Significant New Alternatives Policy Program

Refrigeration and Air Conditioning Sector

Risk Screen on Substitutes for CFC-12, R-502, and HCFC-22 in Retail Food
Refrigeration.

Substitute: Carbon Dioxide (CO2)

This risk screen does not contain Clean Air Act (CAA) Confidential
Business Information (CBI) and, therefore, may be disclosed to the
public.

INTRODUCTION

Ozone-depleting substances (ODS) are being phased out of production in
response to a series of diplomatic and legislative efforts that have
taken place over the past decade, including the Montreal Protocol and
the Clean Air Act Amendments of 1990 (CAAA).  The U.S. Environmental
Protection Agency (EPA), as authorized by Section 612 of the CAAA, is
developing a program to evaluate the human health and environmental
risks posed by alternatives to ODS.  The main purpose of EPA's program,
called the Significant New Alternatives Policy (SNAP) Program, is to
identify acceptable and unacceptable substitutes for ODS in specific end
uses.  

EPA’s decision on the acceptability of a substitute is based largely
on the findings of a screening assessment of potential human health and
environmental risks posed by the substitute in specific applications. 
EPA has already screened a large number of substitutes in many end uses
within all of the major ODS-using sectors, including refrigeration and
air conditioning, solvent cleaning, foam blowing, aerosols, fire
suppression, adhesives, coatings and inks, and sterilization. The
results of these risk screens are presented in a series of Background
Documents that are available in EPA's docket.

The purpose of this report is to supplement EPA’s Background Document
on the refrigeration and air conditioning sector (EPA 1994) (hereinafter
referred to as the Background Document) by adding to the list of
potential substitutes for specific end-uses of CFC-12, R-502, and
HCFC-22 in this sector.  The proposed end-use application considered in
this analysis is retail food refrigeration.  The specific proposed ODS
substitute examined in this report is carbon dioxide (CO2, or R-744). 
The substitute would only be used in new equipment.  

The potential risks associated with use of substitutes in retail food
refrigeration have been examined at length in the Background Document. 
In this risk screen, occupational exposure modeling was performed to
ensure that use of the proposed substitute in the application listed
above did not pose unacceptable risk to workers during equipment
manufacture.  Modeling was performed at the end-use to ensure that
potential catastrophic releases of the substitute did not pose
unacceptable risk to store employees and customers in locations where
the refrigeration systems are in use.  Lastly, general population
exposure modeling was performed to ensure that the proposed substitute
would not pose unacceptable risk to the population at large.  Carbon
dioxide is not flammable, so a flammability analysis was not conducted. 
The reader is referred to the Background Document for a detailed
discussion of the methodologies used to conduct this risk screen.

Section 2 of this report summarizes the results of the risk screen for
the proposed substitute.  The remainder of the report is organized into
the following sections:

Section 3: Atmospheric Assessment

Section 4: Asphyxiation Assessment

Section 5: Toxicity Assessment

Section 6: Volatile Organic Compound Assessment 

Section 7: References

SUMMARY OF RESULTS

Carbon dioxide is recommended for SNAP approval for use in retail food
refrigeration systems.  EPA's risk screen indicates that the use of the
proposed substitute will be less harmful to the atmosphere than the
continued use of CFC-12, R-502, or HCFC-22.   No significant
asphyxiation or toxicity risks to workers, consumers, or the general
population are expected according to occupational and end-use exposure
modeling.  EPA recommends that American Society of Heating,
Refrigerating and Air-Conditioning Engineers (ASHRAE) Standards 15 and
34 be followed, as well as Chapter 11 (“Refrigeration”) of the
International Mechanical Code.

ATMOSPHERIC ASSESSMENT

This section presents an assessment of the potential risks to
atmospheric integrity posed by the use of carbon dioxide in the retail
food refrigeration sector.  The ODP, GWP, and atmospheric lifetime (ALT)
of the proposed substitute are presented in Table 2.  The substitute is
substantially less harmful to the ozone layer and has less climate
impact than CFC-12, R-502, or HCFC-22.  (R-502 is a mixture that is
48.8% HCFC-22 and 51.2% CFC-115 by weight.)

Table   SEQ Table \* ARABIC  1 .  Atmospheric Impacts of Carbon Dioxide
Compared to HCFC-22, CFC-115 and CFC-12.

Refrigerant	Ozone Depleting Potential (ODP)	Global Warming Potential
(GWP)	Atmospheric Lifetime years (ALT)

Carbon Dioxide	0 a	1 a	NA b

HCFC-22	0.055 c	1,500 d	12 c

CFC-115	0.44 c	9,300d	1,700 c

CFC-12	1.0 c	8,100 d	100 c

a Carbon Dioxide SNAP Submission (Hill PHOENIX 2009).

b NA = Not Available; Meehl et al. (2007) report that a lifetime for CO2
“cannot be defined.”

c Available at: http://www.epa.gov/ozone/ods.html.

d IPCC, Second Assessment Report (1996).

ASPHYXIATION ASSESSMENT

The risk of asphyxiation for a reasonable “worst-case” scenario was
investigated for carbon dioxide.  This analysis considers release of the
full charge of carbon dioxide into a store. This analysis does not
consider conditions that are likely to occur that would reduce the
levels to which individuals would be exposed, such as open doors or
windows, fans operating, conditioned airflow (either heated or cooled),
or even seepage between the door and door frame.  

The charge of carbon dioxide necessary to reduce the oxygen levels to 12
percent in air, in a supermarket of volume 17,840 m3 (630,000 ft3) (Hill
PHOENIX 2009), was calculated, assuming horizontal stratification of the
refrigerant and the air.  Horizontal stratification is assumed since
carbon dioxide is denser than air and will settle in higher
concentrations closer to the ground.  Assuming that nitrogen and oxygen
retain the same relative volumes in the rooms with the balance composed
entirely of carbon dioxide, and that the pressure of the room does not
increase significantly with the addition of the refrigerant, a charge of
approximately 221 kg would be necessary to reach 12 percent oxygen in
the lower stratum.  This amount represents almost double the intended
charge of approximately 125 kg for a single carbon dioxide retail
refrigeration unit.  Charge requirements to reach the same effect in the
upper strata would be even higher because of the stratum’s larger
volume.  The results of the asphyxiation assessment are summarized in  
REF _Ref225832351 \h  \* MERGEFORMAT  Table 2  below. 

Table   SEQ Table \* ARABIC  2 .  Asphyxiation Assessment.

Room Type/Appliance Type	Appliance Charge (kg)	Reasonable Worst-Case
Scenario	Asphyxiation Threshold Scenario



Room Size (m3)	Charge Causing Impairment (kg)a	Room Size (m3)	Charge
Causing Impairment (kg)a

Supermarket/Retail Refrigeration Unit	125	17,840 (630,000 ft3)	221
10,150 (358,438 ft3)	125

a Values provided in these columns refer to the charge required to cause
impairment in the lower stratum of the room.

For asphyxiation to be of concern (in the lower stratum of the room)
with the proposed charge size, under the conservative (protective)
assumptions described above, the volume of the grocery store would have
to be about 10,150  m3 (358,438 ft3).  Assuming a square room with a
ceiling height of 5.5 m (18 ft), this equates to a 1,845 m2 (19,913 ft2)
supermarket.   For asphyxiation to be a concern in the upper stratum of
the room (where store employees or customers are more likely to be), the
area of the store would have to be 7.5 m2 (81 ft2).  Armines (2008) and
Hill PHOENIX (2009) report average supermarket areas ranging from 3,252
to 4,400 m2.  Therefore, it is considered unlikely that a supermarket
would be small enough for asphyxiation to be a concern in either of the
room’s strata under a worst-case scenario.  Consequently, EPA does not
believe that the use of carbon dioxide in this end-use poses a
significant risk of asphyxiation or impaired coordination to store
employees or customers in the main area of the store.

While the risk of asphyxiation in supermarkets is minimal, because
carbon dioxide is denser than air, there is the potential for
accumulation of CO2 in low-lying and/or small spaces (such as a
mechanical access area or the bottom of a stairwell) which could
inadvertently cause an oxygen-deficient atmosphere.  For example,
assuming the submitter’s suggested volume of 110.7 m3 (3,910 ft3) for
a mechanical room (Hill PHOENIX 2009), there is the potential for the
risk of asphyxiation in a mechanical room following a catastrophic leak
from a system using the proposed charge size (as this room’s volume is
less than the minimum size that leads to impairment for the given charge
– see   REF _Ref175044517 \h  \* MERGEFORMAT  Table 2 ).  However, as
other refrigerants are already used in this end-use, and the
asphyxiation limit would be the same for all refrigerants (i.e.,
reduction of oxygen levels below 12 percent in air), it is believed that
standard industry precautionary measures will mitigate the risk of
asphyxiation.  Such measures may include leak detectors/alarms and
automatic ventilation which is engaged when refrigerant concentrations
reach a certain level.  Also, a small amount of an odorant gas could be
added to the CO2 refrigerant charge which when detected would warn room
occupants to leave the room.  In their SNAP submission, the manufacturer
has indicated they “recommend to [their] customers that systems using
CO2… should be equipped with the same level of leak detection,
ventilation, and alarming [sic] as they would normally apply to HFCs”
(Hill PHOENIX 2009).  ICF recommends that users of CO2 systems continue
to be advised to use these measures, especially if the user’s
mechanical room is smaller than the suggested size given by the
submitter.   It is also recommended that installation and maintenance
occur when store employees and customers are not present in the building
and that signage be installed in any low-lying areas which have access
to the refrigeration system warning that a CO2 charge is present.  If an
odorant is added to the CO2 charge or an alarm is placed in the room,
the signage should also instruct personnel to vacate the area and ensure
customers vacate the store should they smell the odorant or hear the
alarm.

TOXICITY ASSESSMENT

5. 1.  Toxicity Reference Values

To assess potential health risks from exposure to this substitute in the
retail food refrigeration sector, EPA identified the relevant toxicity
threshold values for comparison to modeled exposure concentrations for
different scenarios.  For the occupational exposure analysis, potential
risks from chronic and acute worker exposure were evaluated by comparing
exposure concentrations to available occupational exposure limits. 
Occupational exposure limits are typically established for either an
eight-hour or ten-hour time period for long-term exposure, or for a 10
to 30-minute period for short-term exposure, as shown in   REF
_Ref222734171 \h  \* MERGEFORMAT  Table 3 .  Because they are designed
to assess risks from acute exposure, emergency guidance levels are used
to assess risks from short-term consumer exposures.  The relevant
toxicity limits are shown in   REF _Ref222734171 \h  \* MERGEFORMAT 
Table 3 .    REF _Ref225834278 \h  \* MERGEFORMAT  Table 4  provides
definitions for acronyms used in   REF _Ref222734171 \h  \* MERGEFORMAT 
Table 3 .  EPA’s approach for identifying or developing these values
is discussed in Chapter 3 of the Background Document. 

Table   SEQ Table \* ARABIC  3 .  Toxicity Levels of Carbon Dioxide.

	Long-term Exposure

ppm	Short-term Exposure

ppm

Carbon Dioxide	5000 a 

(OSHA PEL/NIOSH REL)	30,000 a (NIOSH STEL-REL) a

a  http://www.cdc.gov/Niosh/npg/npgd0103.html

Table   SEQ Table \* ARABIC  4 .  Explanation of Toxicity-Related
Acronyms. a

Organization 	Definition

OSHA	Occupational Safety and Health Administration

NIOSH	National Institute for Occupational Safety and Health

Exposure Limit	Definition	Explanation

PEL	Permissible Exposure Limit	This is an 8-hour time-weighted average
exposure limit set by OSHA. 

REL	Recommended Exposure Limit	This is a 10-hour time-weighted average
exposure limit set by NIOSH.

REL- STEL	Recommended Exposure Limit - Short-term exposure limit	This is
a 15-minute time-weighted average exposure limit set by NIOSH.

aAll information in this table taken from EPA (1994).

5.2.  Occupational Exposure 

Occupational exposure modeling was performed for the proposed substitute
to ensure that use of the substitute does not pose an unacceptable risk
to workers.  The methodology used for this screening assessment is based
on the one used in the occupational exposure and hazard analysis
described in Chapter 5 of the Background Document. A box-model approach
was used to evaluate potential worker exposure to alternative
refrigerants.   This approach has been widely used for many years to
estimate probable exposures of workers to hazardous airborne materials,
and has been described in detail by the National Institute for
Occupational Safety and Health (NIOSH).  This model takes into
consideration the duration and magnitude of the resulting exposure which
is influenced by 1) duration and intensity of the release, 2) rate at
which contaminated air is diluted with uncontaminated air, 3) proximity
of the worker to the source of the release, and 4) the length of time
the worker remains in the affected space.  

Estimates of refrigerant release per event for various release scenarios
and data on number of events were obtained from the Vintaging Model.  
The release per event was conservatively assumed to be 1 percent of the
equipment charge during manufacturing and 3 percent of the equipment
charge during disposal.  The release rate per event was multiplied by
the number of events estimated to occur over a workday.  For equipment
installation, the number of events per workday was assumed to equal 1,
as installation occurs on-site and only one full charge size is present.
 For the purposes of this model, it is assumed that carbon dioxide will
have a market penetration rate of 80% of low-temperature refrigeration
systems (Hill PHOENIX 2009).  For disposal, it was assumed that 1 unit
is disposed of during an 8-hour work day.  

The maximum time weighted average (TWA) exposure was estimated for each
exposure scenario, and this value compared to the long-term exposure
level (i.e., “Workplace Guidance Level,” WGL) for carbon dioxide. 
The modeling results indicate that the short-term (15-minute and
30-minute) and long-term (8-hour) worker exposure concentrations are at
a maximum, about 18 percent of the WGL.   REF _Ref225915175 \h  \*
MERGEFORMAT  Table 5  displays the maximum estimated 15-minute TWA
occupational exposure level.   Even this maximum estimated short-term
occupational level is significantly lower than the 8-hour or 10-hour
long-term WGLs, and therefore occupational exposure to carbon dioxide is
not considered a toxicity threat. 

Table   SEQ Table \* ARABIC  5 .  Occupational Risk Assessment.

	Maximum 15-minute TWA Occupational Exposure Levels (ppm)	Workplace
Guidance Level (ppm) a	Workplace Guidance Level Time Period

Carbon Dioxide	878	5000	8-hour TWA b

a. See Table 3 for more information. 

b. The WGL for Carbon Dioxide is set as a 5000 ppm 8-hr TWA by OSHA and
a 10-hr TWA by NIOSH.  

5.3.  End-Use Exposure

	5. 3. 1. Exposure in the Main Supermarket Area

This section presents estimates of potential store employee and customer
exposures to carbon dioxide in retail refrigeration units.  An exposure
analysis was performed to examine potential catastrophic release of the
substitute in an average-sized supermarket.  The analysis was undertaken
to determine the 15- and 30-minute TWA exposures for the substitute,
which were then compared to the standard toxicity limits presented in  
REF _Ref222734171 \h  \* MERGEFORMAT  Table 3  to assess the risk to
consumers.  However, the TWA values are fairly conservative as the
analysis does not consider opened windows, fans operating, conditioned
airflow (either heated or cooled) and other variables that would reduce
the levels to which individuals would be exposed.

The model involves a refrigerant leak from a retail food refrigeration
unit into a supermarket of volume 17,840 m3 (630,000 ft3) (Hill PHOENIX
2009).  The model assumes that the individual is present at the start of
the leak and the individual remains in the store while the refrigerant
is released. It is also assumed that horizontal stratification causes
most of the refrigerant to settle in higher concentrations closer to the
ground.  Exposure concentrations were calculated using the box model
described in the Background Document, which was adapted to estimate
concentrations on a minute-by-minute basis. In the analysis, the full
charge of the unit is assumed to be emitted over the course of five
minutes.  The results of the assessment are presented in   REF
_Ref225848639 \h  \* MERGEFORMAT  Table 6 .

Table   SEQ Table \* ARABIC  6 . End-Use Exposure Assessment – Main
Supermarket Area.

	15-minute TWA Exposure (ppm)	30-minute TWA Exposure (ppm)

Lower Stratum	20,941	12,216

Upper Stratum	87	51

TWA = Time Weighted Average

All calculated exposures are less than the short-term (15-minute TWA)
exposure limit presented in   REF _Ref222734171 \h  \* MERGEFORMAT 
Table 3  (30,000 ppm).  Therefore, exposure to this substitute is not
expected to pose a toxicity threat to store employees or customers.  

	5. 3. 2. Exposure in the Mechanical Room

Another potential concern for exposure at the end-use is installation or
maintenance workers’ exposure while in a small mechanical room which
has access to the refrigerant charge.  Assuming the submitter’s
suggested volume of 110.7 m3 (3,910 ft3) for a mechanical room (Hill
PHOENIX 2009), exposure concentrations would be higher in the mechanical
room than in the main supermarket area under a reasonable worst-case
leak scenario (as this room volume is smaller than that of the main
supermarket area).  However, as refrigerants with lower toxicity limits
are already used in this end-use, it is believed that standard industry
precautionary measures will mitigate toxicity risks.  Such measures may
include leak detectors/alarms and automatic ventilation which is engaged
when refrigerant concentrations reach a certain level.  Also, a small
amount of an odorant gas could be added to the CO2 refrigerant charge
which when smelled would warn room occupants to leave the room. 

ICF recommends that CO2 systems be designed to adhere to the
requirements in Chapter 11 of the International Mechanical Code which
includes charge limits that are based on the acute toxicity levels of
the refrigerant.  Further, in their SNAP submission, the manufacturer
has indicated they “recommend to [their] customers that systems using
CO2… should be equipped with the same level of leak detection,
ventilation, and alarming [sic] as they would normally apply to HFCs”
(Hill PHOENIX 2009).  ICF recommends that users of CO2 systems continue
to be advised to use these measures, especially if the user’s
mechanical room is smaller than the suggested size given by the
submitter.   Additionally, the installation of signage which warns room
occupants to remain standing (as higher concentrations are likely to
accumulate near the floor) and exit the room immediately should they
hear the refrigerant detector alarm and/or smell the leaked refrigerant
would further minimize toxicity risks.  

5.4.  General Population Exposure

Chronic exposures to the substitute are not expected for the general
population. 

VOLATILE ORGANIC COMPOUND (VOC) ASSESSMENT

Carbon dioxide has been exempted as a VOC under the CAA (40 CFR 51.100).


REFERENCES

Armines.  2008.  Inventory of Direct and Indirect GHG Emissions from
Stationary Air Conditioning and Refrigeration Sources, with Special
Emphasis on Retail Food Refrigeration and Unitary Air Conditioning. 
Provisional Final Report.  June 2008.  Available at: <
http://www.arb.ca.gov/cc/commref/armines_report_03_625.pdf>.

EPA. 2000.  Carbon Dioxide as a Fire Suppressant: Examining the Risks. 
February 2000. Available at: <
http://www.epa.gov/Ozone/snap/fire/co2/co2report.html>.

EPA. 1994.  Significant New Alternatives Policy Technical Background
Document:  Risk Screen on the Use of Substitutes for Class I
Ozone-depleting Substances: Refrigeration and Air Conditioning. 
Stratospheric Protection Division.  March 1994.

Hill PHOENIX. 2009. Significant New Alternatives Policy Program
Submission to the United States Environmental Protection Agency. March
2009. 

ICF. 1997. Physiological Effects of Alternative Fire Protection Agents -
Hypoxic Atmospheres Conference. Stephanie Skaggs prepared the
proceedings of the conference held May 22, 1997 in New London, CT.

Kataoka.  1999.  “Allowable Charge Limit of Flammable Refrigerants and
Ventilation Requirements.”  Draft Proposal.  O. Kataoka/Daikin/Japan,
June, 1999.

Meehl, G.A., T.F. Stocker, W.D. Collins, P. Friedlingstein, A.T. Gaye,
J.M. Gregory, A. Kitoh, R. Knutti, J.M. Murphy, A. Noda, S.C.B. Raper,
I.G. Watterson, A.J. Weaver and Z.-C. Zhao.  2007. Global Climate
Projections. In: Climate Change 2007: The Physical Science Basis
.Contribution of Working Group I to the Fourth Assessment Report of the
Intergovernmental Panel on Climate Change [Solomon, S., D. Qin, M.
Manning, Z. Chen, M. Marquis, K.B. Averyt, M. Tignor and H.L. Miller
(eds.)]. Cambridge University Press, Cambridge, United Kingdom and New
York, NY, USA.

 That is, a scenario which is feasible but has a low probability of
occurring.

 Twelve percent oxygen in air is the NOAEL for hypoxia (ICF 1997). 

 This represents the store size of an actual Hill PHOENIX customer who
is interested in the CO2 technology.

 This accumulation and pooling of CO2 may be a concern in mechanical
areas, storage rooms and other small spaces in the back of the store;
however it is not expected to be a concern in the main area of the store
as such low-lying and/or small areas would not be found there.

 This has been a concern for carbon dioxide fire extinguishing systems
and is addressed in several fire protection regulations (EPA 2000).

 ICF International maintains the Vintaging Model for EPA in order to
simulate the aggregate impacts of the ODS phaseout on the use and
emissions of various fluorocarbons and their substitutes over a period
of several years across more than 40 different applications.  The model
tracks the use and emissions of various compounds for the annual
vintages of new equipment that enter service in each end-use.  The
vintage of each type of equipment determines such factors as leak rate,
charge size, number of units in operation, and the initial ODS substance
that the equipment contained.    

 During disposal it is assumed that only 90 percent of the refrigerant
charge remains in the unit. 

  In order to simulate the horizontal concentration gradient that will
occur because of the weight differential between the refrigerant and
air, it is assumed that 95 percent of the leaked refrigerant mixes
evenly into the bottom 0.4 meter of the room, and the rest of the
refrigerant mixes evenly in the remaining volume (Kataoka 1999).

  HCFC-22. one of the refrigerants which CO2 is proposed to replace in
this end-use, has a short-term exposure limit (STEL) of 1,250 ppm (NIOSH
REL, available at   HYPERLINK
"http://www.cdc.gov/NIOSH/NPG/npgd0124.html" 
http://www.cdc.gov/NIOSH/NPG/npgd0124.html ) while CO2 has a STEL of
30,000 ppm (NIOSH REL, available at
http://www.cdc.gov/Niosh/npg/npgd0103.html).

  	            August 13, 2009

