[Download PDF for Printing]
Vapor Cloud Explosions and Fires
All of us are aware of vapor cloud or gas explosions and
fire and the resulting damage that occurs. Many occur at industrial complexes
which handle fuels or manufacture chemicals. They can also occur at Liquefied
Natural Gas (LNG) Plants, for example, the LNG production plant at Skikda,
Algeria, on 19 January 2004. Residential areas are not immune to explosions
resulting from ignition of natural gas leaks. Transportation accidents may
result in the release of fuels.
Before the 11 September 2001 attacks on the World Trade
Center and the Pentagon, no one seriously considered that terrorists would use
commercial aircraft carrying a full load of fuel to destroy targets on the
The difference between a gas explosion and vapor cloud
explosion has to do whether the material is a gas or liquid at ordinary
temperatures and pressures. Methane and propane would be considered as a gas
but fuels such as gasoline, hexane, and cyclohexane are liquids which can
vaporize forming a vapor cloud. Finely divided combustible solids dispersed in
the air as in a grain or corn silo can also ignite resulting in a serious
The term, “Unconfined Vapor Cloud Explosion” apparently was
coined following the 1 June 1974 accident at a chemical plant near Flixborough,
UK. This accident involved the rupture in a 20-inch bypass pipe dumping almost
40 tons of cyclohexane which vaporized. The resulting vapor cloud found an
ignition source; the blast and fire destroyed the entire chemical complex
killing many people. The word “unconfined” can be misleading. While the
cyclohexane vapor was outdoors as opposed to being inside a tank or pipe, the
chemical equipment, ground surface, and buildings provided some confinement
which influenced the blast overpressure and damage.
Analysis of vapor cloud explosions at industrial facilities
can be complex. For example, the U.S. Chemical and Hazard Investigation Board
investigated the 23 March 2005 explosions and fire at BP’s Texas City refinery
which killed 15 people and injured 170 others. The CSB investigative team
interviewed over 350 witnesses, collected tens of thousands of documents, and
expended about $2 million dollars to conduct the investigation. An impressive
animated video of the accident and causes can be viewed at the CSB website, at
[click on: “ CSB Animation of BP Texas Refinery Accident”
video produced 27 Oct. 2005.]
Before we continue further with vapor cloud and gas
explosions, let’s look at some definitions. We might consider as an example
the chemical hexane. The display in the PEAC tool for hexane under physical
properties is illustrated below:
Figure 1 – PEAC-WMD Chemical Information for n-Hexane
The PEAC tool also displays information using metric units.
- Auto Ignition Temperature: This is the temperature
above which spontaneous combustion can occur without an ignition source
(spark or flame). While many data bases including the PEAC tool lists
auto ignition temperatures for particular chemicals, the actual auto
ignition temperature is also dependent upon the concentration of chemical
in the air and the material that it is in contact. For example, according
to the National Fire Protection Association (NFPA), the auto ignition
temperature of benzene measured in an iron flask is almost 200oF
higher than when measured in a quartz flask. The auto ignition temperature
of 1.5% pentane in air is 129oF higher than that for a 7.65%
concentration in air.
- Flash Point: This is the lowest temperature at
which a flammable liquid exposed to air will burn when exposed to sparks
or flame. The flash point is determined from laboratory tests. Several
laboratory tests are available: (1) closed cup tag, ASTM procedure D-65,
used for materials of 194oF (90oC) flash point or
less; (2) Pensky Martens, (ASTM procedure D-93), used for materials of 194oF
(90oC) or greater flash point; (3) Open cup tag (ASTM procedure
D-1310); and Cleveland test (ASTM procedure D-92). Many listings such as
the NIOSH pocket guide do not list flash points for flammable gases; if a
flashpoint is listed for a gas in a data source the chemical is assumed to
- Flammable Limits: The flammable limit is the range
of composition of material in air which will burn. The Upper Flammable
Limit (UFL) is the highest volume percent composition in air in which the
material will burn. The Lower Flammable Limit is the lowest volume
percent composition in air in which the material will burn.
- Explosive Limits: Same as flammable limits.
- Vapor Pressure: Pressure exerted by liquid (or
solid) on the vapor space. For example, the vapor pressure for hexane at
68oF (20o) is 0.16 atm. At sea level elevation, the
total atmospheric pressure is 1 atmosphere which means that if some hexane
is placed in a container and the container is allowed to vent to the air
the volume percent concentration of hexane in the air space above the
liquid will reach 16% at 68oF. The vapor pressure increases
with increasing temperature. At the boiling point of hexane (156oF),
the vapor pressure will be 1 atmosphere. At a high elevation location
where the total atmospheric pressure is say 0.75 atm the vapor pressure of
hexane is still 0.16 atm at 68oF (20oC), but now the
volume percent vapor in a vented container is now (0.16/0.75) x 100% =
- Reid Vapor Pressure: Vapor pressure measured at
100oF (37.8oC). The vapor pressure of hexane at 100oC
is 0.34 atm or 5 psi. The Reid vapor pressure is used in the petroleum
refining industry especially in the specification of fuel blends.
- Flammable Liquid: The NFPA defines a flammable
liquid as a liquid with a flash point less than 100oF (38oC).
- Combustible Liquid: The NFPA defines a combustible
liquid as a liquid with a flash point greater than or equal to 100oF
- Light end hydrocarbons: The petroleum industry
defines light end hydrocarbons as liquids which vaporize when exposed to
air, in particular, hydrocarbons with a Reid vapor pressure greater than
15 pounds per square inch absolute (psia). Examples are pentanes and
- High Flash Stocks: The petroleum industry defines
high flash stocks as liquids with a flash point greater than or equal to
130oF (55oC) or stored at temperatures of at least
15oF (8oC) below its flash point. An example is
heavy fuel oil.
- Low Flash Stocks: The petroleum industry defines
low flash stocks as liquids with a flash point less than 130oF
(55oC) or stored at temperatures within 15oF (8oC)
below its flash point. An example is Diesel fuel.
- Ignition Source: Fire or flames, electrical
sparks, or even very hot surfaces may serve as ignition sources. The fire
may come from a furnace, flares, welding, sparks from tools, matches and
lighters, or cigarettes. Electrical sparks may originate from switches
and motors, electrical equipment lightning, or buildup of static
electricity. A 1959 fire at the University of Minnesota Chemistry
building was caused by someone transferring a flammable liquid (benzene)
from an ungrounded 55 gallon drum to smaller containers for use in a
laboratory; the pouring of the liquid caused a buildup of static charge
which sparked igniting the liquid. Hot surfaces near or above the auto
ignition temperature (e.g. hot pipes, equipment, vehicle equipment) may
cause fuel to ignite even without spark or flame.
- Deflagration: Propagating reactions in which the
energy transfer from the reaction zone to the unreacted zone is
accomplished thru ordinary transport processes such as heat and mass
transfer. An example is fire, although the term deflagration is usually
used for a fire that burns fast (combustion processes with flame speeds on
the order of 0.5 to 1 meter/second).
- Detonation: Propagating reactions in which energy
is transferred from the reaction zone to the unreacted zone on a reactive
shock wave. The velocity of the shock wave always exceeds sonic velocity
in the reactant. This is the blast or explosion.
- Explosion: Explosions result when combustion
occurs so rapidly that there is a large buildup of pressure from the
combustion by-products resulting in a shockwave. The shockwave expanding
is the explosion (detonation). If combustion is not that rapid we get
deflagration (a fire).
Questions for Discussion
- A sniper using a high powered rifle fires at a tanker
truck carrying gasoline (petrol). Would there be a fire or explosion if
the bullet penetrates the tank?
Answer: Unlikely, unless the
gasoline leaks out from the tank and is ignited outside the tank. Liquid
gasoline does not burn; it is gasoline vapor in contact with air oxygen which
burns or could explode. The concentration of gasoline vapor in air must be
between the lower explosive limit and upper explosive limit for gasoline. The
PEAC tool lists the lower explosive limit as 1.4% and the upper explosive limit
as 7.6%. These numbers might vary somewhat depending how the gasoline was
formulated (another reference source lists 7% as the upper explosive limit).
At 68oF (20oC) the vapor pressure for gasoline is 0.39
atm, which translates to 39% vapor concentration by volume. The Reid Vapor
Pressure of gasoline [recall that this is the vapor pressure at 100oF]
is about 9.5 psi, or 0.647 atm, which converts to 64.7% vapor concentration by
volume. These are above the upper explosive limit, and therefore, the gasoline
does not ignite.
- Are there circumstances where the gasoline inside the
tanker truck can ignite and explode?
Answer: Yes. If the
tanker truck is essentially empty, the vapor concentration in the tank can drop
below 7.6% and ignite and explode if there is an ignition source. This is why
welding of gasoline tanks are especially dangerous; the tank must be purged
with an inert gas. Interestingly, the vapor space in a partially filled tanker
truck could become flammable under very cold temperatures (between 0oF
or 10oF and -45oF for gasoline). The flash point for
gasoline is -45oF (some references say -40oF). At 5oF,
the vapor pressure is about 0.076 atm or 7.6% (the numbers may vary depending
upon the gasoline formulation).
- If gasoline (petrol) spills on a hot surface such as an
automobile exhaust system, will it ignite?
Answer. Probably not, as
the auto ignition temperature of gasoline (from the PEAC tool) is 853oF
(456oC). Exhaust systems are not that hot. As pointed out in the
definitions, there are a number of variables that affect the auto ignition
temperature so any hot surfaces that even near the auto ignition temperature of
a fuel have the potential to ignite the fuel. Of course, a spark can ignite
- Does the same situation hold for diesel fuel?
Answer: Not quite; the
flash point, auto ignition temperature, vapor pressures, lower and upper explosive
limits are different. The flash point for diesel fuel is between 70oF
and 140oF depending upon the formulation [typically 120oF
or 49oC], which makes diesel fuel difficult to ignite if a bullet is
fired into a diesel fuel tanker. The auto ignition temperature of diesel fuel
is considerable less than gasoline (one reference lists 351oF or 177oC)
making it somewhat easier to ignite when contacted with a very hot surface.
The vapor pressure of diesel fuel is much less than gasoline; under normal
temperatures, the vapor pressure in a tanker truck should result in a vapor
concentration less than the lower explosive limit. Upper and lower explosive
limits are different for different formulations [e.g. LEL: 0.4 to 6%; UEL: 8%
- Why is the combination ammonium nitrate-fuel oil (ANFO) an
- Answer: The nitrate supplies the oxygen for the
reaction and the fuel oil supplies the fuel. Air oxygen is not required.
When ammonium nitrate and fuel oil is mixed in the right proportions the
reaction is so rapid that there is a large buildup of pressure from the
combustion by-products (ammonia, nitrogen, carbon dioxide, etc.) resulting
in a shockwave. A combination which results in maximum explosive effect
is blasting grade ammonium nitrate prills soaked in fuel oil (94% ammonium
nitrate, 6% fuel oil) and detonated by a high explosive booster or
blasting cap. The shockwave expanding what produces the explosion. The
- If there is a hypothetical cloud of propane up in the air
away from the ground or anything that might confine the propane, and the
cloud contacts an ignition source, will it detonate or deflagrate?
Answer: Deflagration is the better answer. [left
photo, propane deflagration, from Baker Engineering & Risk Engineering
Inc. website]. We will assume that the cloud concentration is between the
lower and upper explosive limits, although the concentrations at the edge of
the cloud may tail off below the lower explosive limit. The combustion front
will be very rapid but will probably not reach sonic velocities. With
nothing to restrain the front (e.g. ground, structures, a tank), the
large overpressures associated
with a blast may not occur. The result will be a huge fire ball. The same is
true for methane. There would be catastrophic fire damage but relatively low
explosion potential. There is some debate in the literature on how much blast
damage would occur.
- What are the conditions that result in vapor cloud
explosions as opposed to a deflagration?
Answer: This is a complex
issue which is not easily answered. The vapor cloud concentration must be
between the lower and upper explosive limit. The material itself is a factor.
Saturated hydrocarbons containing only hydrogen and carbon (e.g. methane,
ethane, propane, butane, pentane, hexane, etc. are more to result in
deflagration if completely unconfined. Olefins are more likely to detonate
(overpressure produced from combustion products). If the fuel is reactive
(e.g. acetylene) or contains oxygen (e.g. ethers, ethylene oxide) it is even
more likely to detonate. Any nearby structures including the ground which
provides a barrier to the combustion front can result in the buildup of
overpressures resulting in a blast front.
What is the Basis for the PEAC Tool Calculations?
We will look at an example. Let us select propylene as
the chemical. The PEAC tool will then display information about the chemical
including flash point, auto ignition temperature, vapor pressure, upper
explosive limit, and lower explosive limit. Calculations for a vapor cloud
explosion and fireball are initiated by clicking on the fireball icon, which
appears at the left.
The PEAC tool first displays a disclaimer statement and then
asks the user questions about the amount of propylene released as a vapor
cloud. The user can input the mass directly or calculate it based on the
original container size or calculate a mass in the container and assume that
only part of this forms a vapor cloud. The user can select a blast distance
based on either an overpressure or damage profile and a thermal heat distance
based on a threshold for second degree burns or 1% fatality or 50% fatality. A
partial final display may look like the following:
Some of the details on “Minimum Distances for Blast
Overpressure Effects” have been cropped in the display below. Also, two screen
images have been merged into one.
Figure 2 – Fireball Results from the
The PEAC tool follows the methodology used by the U.S.
Environmental Protection Agency (EPA) in their Risk Management Assessment in
calculation of blast and thermal heat distances as used in 40 CFR Part 68.
However, instead of assigning a yield factor of 0.1 for estimating the fraction
of the chemical which produces blast energy, the PEAC tool uses the yield
factors used by the ARCHIE Model. The EPA uses 0.1 as the yield factor. The
yield factors in ARCHIE vary from 0.02 for saturated hydrocarbons to 0.19 for
more reactive chemicals.
There are some differences in the literature of some of the
details of how to implement
these calculations, which means that the ARCHIE model does
not exactly give the same answers as say the EPA model even if the same
overpressures and yield factors are used. If the explosion occurs in the air
well above the ground, the distances are reduced by a factor of 1.26 compared
with an explosion at ground level. If the vapor is partly confined as in a
building or by walls the blast effects could even be greater. As mentioned
before, the PEAC modeling follows the EPA methodology using Yield Factors in
The citation for EPA is “Risk Management Program Guidance
for Offsite Consequence Analysis”, EPA 550-B-99-009, April 1999.
ARCHIE [an acronym for Automated Response for Chemical
Hazard Incident Evaluation] was developed under contract with Hazmat America,
Inc., in Arlington MA, and by scientists at Arthur D. Little, Inc., for the
Federal Emergency Management Agency (FEMA), the U.S. Department of
Transportation (DOT), and the U.S. EPA.
There are critics of this methodology who point out that
vapor cloud is not a point source but a defuse cloud whose concentrations tail
off to essentially zero at the edge of the cloud. The shape of the cloud
before ignition is not spherical but is influenced by wind or air movement and
by nearby structures. As a rough rule of thumb, the fire will burn at all
locations where the vapor cloud concentrations are about half the lower explosive
limit or higher; however, ignition takes place between the lower and upper
explosive limit values. Therefore, anything between about half the lower
explosive limit concentration or higher is expected to be consumed by a fire.
The blast waves release energy from the large volume of the ignited vapor cloud
and produce blast waves different than from detonation of say TNT or ANFO.
Alternative models are (1) TNO Mult-Energy Model, (2) Baker-Strehlow Model, and
(3) the Distributed-Source TNT Model. We will not discuss these models here
but refer to comparison in a company newsletter published by Quest Consultants,
Inc., Norman, Oklahoma, available over the internet at
http://www.questconsult.com/, “The Quest Quarterly, Vol 4 No. 1 and Vol 4 No.
2, 1999, specifically http://www.questconsult.com/99-spring.pdf. Another
paper, presented by Arthur D. Little staff, discusses the TNT Equivalency
method, the TNO Mult-Energy Model, and a computer software calculation package
(called AutoReaGas), was presented by H. Ozog and G.A. Melhem at the
International Conference and Workshop on Process Safety Management and
Inherently Safer Processes, AIChE/CCCP, October 1996, available at
The authors H. Ozog and G.A. Melhem, cited above, commented
that the peak overpressure within the vapor cloud when ignited probably does
not exceed 15 psi above atmospheric. At the edge of the cloud, the
overpressure is likely to be 10 psi or somewhat less. Outside the cloud, the
overpressure is likely to drop off as a function of distance as predicted by
the TNT point source equivalency equation which almost everyone uses including
the PEAC tool. At distances far from the cloud where the overpressure is less
than 0.5 psi, the overpressure blast wave behaves more like an acoustical wave,
and the basic TNT equivalency equation under predicts the distance for a given
The authors, J. Jiang, Z.G. Liu, and A.K. Kim presented a
paper titled “Comparison of Blast Prediction Models for Vapor Cloud Explosion”
sponsored by The Combustion Institute, Canadian Section, 2001 Spring Technical
Meeting, 13-16 May 2001 (see
http://irc.nrc-cnrc.gc.ca/pubs/fulltext/nrcc44715/). They concluded, “Over the
past decades a variety of prediction models have been developed to predict the
blast effect at any given distance from a possible explosion source. However,
these models have a limited range of applications, and they provide reasonable
results only if they are applied correctly within their intended purposes,
otherwise they may produce misleading results. Currently, there is no clear
recommendation on which model is preferred for a specific case under
Because of the many uncertainties in modeling vapor cloud
explosions and fires, the PEAC tool incorporates a safety factor of two in
estimating blast and thermal heat distances.
Aristatek searched the Internet for actual pictures of vapor
cloud explosions. There were many pictures available showing the damage
occurred or of fires following the explosion but the only pictures available of
actual vapor cloud explosions were of specially setup tests such as the photo
below rather than of actual industrial facilities. Notice that the explosion
image is not hemispherical.
Military test photo.