upon the size of the equipment. Equipment that is considered small (i.e., with a charge size less than 200 kg) are assumed to have a 100% release of the charge upon disposal. Equipment that is considered large (i.e., a charge size larger than 200 kg) has a disposal release rate that is consistent with what is currently assumed in the Vintaging Model for HFC refrigerants.
*Excluded from Scenario 1 and Scenario 2
**Excluded from Scenario 2
 
 Appendix B
Table B-1: Scenario 4 Modeling Assumptions
SNAP End-Use
Vintaging Model End-Use
                                      HC*
                             Market Penetration**
                                  Charge Size
                               Annual Leak Rate
                             Disposal Release Rate
Retail Food Refrigeration
Medium Retail Food
                                    Propane
                                     100%
                               1,500 to 12,500 g
                                   8% to 15%
                                     100%

Small Retail Food***
                                    Propane
                                      33%
                                     150 g
                                      1%
                                     100%


                                    R-441A
                                      33%
                                     150 g
                                      1%
                                     100%


                                   Isobutane
                                      33%
                                     150 g
                                      1%
                                     100%

Ice Makers
                                    Propane
                                     100%
                                     50 g
                                      5%
                                     100%
Refrigerated Transport
Refrigerated Transport
                                    R-441A
                                     100%
                                    2,130 g
                                  15% to 20%
                                     100%
Household Refrigerators and Freezers
Refrigerated Appliances
                                   Isobutane
                                      33%
                                     50 g
                                      1%
                                     100%


                                    Propane
                                      33%
                                     50 g
                                      1%
                                     100%


                                    R-441A
                                      33%
                                     50 g
                                      1%
                                     100%
Residential Dehumidifiers
Dehumidifiers
                                     None
                                     None
                                     None
                                     0.50%
                                     100%
Residential and Light Commercial Air Conditioning and Heat Pumps
Window Units
                                    R-441A
                                      33%
                                     1 kg
                                     0.65%
                                     100%


                                    R-443A
                                      33%
                                     1 kg
                                     0.65%
                                     100%


                                    Propane
                                      33%
                                     1 kg
                                     0.65%
                                     100%

Water & Ground Source HPs
                                    R-443A
                                      33%
                                     1 kg
                                      4%
                                     100%


                                    Propane
                                      33%
                                     1 kg
                                      4%
                                     100%


                                    R-441A
                                      33%
                                     1 kg
                                      4%
                                     100%

PTAC/PTHP
                                    R-443A
                                      33%
                                     1 kg
                                      4%
                                     100%


                                    R-441A
                                      33%
                                     1 kg
                                      4%
                                     100%


                                    Propane
                                      33%
                                     1 kg
                                      4%
                                     100%
*R-441A is a blend consisting of propane, n-butane, isobutane, and ethane; R-443A consists of propylene, propane, and isobutane. For purposes of this assessment, both R-441A and R-443A were considered to only contain the components included in this analysis, propane, isobutane, and propylene.
**In the absence of market penetration projections, it was assumed that the market was equally divided by the number of hydrocarbon substitutes available for each end-use.
***Water coolers and vending machines were assumed to be included under small retail food.
