School of Mechanical and Manufacturing Engineering
Hydrocarbon refrigerants have economic, environmental and performance ad-
vantages over nonflammable refrigerants. The Mobile Air Conditioning Societyclaims flammability too dangerous for hydrocarbons to replace R12 in car air-conditioners.
This report describes four ignition tests. The results, hydrocarbon data and
crude assumptions about accident statistics allow estimation of the increase in in-surance risk due to replacing R12 with hydrocarbons. Replacing R12 with hydro-carbons decreases the insurance risk by about $2 per operating year.
2 Ignition Tests
2.1 On engine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2 In passenger compartment . . . . . . . . . . . . . . . . . . . . . . .
2.3 On refrigerant jet . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.4 On puncturing container . . . . . . . . . . . . . . . . . . . . . . . .
3 Operational Risks
3.1 Leaks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2 Fatigue fracture . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3 On Collision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4 Insurance Risk Increment
The advantages of saturated hydrocarbon (alkane) refrigerants such as 67% propane/33%
butane by mass as a replacement for R12 in car air-conditioning are (Maclaine-cross
1. Their ozone depletion potential is zero.
2. In quantity they cost only 2$/kg and their low density means that only 40% of
the mass is required compared to R12 or R134a. For a medium sized Australian
car the charge of hydrocarbon is about 300 g the same as a large aerosol can.
3. Propane and butane are non-polar and so completely compatible with R12, its
lubricants and desiccants. R12 lubricants and desiccants are cheaper than those
for R134a but in any case they do not need to be replaced. My estimate is $50 to
change to propane/butane but $200 to change to R134a.
4. Propane and butane occur naturally in petroleum deposits and so the energy con-
sumption and byproducts of their manufacture are minimal. They are the least
5. Hydrocarbon increases cooling capacity by typically 10% compared with R12
(Maclaine-cross 1993). Recent refrigerant data (Gallagher et al.
a reduction in energy consumption by the car air-conditioning of 5%. Measure-
ments on energy consumption would require considerable funding. This data
also suggests energy savings would be about 5% greater again with cyclopropane
or if the air-conditioner was designed to take advantage of hydrocarbon refriger-
6. The 500 g saving in refrigerant mass using hydrocarbons gives a vehicle fuel and
pollution saving additional to that from superiority in refrigerant properties.
7. It has fewer toxic combustion products than R12. Technicians may be injured
or killed by leaking R12 because it is unodorized, they are welding or brazing
and some of its combustion products are toxic. Hydrocarbons have fewer toxic
combustion products than hydrofluorocarbons like R134a also.
For the consumer the only disadvantage of hydrocarbon refrigerants is that their
vapour is flammable when mixed with air. The two aspects to this disadvantage are
risks in servicing and risks in operating the motor vehicle. The risks and precautions
in servicing are similar to those with LPG used as transport fuel which are well known
(Katz and Lee 1990) and require no research. The mass of refrigerant is only 1% of that
in a fuel tank so the risk is correspondingly reduced. A prudent insurer would inquire
about the total quantity of flammable material being held on site by any automotive
business and require compliance with AS 1596–1989.
People consider it foolish not to fly because airliners crash or to only ride bicycles
because car drivers die in collisions. The risks and benefits must be assessed for any
decision. This applies also to using flammable refrigerant in car air-conditioners.
The risk of creating and igniting a flammable mixture of hydrocarbon refrigerant in
operating the motor vehicle has been claimed very high (Keebler 1993, MACS 1993).
The claims imply such accidents will cause great personal injury and property damage.
My students, colleagues and I have been unable to locate any reports of test results or
accident statistics which support these claims despite an estimate that 50,000 vehicles
are using flammable refrigerants in the US (Keebler 1993) and US bans on flammable
refrigerants as early as the 1950’s (MACS 1993). We would appreciate any help in
locating such literature. The fact that a proposition is widely believed is not evidence
Table 1: Data on the fire and explosion hazard of hydrocarbon refrigerants.
Four simple preliminary tests have been made to allow calculation of the probability
4. on puncture of refrigerant container.
The operational risk claims can be divided into a number of scenarios:
2. Fatigue fracture of circuit in operation with loss of refrigerant;
3. Frontal collision puncturing circuit with loss of refrigerant.
I assume the owner holds both comprehensive and third party injury cover as usual in
Australia. The last scenarios are high probability accidents and the insurer bears the
cost of replacing the refrigerant. The increased cost of nonflammable R12 compares
with the increased cost of damage and injury by fire using flammable alkane in all
scenarios. If the increased cost is greater, flammable refrigerant reduces insurer risk.
2 Ignition Tests
The following tests could be readily repeated by any professional engineer familiar
with the relevant product safety data sheets.
2.1 On engine
On the 18th December 1993 at 5:39 p.m., I drove my 5-door KE Laser (PGV496)
around the block. The air temperature was 29.5 C. I then took a Taymar butane weld-
ing torch ignited it and blew it out. The torch then resembled a large flammable re-
frigerant leak. I played the torch all over the engine, the ignition, the carburettor, the
exhaust and catalytic converter using 150 g of butane. I was not able to reignite the
Table 1 shows that about 500 C is required to ignite a flammable mixture of butane.
An electric arc or fusion is the only possible ignition source. Modern spark plug leads
have plug terminal insulators which prevent a spark from a loose lead. All electrical
faults except a direct electrical short across the battery will cause the failure of fuses
inside the passenger compartment. A direct electrical short across the battery would
cause failure in a day and could only exist for a short period in a car’s life. A collision
is necessary to create ignition sources in a modern engine.
2.2 In passenger compartment
OZ-12 is a commercial propane/butane mix sold as a replacement for R12 in the US
and now banned in many states. Keebler (1993) described an experiment by the Inter-
national Association of Arson Investigators (IAAI). They introduced 150 g of OZ-12
into a vehicle interior and artificially ignited it. The resulting pressure rise blew the
windows out of the car. Cockroach bombs have a propane/butane hydrocarbon propel-
lant. There has been at least one ignition accident in Australia with property damage
from failure to follow the dosage instructions and avoid ignition sources. Neither of
these scenarios is credible for a car but are related to the following.
Car air-conditioning has an evaporator and sometimes a control valve in the pas-
senger compartment behind the dashboard. These are nearly always on the passenger
side to avoid the steering mechanism and controls. A rapid leak from a fatigue fracture
occurring just before garaging the car would result in a passenger compartment filled
with refrigerant. If after an hour, a driver entered the vehicle with a cigarette would
On the 25th December 1993, I wound down the windows of my MF Barina (SCK302)
and sealed them with plastic sheet and adhesive tape. The car was completely sealed
except for the air-conditioning vent with lights and ignition off. The temperature in the
garage was 26.0 C. At 2:41 pm, a 150 g Mortein Roach Bomb was triggered on the
front passenger’s seat and the door closed. At 2:44 pm the spray finished. I lighted a
cigar and entered the driver’s side door at 2:45 pm. No flame, ignition or explosion
occurred even when the cigar was waved near the floor and over the back seat.
The active constituents of the bomb, 10 g/kg Permethrin 25:75 and 6 g/kg Phenoxy-
carb, do not effect the flammability of the hydrocarbon propellant. A Barina is a small
car and 150 g will produce a flammable mixture. Propane and butane are however sig-
nificantly heavier than air so when the driver’s door is opened the flammable mixture
exits at the bottom and fresh air enters at the top. The propane and butane remaining in
the compartment are insufficient to create a flammable mixture.
2.3 On refrigerant jet
The orifice at the top of a cockroach bomb represents a small leak that might occur in
an accident situation. The total charge of the cockroach bomb discharges through such
an orifice in about 3 minutes so a 300 g refrigerant charge would take 6 minutes.
On the 15th January 1994 at about 5:30 pm, the temperature in a sheltered outdoor
position for the experiment was about 21.2 C. I held the bomb and discharged it hori-
zontally at chest height. A lighted match was raised underneath the jet about 100 mm
from the orifice. The jet ignited when the match flame touched its lower edge. The jet
flame blew out when the match flame was removed. If the match flame was raised into
the jet, the match was extinguished.
Chemical engineering (Bird, Stewart and Lightfoot 1960) shows that before igni-
tion, a flammable mixture exists in a layer around the outside of the jet. The high
velocity of the jet causes rapid mixing with air into a nonflammable mixture. The
flame velocity for propane and butane in air at atmospheric pressure is between about
0.1 and 0.5 m/s (Ward 1978). The experimental observations are consistent with this
data. From this experiment and data, I draw two conclusions. An ignition source which
is temporary, like a spark, would ignite a small fraction of the refrigerant amplifying
the effective size of the spark. Ignition will occur only if the ignition source is precisely
2.4 On puncturing container
Rapid plastic deformation and friction can create temperatures sufficient for local igni-
tion. Whether this will ignite the refrigerant can be determined experimentally.
On the 15th January 1994 at 4:51 pm, the temperature in a sheltered outdoor posi-
tion for the experiment was about 21.2 C. A 150 g Mortein Roach Bomb was placed
on its side between two bricks. A crowbar was rammed through it, severing it half
way around its circumference. The bomb emptied in about 0.5 s. No flame, ignition or
At 6:00 p.m. another 150 g Mortein Roach Bomb was placed on its side between
two bricks. A crowbar was rammed into it making a jagged 10 mm by 5 mm hole. The
bomb emptied in about 3 s. No flame, ignition or explosion occurred.
Considerable force was required to make both holes since the bombs were rein-
forced by internal pressure. The sudden release of internal pressure by the hole caused
the hydrocarbon to spontaneously boil, expanding as a bubbly mixture and then break-
ing up into a droplet spray. The expansion of the bubbly mixture was responsible for
rapidly ejecting the liquid from the bomb. The hydrocarbon temperature after ejection
from the bomb would be about -30 C. The velocity of ejection from Bernoulli’s prin-
ciple would be about 40 m/s. The jet creates a small nonflammable hydrocarbon cloud
surrounded by a flammable mixture. In the open air, all the ejected hydrocarbon will
pass though a flammable mixture stage but the mixture becomes nonflammable due to
dilution only seconds after being ejected.
3 Operational Risks
The main sites for leaks in car air-conditioning are the hoses, the joints, the compressor
shaft seal and the charging valves. A single leak of 1 mg/s would empty a 300 g
refrigerant charge in 3.5 days. The probability of any technician missing a leak larger
than this is negligible. It is however too small to support a flame whatever the source
of ignition. Such an ignition source would need to be within 10 mm of a leak. Ignition
3.2 Fatigue fracture
Fatigue is failure of materials under alternating stress. Alternating stress is present
while the engine is operating. The heat exchangers are aluminium and the hoses elas-
tomer and both suffer fatigue fracture down to low stress levels. Poor design or instal-
lation can result in such fractures. They usually result in leaks which cause total loss
A fatigue fractures occurring in the engine bay during operation may be as common
as one in ten thousand operating years. The experiment described in Section 2.1 shows
that ignition sources are not usually present and ignition sources are believed present
on less than 1% of the vehicle population. The experiments in Sections 2.3 and 2.4
show that the quantity of flammable mixture present from a leak is at any time about
10%. The probability of an ignition source contacting and igniting leaking flammable
mixture is estimated as less that one in ten. A flammable mixture might be created and
ignited in this manner once in ten million operating years. The damage due to ignition
would be less than $1000 on average and it will be assumed to be covered by the policy.
Fatigue failures which cause total discharge of refrigerant in minutes are very un-
common. Fatigue stress levels on passenger compartment equipment are low and only
a fraction of the air-conditioning components containing refrigerant are there. Sud-
den fractures are certainly less common than once in a million operating years into the
passenger compartment. Hydrocarbon refrigerants contain 25 mg/kg ethyl mercaptan.
This allows refrigerant to be smelt (van Gemert and Nettenbreiger 1977) at 0.002 of
the lower flammability limit. A fatigue fracture causing a sudden major discharge into
the passenger compartment would have its flow rate limited by the expansion orifice to
below 10 g/s. A non-smoker would have 15 s warning of a flammable mixture to wind
down the driver’s side window and create a nonflammable situation. Even if all vents
were closed, 25 L/s infiltration would create a nonflammable mixture in three minutes.
A smoker may ignore the odour warning and has a source of ignition in hand. Over
the whole driving population smoking is equivalent to about one cigarette taking three
minutes a day or 0.2% of the time. This implies a fire from this risk once in 500 million
operating years with perhaps $100,000 damage on average for each fire.
A fatigue fracture may leak for the first time on garaging the vehicle. 300 g of
propane/butane in a small 30 m3 garage creates a concentration of 0.5% by volume.
This is only 25% of the lower flammability limit so a gas explosion is impossible if this
leaks into the garage. If it remains in the passenger compartment, there is no ignition
source when the vehicle is idle. If the door is opened, Section 2.2 shows ignition is
improbable and the potential injury of first degree burns to exposed skin is too small to
3.3 On Collision
Low velocity front collisions create about one insurance claim every ten operating
years. About one fifth of these are likely to involve loss of refrigerant in the collision.
Another fifth will require the refrigerant to be removed before repairs can commence.
These less serious collisions are more expensive to repair if R12 is used because the
law requires that R12 be recovered by trained and licensed operators before repairs
commence. R12 replacement is estimated at $50 and recovery and later replacement at
$100 more than hydrocarbon refrigerant. Conversion to R134a after an accident creates
Section 2.4 shows that ignition is unlikely from fracture of the refrigerant circuit
and Section 2.1 that ignition is unlikely from an intact engine. I expect damage to
electrical wiring and components to create an ignition source for one in ten accidents.
A flammable fraction of the leaking refrigerant might contact such an ignition source
one in ten times. Ignition of hydrocarbon refrigerant is expected once in a hundred
refrigerant loss accidents. Such fires would frequently add nothing to damage and
injury but it will be assumed here to add $1000.
Front to rear collisions rarely occur at sufficient velocity to fracture the fuel tank
of the vehicle in front. I will assume that this is once in a thousand operating years.
Ignition of hydrocarbon refrigerant is expected once in a hundred such accidents and it
will be assumed this ignites the fuel 50% of the time with a major fire. I assume that
on average this increases the cost of the accident by $100,000.
4 Insurance Risk Increment
Table 2 summarizes the above assumptions and data. Conversion of R12 air-conditioners
to alkane hydrocarbons reduces the insurance risk. This was probably not true ten years
ago when the cost premium for nonflammable R12 was less than $3 and no recovery
was required. An individual insurance company will have a client base and data which
would allow refining of these estimates.
The public may be very interested in whether their air-conditioner will catch fire if
flammable refrigerants are used. If such events occur, they will receive great publicity
because they are rare. Even if the estimates are considerably in error they will still be a
Table 2: Annual insurance risk increment on conversion of R12 air conditioner to
negligible insurance risk. The experiments which measure this risk are however worth
Collisions are very important in determining insurance risk since they may create a
source of ignition and a flammable mixture at the same time. If these are in the same
place a fire results. The cause of the fire will not be obvious if the damage is serious.
Expensive crash testing is necessary to measure this risk with greater accuracy.
Insurance company records and statistics should contain information for refining
the risk estimates. A full-time researcher for one year may be necessary to achieve
Changing from R12 to saturated hydrocarbon refrigerant increases the fire insurance
risk but reduces the refrigerant loss and recovery risk. For the assumptions and data
used here, hydrocarbons reduce insurance risk by $2.30/year. Further research and tests
which are considered desirable may change this conclusion.
Two precautions assumed are that the hydrocarbon refrigerant has been odorized
with 25 mg/kg ethyl mercaptan and that drivers are instructed to wind down their win-
dow completely to remove the odour if it appears.
AS 1596–1989, LP Gas—Storage and handling, Standards Australia, Sydney.
Bird, R. B., Stewart, W. E. and Lightfoot, E. N., 1960, Transport Phenomena, Wiley,
Gallagher, J., McLinden, M., Morrison, G. and Huber, M., 1993, NIST Thermody-
namic Properties of Refrigerants and Refrigerant Mixtures Database (REFPROP),
November, U.S. Department of Commerce, National Institute of Standards and
Technology, Standard Reference Data Program, Gaithersberg MD.
Katz, D. L. and Lee, R. L., 1990, Natural Gas Engineering, McGraw-Hill, New York.
Keebler, J., 1993, Cold fact: A/C gas danger, Automotive News, November, p. 1, 45.
Maclaine-cross, I. L., 1993, Hydrocarbon Refrigerants and Motor Car Air-Conditioning,
22nd November, Paper presented at Green Fridge Quest, Master Class Workshop,
National Science & Technology Centre, Canberra.
MACS, 1993, MACS Refrigerant Update, Automotive Cooling Journal, October, p.30.
van Gemert, L. J. and Nettenbreiger, A. H., 1977, Compilation of Odour Threshold Val-
ues in Air and Water, June, National Institute for Water Supply, Voorburg, Nether-
Ward, C. C., 1976, Gaseous Fuels, Marks’ Standard Handbook for Mechanical Engi-
neers, eds Baumeister et al.
, McGraw-Hill, New York, p.7-23.
Schede tecniche PRESSIONI DI ESERCIZIO MAX AMMISSIBILI sezione: Schede Tecniche catalogo tecnico PRESSIONI DI ESERCIZIO MAX AMMISSIBILI Pressioni di esercizio max ammissibili in funzione della temperatura e del materiali (rating) secondo norme DIN 2401 Valvole flangiate max pressione d'esercizio ammissibile in Kg/Cm2 alla temperatura di °C Pressione acciaio fuso (1
MBR JANUARY 2005 vol. 24 no. 1 EDITORIAL NEWS IN BRIEF HIV ‘explosion’ prompts call for action Over half of the world’s children are severely deprived Missing Prozac papers to be reviewed by FDA Israelis and Palestinians collaborate on health journal US medics in Iraq and Afghanistan may have participated in abusive Millions of people threatened by disease outbreaks foll