Journal of Loss Prevention in the Process Industries 43 (2016) 29e34

Contents lists available at ScienceDirect

Journal of Loss Prevention in the Process Industries

journal homepage: www.elsevier.com/locate/jlp

Experimental evaluation of the possibility of ignition and ame

propagation in accumulated diuoromethane (R32) from a kerosene
cigarette lighter
Tomohiko Imamura a, *, Takuro Sano b, Kazutaka Yuzawa b, Osami Sugawa a
a
Department of Mechanical Engineering, Tokyo University of Science, Suwa 5000-1 Toyohira, Chino, Nagano 391-0292, Japan
b
Department of Mechanical Systems Engineering, Tokyo University of Science, Suwa 5000-1 Toyohira, Chino, Nagano 391-0292, Japan

a r t i c l e i n f o a b s t r a c t

Article history: The possibility of ignition and ame propagation in accumulated diuoromethane (CH2F2, R32) was
Received 21 February 2015 examined experimentally, simulating a situation in which a service operative uses a kerosene lighter for
Received in revised form smoking. To simulate the situation where a kerosene cigarette lighter is used in accumulated R32,
8 March 2016
electrodes xed in the windbreak of the lighter were remotely supplied with electricity to generate
Accepted 9 April 2016
Available online 13 April 2016
sparks of various durations but of similar energies to those of actual sparks generated by rubbing a int
to ignite the fuel in the lighter. We identied several cases of ignition and formation of an open ame in
the windbreak of the lighter, and the ame propagated to the accumulated R32 when it was supplied
Keywords:
Refrigerant
with sufcient energy from the spark. Gas chromatographic analyses conrmed that the mixture in the
Diuoromethane windbreak of the kerosene lighter consisted mainly of vaporized fuel and air, with no R32. Therefore,
Ignition even if the lighter is located in accumulated R32, an open ame can be generated in the windbreak of the
Flame propagation kerosene cigarette lighter through ignition by the spark energy generated by friction between the int
Kerosene cigarette lighter and the int wheel. Our results conrmed that there is a real possibility of ignition and ame propa-
gation when a kerosene cigarette lighter is used in accumulated R32 under the leak rate conditions of the
present experiment.
2016 Elsevier Ltd. All rights reserved.

1. Introduction the alternative refrigerants, such as their ammability limits, their

In the eld of refrigerating and air-conditioning engineering,
there is a worldwide trend toward the development and use of
alternative refrigerants that have no ozone-depleting potential and
low global-warming potential. For example, there are strong ex-
pectations that diuoromethane (R32), 2,3,3,3-tetrauoroprop-1-
ene (R1234yf), and (E)-1,3,3,3-tetrauoroprop-1-ene [R1234ze(E)]
will be used as alternative refrigerants, and several items of
equipment containing these refrigerants have already been
commercialized in Japan.
However, unlike the refrigerants in current use, such as R410A
[diuoromethane (R32)/pentauoroethane (R125) azeotrope], or
R134a (1,1,1,2-tetrauoroethane), these alternative refrigerants
have a degree of ammability, although this is lower than that of
most ammable gases. The fundamental combustion behaviors of
* Corresponding author.
E-mail address: imamura@rs.suwa.tus.ac.jp (T. Imamura).
minimum ignition energies, their burning velocities, and their
quenching distances, have been examined and reported by several
researchers (for example, see: Takizawa et al., 2015, 2009; Saburi
et al., 2014; Spatz and Minor, 2008). The American Society of
Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE)
has dened a refrigerant of class A2L as one that meets all the
following conditions: (1) it exhibits ame propagation at 60
C and
101.3 kPa, (2) it has an LFL of more than 0.10 kg/m3, (3) it has a heat
of combustion of less than 19 MJ/kg, and (4) it has a burning ve-
locity of less than 10 cm/s at 23.0
C and 101.3 kPa (American
Society of Heating, Refrigerating and Air-Conditioning Engineers,
2013). This classication has subsequently been adopted by the
International Organization for Standards (ISO) in its ISO817 stan-
dard (International Organization for Standards, 2014). On the basis
these researches and standards, there has been worldwide progress
in risk assessment and the development of guidelines for A2L re-
frigerants (for examples, see: DOE, 2014; Australian Institute of
Refrigeration, Airconditioning, and Heating, 2013, British
Refrigeration Association, 2012). In Japan (Japan Society of

http://dx.doi.org/10.1016/j.jlp.2016.04.003
0950-4230/ 2016 Elsevier Ltd. All rights reserved.
30 T. Imamura et al. / Journal of Loss Prevention in the Process Industries 43 (2016) 29e34
Refrigerating and Air Conditioning Engineers, 2015), these risk
evaluations of A2L refrigerant based on the physical hazard have
been conducted by the New Energy and Industrial Technology
Development Organization (NEDO), and they have led these re-
searches conducted in the world.
In some handling situation, several A2L refrigerants, such as
R1234yf, are classied in Japan as ammable gases, like hydrogen,
propane, etc., although the actual combustibility of the refrigerants
is markedly lower (for examples, see: Takizawa et al., 2015, 2009).
Therefore, if A2L refrigerants are to be used in air-conditioning
systems, it will be necessary to reconsider their classication and
to relax standards for their handling on the basis of risk manage-
ment for foreseeable actual handling situations and occasional ac-
cident scenarios, and it has been widely required not only in Japan
but also all over the world.
We have therefore conducted a series of experimental evalua-
tions of the physical hazards associated with A2L refrigerant,
assuming occasional accident scenarios in situations in which A2L
refrigerants are likely to be handled, which were assumed by the
risk assessment conducted by The Japan Refrigeration and Air
Conditioning Industry Association (for examples, see: Takaichi and
Taira, 2015; Watanabe, 2015; Yajima, 2015). We have reported the
possibilities of ignition of A2L refrigerants in several service and
maintenance situations, such as the case of leakage of refrigerant
from a pinhole in a pipe or hose, or leakage of a refrigerant into
equipment such as a refrigerant-collection device (Imamura et al.,
2015), and we have also reported experimental evaluations for
the case in which A2L refrigerant in an air conditioner is used at the
same time as a fossil-fuel heating system in a general living space
(Imamura et al., 2012).
In the above risk assessment, the inuences of varieties of igni-
tion source, such as lighter, appliances, electrical circuit, on the
ignition and ame propagation of A2L refrigerant are focused, and to
clarify them is widely expected (for examples, see: Takaichi and
Taira, 2015; Watanabe, 2015; Yajima, 2015). Therefore we have
focused on the situation in which a service operative uses a portable
lighter to smoke in an atmosphere where A2L refrigerant has accu-
mulated, and then the evaluation result using a portable piezoelec-
tric cigarette lighter was already reported (Imamura et al., 2013).
Here, we describe our experimental evaluation of the possibilities of
ignition and ame propagation in the case where a service operative
uses a kerosene cigarette lighter with a int-and-wheel ignition
system to smoke which is listed as the conceivable ignition source by
above risk assessment in atmosphere where R32 has accumulated as
a result of leakage from an air-conditioning systems.
2. Experiments
2.1. Experimental studies on ignition of A2L refrigerant
Fig. 1 is a schematic representation of our experimental setup.
Fig. 1. Schematic representation of the experimental apparatus. Units: mm.
R32 was used as the test A2L refrigerant. The acrylic pool measuring
1000 mm  1000 mm  1000 mm was xed to conduct the ignition
test.
A commercial kerosene cigarette lighter was located at the
300 mm above the center of the base of the pool as an ignition
source. R32 was leaked downwards into an acrylic pool measuring
1000  1000  1000 mm from a position 750 mm above the base of
the pool at a leakage rate of 10 g/min. The total leak amount of R32
into the pool was 220 g. The concentration at the position where
the lighter was located was approximately 16 vol%, which is within
the ammable range and slightly lower than the stoichiometric
concentration of 17.4%. To prevent mixing and turbulence of the
accumulated R32, the gas in the windbreak of the kerosene lighter
was ignited by means of a spark from a pair of electrodes (2-mm-
diameter, stainless steel, gap length: 2 mm) placed near the wick.
The ac electric sparks were generated in the gap between the
electrodes by using an inverter-type neon transformer (CR-N16;
Kodera Electronics, Co., Ltd., Gifu). A solid-state electric relay (GSR-
20L-D32Z; Misumi Group Inc., Tokyo) was inserted between the
circuit of the transformer and the electrodes to control the energy
for igniting the fuel of the lighter. Switching of the electrical supply
to the solid-state relay was controlled by means of a rectangular
wave signal of 5 V dc, generated by a function generator (33120A;
Agilent Technologies, Santa Clara, CA). The time for which elec-
tricity was supplied to the solid-state electric relay (referred to as
the energization time) was 50, 100, or 500 m. These energization
times were determined based on the visually comparison of spark
emission between actual spark generated by rubbing a int against
a int wheel directly and ac spark. In each ignition test, the
switching action to energize the relay was repeated ten times at
intervals of 5 s. The generated voltage between the gap of the
electrodes was measured by means of a high-voltage probe (P6015,
Tektronix Inc., Beaverton, OR), and the generated current was
measured by means of a current transformer (Model 2100, Pearson
Electronics, Inc., Palo Alto, CA). The energy of electric spark was
estimated by integration of the generated voltage and current over
the energization time.
Just before the ignition test, the concentrations of the accumu-
lated A2L refrigerant were measured by means of Fourier-
transform infrared spectroscopy with an FT-IR4200 spectrometer
(JASCO Corp., Tokyo). The sample gas was extracted to the FT-IR
through a 6.35-mm-diameter copper tube attached to a valve
switching box. Concentrations were measured at heights of 0, 100,
300, 500, 750, and 1000 mm above the center of the base of the
pool.
Video recordings of the kerosene lighter and its surroundings
containing R32 were made by using a high-speed camera (FAST-
CAM SA-X; Photron Ltd., Tokyo), a digital video camera (HC-V520M,
Panasonic Corp., Osaka), and a CCD camera (MTV-53KM21H, Min-
tron Enterprise, Co., Ltd., New Taipei City). The timing of the
beginning of energization of the solid-state relay and that of
commencement of the video recording by the high-speed camera
were synchronized by means of a trigger signal generated by a
function generator.
2.2. Analysis of the composition of the gaseous mixture in the
windbreak of the kerosene lighter
The composition of the mixture in the windbreak (volume:
2.72 mL) of the lighter in an atmosphere of accumulated R32 was
analyzed by a gas chromatography/mass spectrometry (GC/MS; GC-
17A, Shimadzu Corp., Kyoto). A 2.0-m-long vinyl tube with an in-
ternal diameter of 4 mm (total volume 25.12 mL) was inserted into
the windbreak. A 26 mL gas sample, which included air from the
extraction tube and the gaseous mixture from the windbreak, was
 T. Imamura et al. / Journal of Loss Prevention in the Process Industries 43 (2016) 29e34 31

Zt2
E IVdt (1)
t1

Fig. 2. An example of proles of voltage and current generated at the gap of the
electrodes.
Energization time: 100 m; approximate energy: 0.38 J.
extracted and being agitated, and a 200 mL portion was analyzed by
GC/MS in a single run. The GC/MS analysis was repeated four times.
3. Results and discussions
3.1. Evaluation of the validity of the supplied spark energy to
simulate the actual spark energy
We used an ac spark to ignite a kerosene cigarette lighter
instead of rubbing a int wheel directly to avoid mixing and tur-
bulence occurrence of accumulated R32 in the acrylic pool. To
evaluate the validity of the energy of an ac spark to simulate the
actual spark energy generated by rubbing a int against a int
wheel directly, we compared the amount of energy supplied by the
ac spark with the actual energy generated by rubbing the int
directly against the int wheel. Fig. 2 shows an example of proles
of generated voltage and current. In addition, because the high
voltage was generated by an inverter-type neon transformer, the
voltage and current proles of the ac electric spark were inter-
mittent shown in Fig. 2, it is possible that the total energy estimated
by integration of the voltage and current prole was not always
available for ignition. Although the estimated value of the energy of
the ac electric spark was only approximate, it was within the range
0.2e2.3 J over the energization time calculated by the following
formula;
where E: estimated spark energy (J), I: current (A), V: voltage (V), t1:
initiation time of spark (s), t2: nish time of spark (s).
The actual spark from the int is believed to be generated by the
following mechanism. First, the int is worn down and the
resulting particles are vaporized by the mechanical energy gener-
ated by rubbing the int on the int wheel. Strictly speaking, only a
part of worn-down particles are vaporized and being ignited, but to
estimate the ratio of worn-down particles changing to the gas is too
difcult. So we assumed that all of the worn-down particles are
changed to the gas and being ignited.
The vapor from the worn-down int is ignited by the mechan-
ical energy and is oxidized by combustion to form oxides as com-
bustion products, with concomitant radiation of the heat of
combustion. This heat of combustion (heat of formation of the
oxide combustion products) is the main component of the spark
energy generated by rubbing the int. Generally, the type of int
used in a kerosene lighter consists of an alloy of 30 mass% of iron
and 70 mass% of rare-earth elements such as cerium (Japan
Smoking Articles Corporate Association, 2013). We assumed that
the composition of the int is of this type. The heat of formation of
cerium (IV) oxide at 25
C and 101.325 kPa is 1088.7 kJ/mol, and
that of ferric oxide is 824.2 kJ/mol (Yokokawa, 2004).
To measure the mass of int particles generated by the one turn
of the int wheel, we rubbed a int 500 times and found that its
mass decreased by 0.06 g. Therefore the heat of combustion of the
worn-down int particles was therefore about 1.2 J per turn of the
int wheel which was calculated by the above heat of formation of
cerium (IV) oxide and ferric oxide. This value was of a similar
magnitude to the energy of the ac electric spark used to ignite the
kerosene lighter in our experiments (0.2e2.3 J). These experi-
mental results and discussions therefore conrm that the ac elec-
tric spark that we used in our experiments as a substitute for
rubbing the int against the int wheel did, indeed, approximately
reproduce the situation in which a kerosene lighter is used in the
normal manner, for example, to light a cigarette.
3.2. Results of experiments on the ignition of R32 by a kerosene
lighter
Table 1 lists our experimental results, and Fig. 3 shows the
vertical distribution of the R32 concentration in just before ignition
in each test case for various energization times. The repeatability of
the concentration prole was generally good which was conrmed
by six times of experiments.
Fig. 4(a) shows photographs of the lighter and its surroundings
containing R32 for sparks with an energization time of 50 m (en-
ergy ~0.2 J). Although the wick of the lighter in the windbreak

Table 1
Results of ignition experiments.

Exp. Refrige Room Relative Ignition source Energization Ignition Notes

No rant temp. humidity time [ms]
[
C] [%]

1 24.0 79 500 B An open ame formed and ame propagated at rst time of switching.
2 26.5 68 Kerosene cigarette lighter 50 
3 24.3 62 100 B An open ame formed and ame propagated at third time of switching.
4 R32 16.9 67  The ac spark was not observed.
5 17.4 62  The ac spark was not observed.
Only ac sparks
6 17.1 80 500  No open ame occurred and no ame propagation to R32 was observed
although an ac spark was generated.
32 T. Imamura et al. / Journal of Loss Prevention in the Process Industries 43 (2016) 29e34

ignited for an instant, no steady open ame formed, and no ame

Fig. 3. Vertical distribution of the concentration of R32.
propagation to the rest of the accumulated R32 occurred in any
instance of the switching action.
Fig. 4(b) similarly shows photographs for sparks with a 100 m
energization time (energy ~0.4 J). A steady open ame formed near
the wick at the third cycle of the switching action, and the gener-
ated open ame propagated smoothly to the surrounding R32.
Fig. 4(c) similarly shows photographs for sparks with an ener-
gization time of 500 m (energy ~2.3 J). A steady open ame formed
at the rst switching action. Flame propagation occurred more
immediately than was the case for an energization time of 100 m.
When an ac electric spark was generated in isolation at a posi-
tion 300 mm above the bottom of the acrylic pool in the accumu-
lated R32, no ignition or ame propagation to the body of R32 was
observed, regardless of the energization time (Table 1). This
conrmed that ignition of the R32 and ame propagation in the gas
were not caused directly by the electric spark but were, instead,
caused by the open ame in the windbreak of the lighter, produced
in turn through ignition of the lighter fuel by the electric spark. One
possible reason why the accumulated R32 was not ignited by the ac
spark in isolation, although its energy was more than the minimum
energy required for the ignition of R32, was that the accumulated
R32 had a vertical distribution of concentration although almost of
data of the minimum ignition energy were obtained under the ideal
condition that the concentration in the vessel is uniform.

Fig. 4. Photographs of R32 around the kerosene lighter ignited by ac electric sparks with various energization times. t: time since spark generation. R32 concentration at the height
of the lighter: 15.5 vol%.
 T. Imamura et al. / Journal of Loss Prevention in the Process Industries 43 (2016) 29e34
Fig. 5. Charts of the GC/MS analysis of R32 and simple components of the lighter fuel
and the mixed gas in the windbreak of the lighter.
Furthermore, it was considered to be that the occurrence of ignition
and ame propagation depended on the energy density of the
ignition source. Takizawa (Takizawa et al., 2015) said that the dif-
culty in determining the reliable MIE is that it is very dependent
on the electrode size, the gap between the electrodes, and the
ignition spark density and duration.
In some cases, no ac electric spark was observed in the accu-
mulated R32. This suggests that the voltage required to produce
electrical breakdown might differ between that in the mixture
present in the windbreak of the lighter and that in the accumulated
R32 in the absence of the lighter. In other words, it is possible that
the mixture in the windbreak of the lighter did not contain any R32.
In the present experiments, the rate of leakage of R32 was very slow
(10 g/min), so it was likely that there was little mixing between the
R32 and the gas in the windbreak.
3.3. GC/MS analysis of the gas in the windbreak
We analyzed the composition of the gas in the windbreak by
means of GC/MS. Fig. 5 shows the GC/MS charts for the sample gas
(four tests) as well as those for R32 alone and for the lighter fuel
alone. With R32 alone, a signicant peak appeared approximately
15.5e17 min after commencement of the GC/MS analysis, whereas
in the case of the lighter fuel alone, signicant peaks appeared
during the period 5e11 min after commencement of the analysis.
On the other hand, with the sample gas extracted from the wind-
break, although some scatter occurred, peaks that were similar to
the signicant peaks for the fuel were generally observed, whereas
those for R32 were barely visible in any of the sample gases. The
identities of the components corresponding to the peaks that
appeared during the periods 12e13 and 15e15.5 min after the start
of the analysis were unclear, but they did not correspond to R32 or
the lighter fuel. These results, coupled with our observations that
an ac electric spark was not always observed in the accumulated
R32 due to differences in the electrical breakdown voltage,
conrmed that R32 was not present in the mixture inside the
windbreak under the experimental conditions that we used.
Therefore, the gas mixture in the windbreak consisted of vaporized
kerosene and air, even when the lighter was positioned in the
accumulated R32. It is therefore reasonable that an open ame was
produced at the wick of the lighter by an electric spark of a similar
energy to that produced by rubbing the int, and that this ame
propagated to the body of the R32. However, if the leak rate of R32
33
is more rapid than the present experimental condition, leaked R32
and fuel mixture existed in the windbreak of a kerosene cigarette
lighter is mixed more. So the oxygen to ignite the mixture existed in
the windbreak, which consists of R32, fuel of a lighter and air, is
insufcient, therefore the ignition and ame propagation to R32
may not be always observed by using a kerosene cigarette lighter in
the accumulated R32.
4. Conclusions
We conducted a series of experiments to evaluate the possibility
of ignition of and ame propagation to R32 as result of operation of
a kerosene cigarette lighter. Ignition and ame propagation
occurred in experiments in which we used ac electric sparks as a
surrogate source of ignition for the usual generation of sparks by
rubbing of the int against the int wheel. In addition, GC/MS
analysis showed that the mixture in the windbreak of the kerosene
cigarette lighter consisted mainly of vaporized lighter fuel and air,
even when the lighter was located in accumulated R32. Our results
conrm that use of a kerosene cigarette lighter in accumulated R32
might be capable of causing ignition of and ame propagation to
R32 under the leak rate of the present experimental condition. The
similar evaluations for the possibility of ignition and ame propa-
gation of other A2L refrigerants, like R1234yf, R1234ze(E), are ex-
pected, it is the future work.
Acknowledgements
This study was conducted as a part of a research project on the
Technology Development of High-Efciency Non-uorinated Air
Conditioning Systems conducted by New Energy and Industrial
Technology Development Organization (NEDO). We would like to
thank Dr. Kenji Takizawa and Dr. Tei Saburi of the National Institute
of Advanced Industrial Science and Technology (AIST) for their
valuable suggestions; we also thank Dr. Kyoko Kamiya of the Japan
Specialist in Law Engineering Corporation for her kind supports
with the GC/MS analysis.
References
American Society of Heating, Refrigerating and Air-Conditioning Engineers, 2013.
Designation and Safety Classication of Refrigerants. ANSI/ASHRAE STANDARD
34e2013. American Society of Heating, Refrigerating and Air-Conditioning
Engineers, Inc., Atlanta, GA.
Australian Institute of Refrigeration, Airconditioning, and Heating, 2013. Flammable
Refrigerants: Safety Guide. Available at: http://www.airah.org.au/imis15_prod/
Content_Files/TechnicalPublications/Flammable-Refrigerant-Safety-Guide-
2013.pdf (Last accessed 19.01.15.).
British Refrigeration Association, 2012. Guide to Flammable Refrigerants. Available
at: http://www.feta.co.uk/uploaded_images/les/BRA%20Guide%20to%
20Flammable%20Refrigerants%20-%20Issue%201%20-%20Oct%2012.pdf (Last
accessed 19.01.15.).
DOE, 2014. Energy Conservation Standards for Commercial Refrigeration Equip-
ment: 10 CFR Part 431. EERE-2010-BT-STD-0003. Available at: http://energy.
gov/sites/prod/les/2014/02/f8/cre_ecs_nal_rule.pdf (Last accessed 19.01.15.).
Imamura, T., Kamiya, K., Sugawa, O., 2015. Ignition hazard evaluation on A2L re-
frigerants in situations of service and maintenance. J. Loss. Prev. Process Ind.
http://dx.doi.org/10.1016/j.jlp.2014.12.018.
Imamura, T., Morimoto, T., Kamiya, K., Miyashita, T., Sugawa, O., 2012. Physical
hazard evaluation for air conditioning systems having low-ammable re-
frigerants used simultaneously with a fossil-fuel heating system. J. Jpn. Soc.
Refrig. Air Cond. Eng. 29, 401e411 (in Japanese).
Imamura, T., Miyashita, T., Kamiya, K., Sugawa, O., 2013. Evaluation on ignition
hazard of leaked A2L refrigerants by a commercial piezoelectric lighter. J. Jpn.
Soc. Saf. Eng. 52, 91e98 (in Japanese).
International Organization for Standards, 2014. ISO817: Refrigerants: Designation
and Safety Classication, third ed. Available at: http://www.iso.org/obp/ui/#iso:
std:iso:817:ed-3:v1:en (Last accessed 19.01.15.).
Japan Smoking Articles Corporate Association, 2013. Lighter and Smoking Goods
Manual. Available at: http://www.jsrae.or.jp/info/2012progress_report_e.pdf
(Last accessed 19.01.15.).
Japan Society of Refrigeration and Air Conditioning Engineers (2015). Risk
34 T. Imamura et al. / Journal of Loss Prevention in the Process Industries 43 (2016) 29e34
assessment of mildly ammable refrigerants: 2014 Progress report. [ONLINE]
Available at: Our results conrm that use of a kerosene cigarette lighter in
accumulated R32 might be capable of causing ignition of and ame propagation
to R32 under the leak rate of the present experimental condition. [Last Accessed
Oct. 23th, 2015].
Saburi, T., Matsugi, A., Shiina, H., Takahashi, A., Wada, Y., 2014. Flammable behavior
of A2L refrigerants in the presence of moisture. In: Proceedings of Tenth In-
ternational Symposium on Hazards, Prevention and Mitigation of Industrial
Explosions. GexCon AS., Bergen, pp. 327e334.
Spatz, M., Minor, B., 2008. HFO-1234yf: a Low GWP Refrigerant for MAC. Honey-
well/DuPont Joint Collaboration. SAE World Congress, Detroit, Michigan.
Available at: http://www2.dupont.com/Refrigerants/en_US/assets/downloads/
SmartAutoAC/MAC_SAE_HFO_1234yf.pdf (Last accessed 19.01.15.).
Takaichi, K., Taira, S., 2015. Procedure for risk assessment of mildly ammable
refrigerant, Risk assessment of mini-split air conditioners. In: Japan Society of
Refrigeration and Air Conditioning Engineers (2015). Risk Assessment of Mildly
Flammable Refrigerants: 2015 Progress Report. Chap. 5 & 6, pp. 77e103.
Available at: http://www.jsrae.or.jp/committee/binensei/2014PR_e.pdf (Last
accessed 26 10.15.).
Takizawa, K., Igarashi, N., Takagi, S., Tokuhashi, K., Kondo, S., 2015. Quenching dis-
tance measurement of highly to mildly ammable compounds. Fire Saf. J. 71,
58e64.
Takizawa, K., Tokuhashi, K., Kondo, S., 2009. Flammability assessment of
CH2CFCF3: comparison with uoroalkenes and uoroalkanes. J. Hazard. Mater
172, 1329e1338.
Watanabe, T., 2015. Risk assessment for split air conditioners (Commercial package
air conditioners. In: Japan Society of Refrigeration and Air Conditioning Engi-
neers (2015). Risk Assessment of Mildly Flammable Refrigerants: 2014 Progress
Report. Chap. 7, pp. 104e120. Available at: http://www.jsrae.or.jp/committee/
binensei/2014PR_e.pdf (Last accessed 26 10.15.).
Yajima, R., 2015. Progress of risk assessment for VRF system. In: Japan Society of
Refrigeration and Air Conditioning Engineers (2015). Risk Assessment of Mildly
Flammable Refrigerants: 2014 Progress Report. Chap. 8, pp. 121e143. Available
at: http://www.jsrae.or.jp/committee/binensei/2014PR_e.pdf (Last accessed 26
10.15.).
Yokokawa, H., 2004. Handbook of Chemistry, Pure Chemistry, fth ed. Maruzen,
Tokyo. Table 10.127 (in Japanese).