See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/312188545

Diesel combustion of oil and refrigerant mixture during pump-down of air

conditioners

Article in International Journal of Refrigeration · January 2017

DOI: 10.1016/j.ijrefrig.2017.01.003

CITATIONS READS
8 1,805

4 authors, including:

Tomohiro Higashi Chaobin Dang

The University of Tokyo University of Fukui
6 PUBLICATIONS 112 CITATIONS 180 PUBLICATIONS 4,276 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Chaobin Dang on 28 November 2017.

The user has requested enhancement of the downloaded file.

Accepted Manuscript

Title: Diesel combustion of oil and refrigerant mixture during pump-down of

air conditioners

Author: Tomohiro Higashi, Shizuo Saitoh, Chaobin Dang, Eiji Hihara

PII: S0140-7007(17)30005-1
DOI: http://dx.doi.org/doi: 10.1016/j.ijrefrig.2017.01.003
Reference: JIJR 3515

To appear in: International Journal of Refrigeration

Received date: 10-8-2016

Revised date: 16-12-2016
Accepted date: 4-1-2017

Please cite this article as: Tomohiro Higashi, Shizuo Saitoh, Chaobin Dang, Eiji Hihara, Diesel
combustion of oil and refrigerant mixture during pump-down of air conditioners, International
Journal of Refrigeration (2017), http://dx.doi.org/doi: 10.1016/j.ijrefrig.2017.01.003.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service
to our customers we are providing this early version of the manuscript. The manuscript will
undergo copyediting, typesetting, and review of the resulting proof before it is published in its
final form. Please note that during the production process errors may be discovered which could
affect the content, and all legal disclaimers that apply to the journal pertain.
 Diesel combustion of oil and refrigerant mixture during pump-down of air
conditioners

Tomohiro Higashia,*, Shizuo Saitohb, Chaobin Danga, and Eiji Hiharaa

a
Department of Human and Engineered Environmental Studies, Graduate School of Frontier Sciences,
The University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa-shi, Chiba 277-8563, Japan
b
Department of Mechanical Engineering, Graduate School of Engineering, The University of Tokyo,
7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan
Highlights

 Compressor destruction accidents during pump-down was experimentally investigated.

 Mixture of refrigerant, air and lubricating oil was adiabatically compressed.
 It was revealed that the accidents are caused by diesel combustion of mixture.
 Even non-flammable refrigerants could burn in the compressor.
 The oil was necessary for burning, however, the refrigerant led intense combustion.

ABSTRACT

Compressor-destruction accidents during the pump-down operation of air conditioners were experimentally
investigated. Assuming air penetration into refrigerant tubes, the gaseous mixture of the air, refrigerant, and
lubricating oil for a compressor was compressed by the compressor with different refrigerant concentrations,
and the diesel combustion of the mixture was examined. The compressor was simulated by a small-scale
engine. R1234yf, R32, R410A, R134a, R22, and R125 were tested as refrigerants. The mixture burned via
adiabatic compression when the refrigerant concentration was low, which means that accidents during the
pump-down were caused by the diesel combustion of the mixture. The refrigerant burned and caused intense
pressure increase. The mixture without the oil did not burn under any refrigerant concentration, which
suggests that oil is necessary for the combustion. These phenomena were observed in the results for R1234yf,
R32, R410A, R134a, and R22. Thus, combustion was observed under certain conditions even for refrigerants
categorized as non-flammable.

Keywords: Heat pump, Pump-down, Low-GWP refrigerant, Mildly flammable refrigerant, Diesel combustion,
Safety

1. Introduction

The Montreal Protocol abolished refrigerants with ozone-depleting potential (ODP) for ozone-layer protection,
and a transition to hydrofluorocarbons (HFC) refrigerants is underway. However, most HFC refrigerants have
a large global-warming potential (GWP), and the emission of the HFC refrigerant into the air causes global
warming. It is recognized that the transition to low-GWP refrigerants is extremely important across the globe.
New low-GWP refrigerants—R1234yf (CH2 = CFCF3) and R32 (CH2F2)—are attracting attention. In the
European Union, DIRECTIVE 2006/40/EC for mobile air conditioning and REGULATION (EC) No
842/2006 for residential air conditioning regulate the usage of refrigerants with a high GWP [1] [2]. North
American countries (the United States, Canada, and Mexico) proposed a step-down schedule for the
production of HFC refrigerants at the Conference on Montreal Protocol [3]. In Japan, the Act on Rational Use
and Proper Management of Fluorocarbons was enforced in April 2015, which demanded the production of
lower-GWP refrigerants, the development of refrigeration and air conditioning equipment that use lower-GWP

*
Corresponding author. Department of Human and Engineered Environmental Studies, Graduate School of Frontier
Sciences, The University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa-shi, Chiba 277-8563, Japan. Tel: +81 4 7136 4630; fax:
+81 4 7136 4631.
E-mail address: t.higashi@hee.k.u-tokyo.ac.jp (T.Higashi).

Page 1 of 18
refrigerants, leakage prevention, and the collection and destruction of refrigerants from end-of-life equipment
[4]
.
R1234yf and R32 are promising low-GWP refrigerants for replacing conventional refrigerants. Table 1
lists the environmental and flammability properties of typical refrigerants [5] [6] [8]. Comparatively, R1234yf and
R32 have a low GWP and good environmental performance along with low flammability; they are known as
mildly flammable refrigerants. In ASHRAE Standard 34, the rank 2L was newly founded for mildly
flammable refrigerants whose heat of combustion is lower than 19 MJkg-1, and their burning velocity is lower
than 10 cms-1. Both R1234yf and R32 are classified as 2L [7]. In Table 1, LFL represents the lower flammable
limit, UFL represents the upper flammable limit, BV represents the burning velocity, and MIE represents the
minimum ignition energy. Non-flammable refrigerants have a high GWP, and low-GWP refrigerants are
flammable. While the conventional refrigerants—R410A and R134a—are non-flammable and have a high
GWP, new refrigerants such as R32 and R1234yf are mildly flammable. More hydrogen atoms or less fluorine
atoms in a molecule reduces the molecular lifetime of a refrigerant in the atmosphere because of decreased
photostability. This tendency leads to a lower GWP and higher flammability. Thus, safety measures for low-
GWP refrigerants must be investigated extensively. Imamura et al. experimentally investigated the hazards of
burning caused by the leakage of R32, R1234yf, and R1234ze(E) from an air conditioner [9]. Takizawa et al.
measured the burning velocity, flammability limits, and heat of combustion for R1234yf, R1243zf, R32, and
R152a by schlieren photography and the spherical-vessel method [10]. Yang et al. theoretically and
experimentally investigated the flammability of R1234ze(E) and its blends by mixing it with R161 and R152a
[11]
. Jia et al. experimentally analyzed the leakage variation of R32 by running an indoor unit [12]. Regarding the
safety evaluation of hydrocarbons as low-GWP refrigerants, Zhang et al. investigated several hazards of an air
conditioner with R290, such as refrigerant leakage from the air conditioner and overpressure arising from
ignition [13].

This study focuses on compressor-destruction accidents during the pump-down of air conditioners as part
of the safety research for mildly flammable refrigerants. When a separate type air conditioner is relocated or
discarded, it is necessary to pump-down the unit for preventing the leakage of the refrigerant into the
atmosphere. Pump-down is an operation for allowing all of the refrigerant to accumulate in the outdoor unit by
closing the outlet valve of the outdoor unit and compulsory driving the air conditioner. In the pump-down
operation, the suction line of the compressor becomes vacuum. If penetration of air into the refrigerant tube
occurs because of an operational error of the three-way valve or damaged tubes, both the pressure and
temperature inside the compressor increase. Such accidents occurred five times from 2008 to 2011 in Tokyo,
Japan [14].
These accidents occur with conventional refrigerants, such as R22 and R410A, which are categorized as
non-flammable. Regarding the pump-down, even conventional non-flammable refrigerants can be dangerous.
Using mildly flammable refrigerants instead of non-flammable refrigerants changes the conditions or hazards
of explosion accidents during pump-down. The mechanism and conditions of accidents related to pump-down
have not been widely investigated. To introduce these mildly flammable refrigerants all over the world in
future, it is necessary to systematically examine the phenomena of explosions that occur during pump-down
and prepare safety guidelines for new refrigerants with a lower GWP than the conventional ones.
In this study, the conditions during the ignition of the mixture of the refrigerant, air, and lubricating oil of a
compressor were measured, and the effects of the flammability of the refrigerant were experimentally
investigated. The compressor will be broken if the mixture is self-ignited. To avoid breakage in the
experimental setup, a small engine was used as a simulated compressor.

Nomenclature

cv Specific heat at constant volume kJkg-1K-1

cp Specific heat at constant pressure kJkg-1K-1
p Absolute pressure MPa
T Temperature K
w Mass fraction -

Page 2 of 18
x Volume fraction of refrigerant -
S Spuriously polytropic index -
Greek letters
 Compression ratio -
 Specific-heat ratio -
 Density kgm-3
Index
air Air
atm Atmospheric pressure
ref Refrigerant
th Theoretical value
' Normalization
0 Standard pressure

2. Experiments

2.1 Experimental apparatus

Figure 1 shows a schematic of the experimental apparatus, which was designed to reproduce the combustion
inside a real compressor during pump-down accidents. It mainly consists of a compressor driven by a motor,
an air-supply system, a refrigerant-supply system, a temperature-control system, and a lubricating oil-supply
system. A model engine was used as the compressor instead of a real compressor for safety, as a real
compressor may explode because of self-ignited combustion.

(a) Compressor
A model engine (R155-4C, ENYA; 4 stroke, stroke volume: 25.42cc, compression ratio: 16.0) was used as a
simulated compressor and was driven by a motor (MELSERVO-J3, Mitsubishi Electric) connected to the shaft
of the engine. Figure 2 shows the cut model of the engine. The rotating speed of the engine was changed to
that of the motor. Gears were attached to the shaft. An encoder was also attached to the shaft, and the position
of the Z-phase pulse was set to the top dead center of the engine.

(b) Air-supply system

Air was compressed to 0.7 MPa using a compressor and then passed through a dehumidifier. Its pressure was
reduced to 0.3 MPa using a pressure regulator. The flow rate was controlled using a mass-flow controller
(MODEL8550MC-0-1-1, Kofloc; accuracy: ± 1.5% F.S., repeatability: ± 0.5% F.S., flow-rate range: 2 to 100
Lmin-1).

(c) Refrigerant-supply system

Vapor refrigerant supplied from a refrigerant cylinder was reduced to a pressure of 0.3 MPa using a pressure
regulator. The flow rate was controlled using a mass-flow controller (FCST1050LC-4F2-F50L-N2, Fujikin ;
accuracy: ± 2% F.S., repeatability: ± 0.2% F.S., flow-rate range: 2 to 100 Lmin-1). The conversion factor
(C.F.) of the mass-flow controller was used to calculate the flow rate. Table 2 shows the C.F. of each
refrigerant.

(d) Temperature-control system

The temperature of the gaseous mixture of the air and the refrigerant was measured by a sheathed
thermocouple and was controlled using heater 1 and a temperature controller. Heater 2 was used for equalizing
the temperature of the model engine with that of the intake gas of the engine. These heaters were used to
establish the experimental conditions required to constantly induce the explosion.

(e) Lubricating oil-supply system

PAG (VG46) oil was used as the test oil and was passed through a flow meter (micro-flow meter MODEL213-
311/295, Toyo Control). The pressure of the oil was increased to 150 MPa using an oil-injection system
(Common Rail Electric Control Fuel Injection System, FC Design), and the oil was injected in the form of a
spray into the intake port of the engine. The timing of the injection was determined using the encoder and a

Page 3 of 18
stroke sensor. The stroke sensor was located above the intake valve of the engine and detected the opening of
the valve. The clank angle when the oil was injected was controlled by assuming the Z-phase pulse of the
encoder, immediately after the signal from the stroke sensor was at 360°. The oil quantity per injection was
changed by according to the injection time. The injection started and ended automatically with the rotation of
the engine; the rotating speed was detected using the encoder.

(f) Measurement system

The pressure in the engine was measured using a pressure sensor (6045A, Kistler; linearity ≦ ± 0.4% FSO).
Considering the thermal behavior of the pressure sensor, the measured pressure was compensated by assuming
that the pressure was atmospheric when the intake valve was open. The pressure waveform was compensated
using signals from the encoder because the rotating speed of the engine became unsteady when combustion
occurred.
The exhaust gas was analyzed using a Fourier transform infrared (FT-IR) spectrometer (FT/IR-4700,
Jasco; resolution: 0.5 cm-1, S/N: 35000:1). The temperature of the FT-IR gas cell was maintained at 70 °C, and
the exhaust gas from the engine was kept in the gas cell during the measurement. The windows of the gas cell
were made of calcium fluoride (CaF2). The measurement range was 7,800 to 350 cm-1. Because of the single-
beam FT-IR method, the infrared absorption spectrum was obtained via comparison with the background
spectrum, which was measured before the experiments. Figure 3 shows an example of the background
spectrum. A wavenumber of ~2,350 cm-1 was not applied to the analyses because of absorption of carbon
dioxide (CO2) in the air. No wavenumber smaller than 950 cm-1 applied because of the absorption by the
windows of the gas cell. Data obtained from the Northwest-Infrared were applied to identify the combustion
products [15]. We used a spectrum of 4,039 cm-1 for the quantitative analysis of hydrogen fluoride (HF) to
avoid the noise caused by H2O and CO2. The HF concentration was calculated by assuming that the
concentration was proportional to the logarithm of the absorbance.
These measurements, along with the injecting signal of the oil, were recorded using a data-logger system
(NR-2000, Keyence; accuracy: ±0.1% F.S.). The sampling rate was set at 40 kHz.

Figure 4 shows a picture of the apparatus. The engine is located at the center and is connected to the motor
through the shaft. The gaseous mixture flows from the right side to the intake port of the engine. The injector
supplies the oil to the intake port. The exhaust gas flows to the left side and is analyzed using FT-IR.

The uncertainty of the inlet-temperature measurement is ± 1.5%, that of the pressure measurement is 4% (at
2.5 MPa), and that of the refrigerant-concentration measurement is 50% ± 1.3% (R1234yf at 50%).

2.2 Experimental conditions

Table 3 lists the experimental conditions. Table 4 lists the refrigerants and lubricating oils that were used. In
addition to R410A (conventional refrigerant) and R1234yf and R32 (new refrigerants), pure, non-flammable
refrigerants such as R134a, R22, R125, and nitrogen gas (inert gas) were used for comparison.
The inlet gas was heated before reaching the intake port of the engine. We could not use a real compressor
for this experiment because a real one can explode in the case of accidents. The model engine was used as a
compressor; however, its compression ratio was not high enough to increase the temperature of the gaseous
mixture sufficiently for burning. Therefore, we heated the inlet gaseous mixture.
The flow rate of the oil was determined according to the theoretical air/fuel ratio, which was calculated
using the air flow rate, as determined from the rotating speed and stroke volume of the engine. The calculated
theoretical air/fuel ratio was 9.5, according to the results of a component analysis of the oil conducted by SVC
Tokyo. The experimental parameters were determined according to the type and concentration of the
refrigerant used.

2.3 Experimental methods

The experimental procedure was as follows:
1) Set the air flow rate at a certain value, and heat the air at the intake side and the engine. Then, drive the
engine at ~50 rpm.
2) Set the air and refrigerant flow rates at the target values after the inlet temperature becomes steady and
reaches the predetermined value.

Page 4 of 18
3) Set the rotating speed of the engine at the target value after the flow rates of the air and refrigerant
become steady.
4) Confirm that the oil is injected, and then start collecting data using the data logger.
5) Close the outlet valve of the gas cell of the FT-IR, and start analyzing the exhaust gas.

2.4 Experiment contents

This study comprised two experiments. In Experiment 1, we investigated the self-ignition of a gaseous
mixture of air and lubricating oil. The flow rate of the oil was set to the theoretical air/fuel ratio. We measured
the pressure variation in the engine caused by self-ignition of the mixture. In Experiment 2, we measured the
pressure variation in the engine when the refrigerant was supplied with air. The concentration of the
refrigerant in the air was changed systematically. The flow rate of the oil was fixed for all the refrigerant
concentrations. In Experiment 2, we analyzed the exhaust gas resulting from the combustion of the gaseous
mixture via FT-IR.

3. RESULTS

In Experiment 1, we examined the pressure variation in the engine without lubricating oil. In Experiment 2, we
reproduced the mixing of the air during pump-down by varying the refrigerant concentration. We investigated
the effects of the refrigerant concentration by measuring the pressure in the engine and analyzing the exhaust
gas.

3.1 Experiment 1: without refrigerant

Figure 5 shows the typical pressure variations that occurred in the engine when the gaseous mixture of the air
and lubricating oil was compressed. The horizontal axis indicates the crank angle of the engine, the top dead
center at 360°. The pressure is plotted in 0.5° increments. The engine ran smoothly when no oil was injected,
indicating that no combustion occurred. The pressure increased drastically with oil injection. Additionally, the
exhaust gas became white, and a loud noise and strong vibrations were noticed. Figure 6 shows the typical
spectra of the exhaust gas with oil injection. In Fig. 6, CO was observed from 2,250 to 2,000 cm-1. In this case,
we assumed that the lubricating oil self-ignited because of the temperature increase induced by the adiabatic
compression.

3.2 Experiment 2: with refrigerant

Figure 5 (a) shows the typical pressure variations that occurred in the engine when the gaseous mixture of the
air and refrigerant was compressed. The refrigerant used was R1234yf. No oil was injected. The blue, red, and
green dots indicate the results for refrigerant concentrations of 0%, 10%, and 65%, respectively. The
maximum pressure increased as the refrigerant concentration decreased, presumably because of the higher
specific-heat ratio of the air compared to the refrigerant. Combustion did not occur, and the rotation of the
engine was smooth regardless of the refrigerant concentration. These tendencies were exhibited by all other
refrigerants. The results show that the refrigerants do not self-ignite in the absence of lubricating oil.
Figure 7 (b) shows the typical pressure variations that occurred in the engine when the gaseous mixture of
the air, refrigerant, and lubricating oil was compressed. The refrigerant used was R1234yf. The blue, red, and
green dots indicate the results for refrigerant concentrations of 0%, 10%, and 65%, respectively. When the
refrigerant concentration was high, e.g., 65%, no drastic increase in pressure was observed. At a concentration
of 10%, the pressure in the engine increased drastically, and an intense noise and vibrations were noticed. The
color of the exhaust gas was black at this point. We assumed that the refrigerant itself combusted. Similar
results were obtained with R32, R410A, R134a, and R22, although the refrigerant concentration range in
which the combustion occurred differed depending on the refrigerant. For R125, no combustion was observed
in the entire concentration range.

Figure 8 shows typical exhaust-gas spectra. The refrigerant was R1234yf, and its concentration was 60%
for (a) and 10% for (b). The chemical formula for the combustion of R1234yf is expressed by Eq. (1), where
the parameter x indicates the effect of the humidity in the air and has a value between 0 and 1. The main
reactive products of the combustion of R1234yf are HF and COF2. In Fig. 8 (a), typical R1234yf spectra
between 1,800 and 1,000 cm-1 are shown. No combustion occurred at a concentration of 60%. In Fig. 8 (b),

Page 5 of 18
where combustion occurred, spectra of HF, COF2 and CO were observed at 4,200 to 3,600; 1,980 to 1,880;
and 2,250 to 2,000 cm-1, respectively.

Page 6 of 18
 5
C H 2 C FC F3  O 2  xH 2 O
2 (1)
 (2  2 x )H F  (1  x )C O F2  (2  x )C O 2

3.3 Theoretical increase of pressure

When the air and refrigerant mixture is compressed adiabatically, the theoretical maximum pressure is
calculated as follows, and the mass fractions of the air and refrigerant are calculated according to the
volumetric fraction.
 ref x
w ref  and
 ref x   a ir (1  x ) (2)

 air (1  x )
w air 
 ref x   air (1  x ) (3)

The specific heats of the mixture at a constant volume and pressure are
c v  c v , ref w ref  c v , air w air and (4)
c p  c p , ref w ref  c p , air w air (5)

The specific-heat ratio of the mixture is calculated as follows:

cp c p , ref w ref  c p , a ir w a ir
  
cp c v , ref w ref  c v , a ir w a ir (6)

The pressure increase caused by the adiabatic compression was calculated by assuming a polytropic
change. The spuriously polytropic index ( S ) is introduced as follows:
S
pv  const. (7)

The spuriously polytropic index is determined by the measured data for the suction pressure ( p atm ),
maximum pressure ( p 0 ), and compression ratio (   at a concentration of 0%.
S  air
p 0  p atm  (8)
ln  p 0 p a tm 
S  (9)
 a ir ln 

Consequently, the theoretical maximum pressure of the air and refrigerant mixture is calculated using the
following equation:
S
p th  p atm  (10)

3.4 Relationship between maximum pressure and refrigerant concentration

Figure 9 summarizes the results of Experiment 2. The horizontal axis shows the volumetric concentration of
the refrigerant, and the left vertical axis shows the maximum pressure during the cycle, which was normalized
with respect to p 0 . The right vertical axis shows the HF concentration in the exhaust gas. “w/ oil” denotes the
results for the mixture of the air, refrigerant, and oil, and “w/o oil” denotes the results for the mixture of the air
and refrigerant, without the oil.

Page 7 of 18
 p m ax
p ' m ax 
p0 (11)

All experiments were conducted several times under the same conditions. Figure 9 shows the results.
For all refrigerants, no combustion occurred at high refrigerant concentrations. Under these conditions, the
maximum pressure with the lubricating oil was approximately the same as that without the oil. Thus, as the
refrigerant concentration was reduced, a large amount of air entered the compressor, the lubricating oil self-
ignited, and the combustion propagated to the refrigerant, regardless of the type of refrigerant. As the
refrigerant concentration decreased, the pressure increased and was maximized at a certain concentration. HF,
which resulted from the burning of the refrigerant, arose within the range of concentrations in which a large
pressure increase occurred. The HF concentration increased with the maximum pressure. These results imply
that not only the lubricating oil but also the refrigerant was burning. Although the range of the flammable
concentration differed slightly depending on the refrigerant, it was similar among all the refrigerants.
Furthermore, even refrigerants categorized as non-flammable would burn. For R125, which has the lowest
flammability among popular refrigerants, intense combustion did not occur at any refrigerant concentration,
and only very small amounts of HF were detected.
The criteria for flammability in ASHRAE 34 are based on the propagation of flames in the refrigerant and
air mixture. For the diesel-combustion experiments, the self-ignited lubricating oil was the ignition source.
The refrigerant continued to burn, and an explosively large pressure increase occurred. Kondo et al. reported
that high humidity can extend the flammable range of refrigerants and even cause non-flammable refrigerants
such as R410A to enter the flammable range [16]. The combustion of the lubricating oil can produce water in
the compressor during pump-down accidents. Kondo et al. reported that high pressure can extend the
flammable range of R1234yf and R32 [17]. Because of the high pressure and water due to the combustion of the
oil, the refrigerants may be more likely to burn in the compressor during pump-down accidents.

(a) R1234yf
Figure 9 (a) shows the dependence of the maximum pressure and HF concentration in the exhaust gas on the
R1234yf concentration. Combustion did not occur when the refrigerant concentration was higher than 20%. At
these conditions, the maximum pressure with the lubricating oil was approximately the same as that without
the oil. Combustion was observed when the refrigerant concentration was less than 20%. In this range, the
maximum pressure was 5 MPa at a concentration of 10%. HF was produced when the combustion occurred.
The concentration of HF increased with the maximum pressure, and its maximum value was ~3.5 vol%.
As indicated by Table 1, the UFL of R1234yf is 12.3 vol%; however, the highest flammable concentration
in this research was ~20 vol%, which is higher than the UFL. The LFL shown in Table 1 is 6.2 vol%; however,
a mixture with a lower R1234yf concentration combusted in this study. Even though the concentration was
toward the exterior of the flammable range shown in Table 1, HF was observed when the mixture burned,
which suggests that R1234yf itself burned.

(b) R32
Figure 7 (b) shows the results for R32. Combustion did not occur when the refrigerant concentration was
higher than 30%. When the concentration was lower than 30%, combustion occurred, and the maximum
pressure increased with the refrigerant concentration. The pressure reached a maximum of 5 MPa at a
concentration of 30%. The borderline between the ranges of burning and non-burning was very clear. The
concentration of HF increased with the maximum pressure.
The UFL shown in Table 1 is 29.3 vol%, which is almost the same as that observed in this study: 30 vol%.
The LFL shown in Table 1 is 13.3 vol%; however, a mixture with a lower R32 concentration combusted in
this study, similar to the case of R1234yf.

(c) R410A
Figure 9 (c) shows the results for R410A. R410A is categorized as a non-flammable refrigerant; however,
combustion occurred when the concentration was less than 20%. In this region, the maximum pressure
increased with the refrigerant concentration. The borderline between the ranges of burning and non-burning
was relatively clear. The concentration of HF increased with the maximum pressure. We assume that real-

Page 8 of 18
world accidents can occur because of air conditioners using R410A; hence, these results reflect actual
accidents.

(d) R134a
Figure 9 (d) shows the results for R134a. Although weak combustion occurred when the concentration was
less than 30%, the maximum pressure was lower than that obtained with the other refrigerants and was
approximately equal to the result obtained without a refrigerant. In this range, we believe that the refrigerant
did not burn and that only the lubricating oil burned. When the refrigerant concentration was reduced to less
than 7.5%, intense pressure was observed, which suggests that the refrigerant was also burning. The HF
concentration was high around this range. We attribute the lower reactivity of R134a compared to the other
refrigerants to the larger amount of fluorine in its molecular composition.

(e) R22
Figure 9 (e) shows the results for R22. Combustion occurred when the concentration was less than 50%. A
wider range of the combustible concentration was measured compared to the other refrigerants.
Park et al. investigated the thermal decomposition of R22 and reported that 20% of R22 was decomposed
by heating at 650 °C for 0.8 s [18]. The chemical reaction for the thermal decomposition of pure R22 is
2CHClF2  CF2  CF2  2HCl (12)
Edwards et al. investigated the thermal decomposition of R22 at 750–950 °C with a heating time of 0.1 to 0.25
s. They reported that the first breakdown step was described by the following equation and that the CF2
radicals played an important role in the decomposition of R22 [19].
C H C lF2  C F2 +H C l (13)
The second breakdown step is believed to be the dimerization of CF2 to C2F4.
2C F2  C F2 =C F2 (14)
[20]
Broyer et al. reported that 51% of R22 was decomposed by heating at 510 °C for 500 s .
Although R22 undergoes a decomposition reaction at a low temperature, as previously described, it is
incombustible in the American Society for Testing and Materials combustion test [21] because of the following
reason. Even if the reaction of Eq. (13) is induced by the energy of the ignition source in the combustion test,
the energy necessary for flame propagation is not continuously supplied because Eq. (14) is an endothermic
reaction of -212 kJ mol-1. On the other hand, many types of combustion products of oil exist in the current
experiments. Therefore, the CF2 radicals may react with substances other than those in the reaction of Eq. (14).
HCl in Eq. (13) may also contribute to exothermic reactions. It is considered that the gaseous mixture of R22
and the lubricating oil may explode because of the intermediate products with high reactivity and exothermic
reactions. For refrigerants that do not contain chloride in the molecule, such as R32, the combustion heat of oil
induces the production of CFx radicals as an elementary process in the combustion of the refrigerant, and the
mixture of the refrigerant and the oil is self-ignited under certain conditions.
The pressure reached a maximum at a concentration of 22.5%. The HF concentration exhibited a tendency
similar to pressure. When the refrigerant burned, HCl was generated. Figure 10 shows the spectra of the
exhaust gas obtained via FT-IR analysis. Figure 11 shows the relationship between the refrigerant and HCl
concentrations in the exhaust gas. The FT-IR spectra include typical HCl absorption spectra in the range of
3,100–2,600 cm-1. We used a wavenumber of 2,844 cm-1 for the spectral quantification of HCl. The
concentrations of HF and HCl exhibit similar tendencies, suggesting that HCl was also a combustion product
of R22.

(f) R125
Figure 9 (f) shows the results for R125. Combustion did not occur when the refrigerant concentration was
higher than 10%. Although combustion occurred when the concentration was lower than 5%, only very weak
combustion was observed, and the maximum pressure was very small. A small amount of HF was detected.
R125 has more fluorine than the other refrigerants studied, suggesting that it has the highest stability even in
the compressor.

(g) Nitrogen
Figure 9 (g) shows the results for N2. Because the properties of nitrogen are similar to those of air, the
theoretical value of the maximum pressure was approximately constant. Weak combustion, i.e., the

Page 9 of 18
combustion of the oil, occurred when the concentration of nitrogen was lower than 70%, and the maximum
pressure increased slightly when the concentration was lower than 50%.

10

Page 10 of 18
 4. CONCLUSIONS

The diesel-combustion phenomena that occurred during a pump-down operation were investigated using an
experimental apparatus that reproduced the adiabatic compression of the gaseous mixture of the air, refrigerant,
and lubricating oil during pump-down. R1234yf, R32, R410A, R134a, R22, and R125 were tested as
refrigerants. It was revealed that accidents that may occur during the pump-down of air conditioners are
caused by the diesel combustion of a mixture of the air, refrigerant, and lubricating oil. Although a mixture of
the air and oil can burn under certain conditions, the pressure increase resulting from this combustion was
moderate unless a refrigerant was added. This suggests that the refrigerant itself burns, leading to a sudden and
significant increase in the pressure.
According to ASHRAE 34, R22, R125, R134a, and R410A are non-flammable, and R32 and R1234fy are
mildly flammable. However, experiments simulating a diesel explosion during pump-down indicated that the
diesel combustion can occur with all the refrigerants except for R125. Compared to the conventional
refrigerants, including R410A and R22, the new lower-GWP, mildly flammable refrigerants R1234yf and R32
exhibited insignificant differences in their flammable range and maximum pressure.
The mechanism of ignition of refrigerants and the marginal concentration boundary for diesel combustion
are not well-understood. It is concluded that the risk of diesel explosion does not coincide with the
flammability category in ASHRAE 34 at this point.

ACKNOWLEDGMENT

The authors gratefully acknowledge the financial support of the New Energy and Industrial Technology
Development Organization (NEDO).

REFERENCES

1). Official Journal of the European Union, “DIRECTIVE 2006/40/EC OF THE EUROPEAN PARLIAMENT AND OF THE
COUNCIL”, 2006.
2). Official Journal of the European Union, “REGULATION (EC) No 842/2006 OF THE EUROPEAN PARLIAMENT AND OF
THE COUNCIL”, 2006.
3). United States Environmental Protection Agency, “Summary: North American 2015 HFC Submission to the Montreal Protocol”,
2015, http://www3.epa.gov/ozone/intpol/HFC_Amendment_2015_Summary.pdf.
4). Ministry of the Environment Japan, “Law on regulation of management and rational use of fluorocarbons”, 2015,
http://www.env.go.jp/earth/ozone/cfc/law/kaisei_h27/index.html.
5). Japan Fluorocarbon Manufactures Association, “Environmental and safety properties of CFC, HCFC and fluorocarbons”, 2016
http://www.jfma.org/database/table.pdf.
6). IPCC, “Climate Change 2013: The Physical Science Basis”, 2013, Chapter 8: p.731 – 732.
7). American Society of Heating, Refrigerating and Air-conditioning Engineers, Inc. “ANSI/ASHRAE Standard 34-2013”, 2013.
8). Japan Society of Refrigerating and Air Conditioning Engineers, “Risk Assessment of Mildly Flammable Refrigerants 2012
Progress Report”, 2013: p. 4.
9). Imamura, T., Kamiya, K. and Sugawa, O., Ignition hazard evaluation on A2L refrigerants in situations of service and
maintenance, Journal of Loss Prevention in the Process Indusries, 2015, 36: p. 553–561.
10). Takizawa, K., Kazuaki, T. and Kondo, S., Flammability assessment of CH2 CFCF3: Comparison with fluoroalkenes and
fluoroalkanes, Journal of Hazardous Materials, 2009, 172: p. 1329–1338.
11). Yang, Z., Wu, X. and Tian T., Flammability of Trans-1, 3, 3, 3-tetrafluoroprop-1-ene and its binary blends, Energy, 2015, 91: p.
386–392.
12). Jia, L., Jin, W. and Zhang, Y., Analysis of Indoor Environment Safety with R32 Leaking from a Running Air Conditioner, Procedia
Engineering, 2015, 121: p. 1605–1612.
13). Zhang, W., Yang, Z., Li, J., Ren, C., Lv, D., Wang, J., Zhang, X. and Wu, W., Research on the flammability hazards of an air
conditioner using refrigerant R-290, International Journal of Refrigeration, 2013, 36(5): p. 1483–1494.
14). Tokyo metropolitan conference on safety of products, “Report on safety ensuring at disassembly of air conditioner”, 2012: p. 2-3.
http://kanagawa-ds.org/wp-content/uploads/f86d1c8dbc80627d75dd17aae7d656a7.pdf.
15). Pacific Northwest National Laboratory, “Vapor phase infrared spectral library”, 2013

11

Page 11 of 18
 https://secure2.pnl.gov/nsd/nsd.nsf/Welcome.
16). Kondo, S., Takizawa, K. and Tokuhashi, K., Effect of high humidity on flammability property of a few non-flammable refrigerants,
Journal of Fluorine Chemistry, 2014, 161: p. 29–33.
17). Kondo, S., Takahashi, A., Takizawa, K. and Tokuhashi, K., On the pressure dependence of flammability limits of CH2=CFCF3,
CH2F2 and methane, Fire Safety Journal, 2011, 46: p. 289–293.
18). Park, J. D., Benning, A. F., Downing, F. B., Laucius, J. F. and McHarness, R. C., Synthesis of Tetrafluorethylene -Pyrolysis of
Monochlorodifluoromethane, Industrial & Engineering Chemistry, 1947, 39 (3), p. 345–358.
19). Broyer, E., Bekker, A. Y. and Ritter, A. R. Kinetics of the Pyrolysis of Chlorodifluoromethane, Industrial & Engineering
Chemistry Research, 1988, 27 (1): p. 208–211.
20). Edwards, J. W. and Small, P. A., Kinetics of the Pyrolysis of Chlorodifluoromethane, I & EC Fundamentals, 1965, 4 (4): p. 396–
400.
21). ASTM International, “Concentration Limits of Flammability of Chemicals (Vapors and Gases),” E681 − 09 (Reapproved 2015).

12

Page 12 of 18
 Scrubber

Mass-flow
Refrigerant
controller FT-IR
cylinder

Gas cell
Heater 1
Compressor
(model engine)

T Heater 2 Motor
Dehumidifier Mass-flow
Air compressor controller
Flow
Oil-injection P
meter
system
Encoder
Stroke sensor
Oil tank

Fig. 1 Schematic of experimental apparatus

Fig. 2 Cut model of the engine

10
Intensity [a.u.]

4500 3500 2500 1500 500

-1
Wavenumber[cm ]

Fig. 3 Infrared absorption spectrum of background

13

Page 13 of 18
Exhaust gas
Gaseous mixture

Motor
Injector

Encoder Engine

Fig. 4 Photograph of apparatus

2.5
without oil
2.0 with oil
Pressure [MPa]

1.5

1.0

0.5

0.0
180 240 300 360 420 480 540

Angle [degree]

Fig. 5 Pressure change in the engine

1.0
with oil
0.8 Spectra of CO
0.6
Abs [-]

0.4

0.2

0.0

-0.2
4500 3500 2500 1500 500
-1
Wavenumber [cm ]

Fig. 6 Infrared absorption spectrum of exhaust gas with oil injection

14

Page 14 of 18
 6 6
R1234yf R1234yf
Pressure [MPa]

Pressure [MPa]
4 4

2 2

0 0
180 240 300 360 420 480 540 180 240 300 360 420 480 540

Angle [degree] Angle [degree]

0% 10% 65% 0% 10% 65%

(a) Without oil (b) With oil

Fig. 7 Pressure in the engine (a) without oil and (b) with oil

5
Spectra of R1234yf
4

3
Abs[-]

-1
4500 3500 2500 1500 500
-1
Wavenumber[cm ]

(a) 65% R1234yf concentration

5
Spectra of CO
4 Decrease of R1234yf
Spectra of COF2 3
Abs[-]

1
Spectra of HF 0

-1
4500 3500 2500 1500 500
-1
Wavenumber[cm ]

(b) 10% R1234yf concentration

Fig. 8 Infrared absorption spectra of the exhaust gas at R1234yf concentrations of (a) 60% and (b) 10%

15

Page 15 of 18
 3 0.6 3 0.6
R1234yf R32
0.5 0.5

HF [vol%]

HF [vol%]
2 0.4 2 0.4
p'max [-]

p'max [-]
0.3 0.3
1 0.2 1 0.2
0.1 0.1
0 0.0 0 0.0
0 20 40 60 80 100 0 20 40 60 80 100

Refrigerant concentration [vol%] Refrigerant concentration [vol%]

w/ oil w/o oil pth HF w/ oil w/o oil pth HF

(a) R1234yf (b) R32

3 0.6 3 0.6
R410A R134a
0.5 0.5

HF [vol%]
HF [vol%]
2 0.4 2 0.4

p'max [-]
p'max [-]

0.3 0.3
1 0.2 1 0.2
0.1 0.1
0 0.0 0 0.0
0 20 40 60 80 100 0 20 40 60 80 100

Refrigerant concentration [vol%] Refrigerant concentration [vol%]

w/ oil w/o oil pth HF w/ oil w/o oil pth HF

(c) R410A (d) R134a

3 0.6 3 0.6
R22 R125
0.5 0.5
HF [vol%]

HF [vol%]
2 0.4 2 0.4
p'max [-]

p'max [-]

0.3 0.3
1 0.2 1 0.2
0.1 0.1
0 0.0 0 0.0
0 20 40 60 80 100 0 20 40 60 80 100

Refrigerant concentration [vol%] Refrigerant concentration [vol%]

w/ oil w/o oil pth HF w/ oil w/o oil pth HF

(e) R22 (f) R125

3
N2

2
p'max [-]

0
0 20 40 60 80 100

Refrigerant concentration [vol%]

w/ oil w/o oil pth

(g) N2
Fig. 9 Maximum pressure and HF concentration with respect to the refrigerant concentration

16

Page 16 of 18
 5

Spectra of HCl 4 Spectra of R22

Abs[-]
2

1
Spectra of HF 0

-1
4500 3500 2500 1500 500
-1
Wavenumber[cm ]

Fig. 10 Infrared absorption spectrum of the exhaust gas at an R22 concentration of 20%

3 3.0
R22

HCl [vol%]
2 2.0
p'max [-]

1 1.0

0 0.0
0 20 40 60 80 100

Refrigerant concentration [vol%]

w/ oil w/o oil pth HCl

Fig. 11 Maximum pressure and HCl concentration with respect to the R22 concentration

17

Page 17 of 18
Table 1 Environmental and flammability properties of refrigerants
Refrigerant ODP GWP LFL UFL BV MIE ASHRAE
vol% vol% cms-1 mJ flammability
category
R1234yf 0 <1 6.2 12.3 1.5 200 2L
R32 0 677 13.3 29.3 6.7 15 2L
R410A 0 2,090 non-flammable 1
R134a 0 1,300 non-flammable 1
R22 0.055 1,760 non-flammable 1
R125 0 3,170 non-flammable 1

Table 2 C.F. values of refrigerants

Refrigerant C.F. [-]
R1234yf 0.248
R32 0.608
R410A 0.438
R134a 0.304
R22 0.460
R125 0.267
N2 1 (definition)

Table 3 Experimental conditions

Rotating speed of engine, rpm 1500
Mixture flow rate , Lmin-1 18.75
Inlet gas temperature, °C 260
Oil flow rate, Lmin-1 2.295 × 10-4
Refrigerant concentration, vol% 0-100

Table 4 Refrigerants and lubricating oil

Item Type
Refrigerant R1234yf, R32, R410A,
R134a, R22, R125, and
N2
Lubricating oil PAG (VG46)

18

Page 18 of 18

View publication stats