Wednesday July 1, 2009 - Vol. VIII Issue 7
[Download PDF for Printing]
Real-World Explosions and Fire Examples
We will take a snapshot look at example explosions and fires
involving chemicals including fuels. Many of the examples chosen occurred at U.S.
industrial facilities. Major accidents at these locations are investigated by
governmental agencies such as the Chemical Safety Board, the Department of
Transportation, Occupational Safety and Health Administration, and the
Environmental Protection Agency, and their reports are published and available
in the public domain. Other investigations are done by insurance companies. Similar
investigations are done by appropriate governmental agencies in several other
countries, and the reports are sometimes available. Several of these events are
vapor cloud explosions. Some involve tank or vessel explosions and some examples
are cascading chain-event accidents. Both catastrophic and smaller scale examples
are illustrated.
Flixbourough England 1974 Vapor Cloud Explosion. 28 or
29 Fatalities depending upon account
On the afternoon of 1 June 1974 at an industrial facility manufacturing
a nylon precursor chemical, a 20-inch diameter pipe carrying cyclohexane at 150
oC
(302
oF) temperature and 1 MPa (145 psi) pressure ruptured (possibly
the result of a pressure surge) sending a large amount of vaporized cyclohexane
into the atmosphere. At this temperature and pressure, the cyclohexane would
have been liquefied, but would have flashed almost instantaneously with the
pressure released. The vapor cloud according to reports (see
http://chemicalprocesssafety.org/flixborough.aspx)
was estimated to be 100 to 200 meters (320 to 650 feet) in diameter. The
vapor cloud found a nearby ignition source, possibly a furnace at a nearby
hydrogen production plant, and exploded resulting in a fireball. The blast
completely destroyed or severely damaged 1800 buildings within a mile radius
from the site; secondary fires burned for over a week. All 18 employees in a
nearby control room were killed; 9 other site personnel were also killed, plus
2 other people counting a heart attack victim in a vehicle. There were 36
people seriously injured. Fortunately, the accident occurred on a weekend when
almost all employees were away. The TNT equivalent of the fuel-air mixture was
estimated to be 15 metric tons based on blast damage (which is a greater TNT
equivalent than what would be calculated based on a yield factor of 0.06 listed
in the ARCHIE manual and one estimate of 40 metric tons of cyclohexane in the
vapor cloud). The blast resulted in significant structural damage to
buildings in the town of Scunthorpe, 3 miles away, and was heard 25 miles away.
Houston (Pasadena) Texas 1989 Petrochemical Vapor
Cloud Explosion and Fire, 23 Fatalities
Image from www.mpri.lsu.edu/workshop/Phillips66Talk.ppt
|
On 23 October 1989, at about 1 PM, a vapor cloud explosion
and fireball occurred at the Phillips 66 Company polyethylene production
plant near Pasadena, Texas, killing 23 workers, one worker missing, and
injuring more than 130 others (one account said 314 people injured). The
accident was investigated by Occupational Safety and Health Administration,
the Environmental Protection Agency and other agencies; the OSHA report is
available at http://ncsp.tamu.edu/reports/phillips/first%20part.pdf
and a book on the incident is available through http://catalogue.nla.gov.au/Record/4183313?&#details.
|
The vapor cloud resulted as a sudden gas release of an
estimated 85,000 lbs of a flammable gas mixture through an 8-inch open valve;
the gas was a mixture of hydrogen, ethylene, hexane, and isobutylene. The
accident apparently occurred when operators were trying to remove solidified
polyethylene from the system. The vapor cloud (estimated to be roughly 1100
feet long, 800 feet wide, and 5 feet high) drifted downwind for about 90 to 120
seconds (another account said 5 minutes) before contacting an ignition source.
The wind was from the south at 10 to 16 mph. The vapor cloud plus additional
flammable gas inside tanks at the valve source exploded with a force equivalent
to 2.4 tons of TNT based on blast damage. Two isobutene storage tanks exploded
10 to 15 minutes later. A third explosion occurred 15 to 30 minutes later when
a polyethylene plant reactor failed catastrophically. The two polyethylene
production plants covering 16 acres near the source of the blast were
destroyed. Windows were shattered and bricks were ripped out in an
administration building 0.5 miles away. Fragments were thrown as far as six
miles away. The OSHA investigation uncovered a number of deficiencies
resulting in fines against the petrochemical company and its operating
contractor.
House, Natural Gas Explosion, 2008, Pennsylvania
location
Usually if a natural gas explosion levels a home, the cause
may be a defective furnace or stove or other leak inside the house. Odorants
(small amounts of sulfur-containing hydrocarbons giving a rotten egg odor) are
purposely added to warn homeowners of the leak. But sometimes a natural gas
explosion occurs without warning, and no odor was detected by neighbors or by a
postal worker at the home 30 minutes before the explosion. On 5 March 2008, shortly
after 1 PM, an explosion leveled this Pennsylvania home killing one man inside
and seriously injuring a 4-year-old girl. Two other houses were destroyed and
11 houses were seriously damaged. The mysterious accident was investigated by
the National Transportation Safety Board (report
http://www.ntsb.gov/publictn/2008/PAB0801.pdf).
From
National Transportation Safety Board Report DCA-08-FP-006
|
The
cause of the explosion was a corroded and cracked 2-inch natural gas
pipeline, at 10 psig, which failed resulting in natural gas entering the home
via a porous backfill for a nearby sewer line from the house. All this
occurred underground with no advanced warning. Apparently the home filled
quickly with natural gas as no one in the neighborhood noticed the natural
gas odorant. The concentration of natural gas reached the lower explosive
limit, the gas found an ignition source, and the house exploded.
|
Cracked natural gas pipeline feeding house
|
The report concluded excavation 5 years earlier for a
sewer line likely damaged the protective coating of the pipeline allowing
corrosion and eventual pipe failure to occur. The pipeline itself was not
affected by the house explosion.
|
Tank Flash Fire and Explosion, 2004, Houston Texas, no
one killed but several people injured by flying fragments
|
On
3 December 2004, a tank flash fire and explosion occurred at a processing
facility operated by Marcus Oil and Chemical on the southwest side of
Houston. The horizontal tank itself (12 foot in diameter and 50 foot long)
was propelled 150 feet where it impacted a warehouse (see yellow arrow at
left). The tank was part of a polyethylene wax processing operation, and
normally operated at 300oF and under a mostly nitrogen atmosphere
at about 45 psig. Later investigation by the U.S. Chemical Safety Board (CSB)
showed that that tank likely failed due to a defective weld at a tank patch
plate at possibly 69 psig, and that the nitrogen generator used to pressurize
the system contained as much as 18% oxygen. A fire resulted which burned for
7 hours. The blast caused structural damage and glass breakage up to 0.25
miles away. A 120 pound steel plate was found in a pasture 900 feet away. A
2-pound steel plate was found in the yard of a resident ¼ mile away. The
full report is available at
http://www.csb.gov/completed_investigations/docs/CSBMarcusOilCaseStudy.pdf.
|
Normally, the wax-like material inside the tank was not
considered flammable but is combustible. The CBS report concluded that at
about 69 psig, the defective weld failed severing a tank patch plate from the
tank. When the patch plate struck the concrete pad, sparks were generated
which ignited the wax and hydrocarbon vapors coming from the tank. The flame
flashed back into the tank. An internal deflagration blew the vessel head into
multiple fragments.
20-lb tank fragment
embedded in a wall 300 feet from company property line
|
750-lb tank fragment landed
near employees
(both photos from CSB
report)
|
Utah Refinery, 2009, Vapor Cloud Flash Fire, no
explosion, 4 people burned
This accident is under investigation by the U.S. Chemical
Safety and Hazard Investigation Board (image from
http://www.csb.gov/news_releases/docs/CSBSilverEagleStatement1.30.09.pdf.)
|
On
12 January 2009, at the Silver Eagle Refinery, Woods Cross, Utah, flammable
vapors were released from the tank vent at left while the tank was being
filled with light naphtha under atmospheric pressure. The resulting vapor
cloud found an ignition source, resulting in an ensuing flash fire 230 feet
long. Two refinery operators and two contractors suffered serious burns.
The tank contents, which contained an estimated 440,000 gallons of naphtha, did
not catch fire.
|
California Ethylene Oxide Explosion, 2004, 4 workers
Injured
On 19 August 2004, an explosion took place inside an air
pollution control device and medical products sterilization chamber at a
company in Ontario, California, injuring 4 workers and severely damaging the
facility. Ethylene oxide is used as the sterilization chemical for the medical
products; the chemical is both highly flammable and toxic (boiling point 51
oF,
Lower Explosive Limit 2.6%, Upper Explosive Limit 100%, Flash point -20
oG,
OSHA 8-hour permissible exposure limit 1 ppm). The Chemical Safety and Hazard
Board (CSB) Report is available at
http://www.csb.gov/completed_investigations/docs/CSBSterigenicsFinalReport.pdf.
Normal sterilization procedure is to first subject the medical products
contained in their normal packaging to high humidity and warm (80
o
to 120
oF) temperatures for 6 to 24 hours (depending upon the
product), followed by injecting gaseous ethylene oxide into specially-designed
chambers containing the products. Under these conditions, the ethylene oxide
penetrates the packaging to sterilize the medical products. When the
sterilization is complete, the chamber is evacuated followed by a series of
nitrogen and/or nitrogen-air flushes to reduce the concentration of ethylene
oxide to well below the lower explosive limit; the exhaust gas containing
ethylene oxide is routed to a scrubber. Additional air flushes are done to
decrease the ethylene oxide concentration in the chamber to below toxic levels
before workers can enter the chamber to remove the sterilized medical products.
The final air venting (containing small amounts of residual ethylene oxide) passes
through a catalytic oxidizer before exhausting to the outside.
However, on the day of the explosion, during a system
troubleshooting checkout with no medical products in the chamber, operators
bypassed some of the flushing steps to save time (a bad decision).
Furthermore, chamber had no monitoring or detection equipment to warn operators
that an explosive mixture of ethylene oxide remained in the chamber when the
chamber door was opened. This caused an estimated 50 lbs of ethylene oxide to
vent to the catalytic oxidizer and also into the building setting off nearby
LEL alarms. The ethylene oxide ignited in the catalytic oxidizer causing the
flame to flash back to the chamber resulting in a powerful explosion.
Fortunately, there were no employees in the chamber area at the time of the
explosion.
Illustrations, shown below, taken from the CSB report, are
useful in the understanding of what happened:
Sterilization chamber, after explosion
|
Catalytic oxidizer, after explosion
|
BP Refinery Explosion and Fire, Texas City, Texas,
2005, 15 killed, 180 injured.
This example was the subject of an earlier PEAC Newsletter
article, available at http://www.aristatek.com/newsletter/0612December/TechSpeak.aspx.
After we wrote the article in 2006, the CSB issued its final report on the
accident (March 2007) which is avilable at
http://www.csb.gov/completed_investigations/docs/CSBFinalReportBP.pdf.
On 23 March 2005, the worst U.S. workplace accident in 16
years occurred at the BP Texas City refinery when flammable vapors from a blow
down vent formed a vapor cloud and ignited, apparently from an idling diesel
pickup located about 25 feet away. The resulting explosion and fire killed 15
workers and injured 180. The accident occurred during startup of the
refinery’s octane-boosting isomerization unit (boosts octane content of
gasoline), when a distillation tower and attached blowdown drum were overfilled
with highly flammable liquid hydrocarbons (the major part of gasoline). Because
the blowdown drum was vented to the atmosphere, there was a geyser-like release
of flammable liquid and vapor onto the grounds nearby causing a series of
explosions and fires that killed 15 workers in and around nearby trailers.
Houses were damaged up to ¾ miles away from the site. The blast was felt up to
5 miles away. A shelter-in-place order was issued that required 45,000 people
to remain indoors.
Blast overpressure map, based on site observations,
structural analysis, and blast modeling. Over pressure circles at 10+, 5+, and
2.5+ psi; also 2.0, 1.0 (320 feet from center) 0.5 (540 feet), and 0.25 psi
(940 feet). The 0.1 psi overpressure line (not shown) occurred at about 2000
ft. from the center, consistent with observations of furthest identifiable
glass damage. Unfortunately, the CSB did not provide a scale for the map so
these are our estimates based on text statements in the report.
Estimated Vapor Cloud location. Field observations (red
line), dispersion modeling (solid colors).
From CSB report. Unfortunately, the CSB did not provide a
distance scale for the sketch.
The CSB report estimated that as much as 6730 gallons of
hydrocarbons was released to the outside as a vapor cloud and liquid which
pooled before the explosion and additional 855 gallons after the explosion.
The hydrocarbon mixture was 30.7% n-heptane, 29.5% iso-hexane, 15.2% n-hexane,
13% n-octane, 4.1% n-nonane, 1.4% n-decane, 3.8% n-pentane, and 2.6%
iso-pentane. The CSB modeled the vapor cloud using a class A stability and 80
oF
temperature. Based on explosion damage distance observations, our modeling using
the PEAC tool suggested that considerably less than 6730 gallons participated
in the explosion based on a TNT yield factor of 0.03; the rest formed a liquid
pool and a vapor flash fire (deflagration). The liquid pool was noted in the
CSB report, which estimated the pool radius increased to 27 meters at 106
seconds after the release, the time of the blast. The burned area was 200,000
square feet.
Illiopolis, IL Vinyl Chloride Monomer Explosion, 2004,
5 dead, 3 Injured, Community Evacuated
Image from CSB Report available at
http://www.csb.gov/completed_investigations/docs/Final%20_Amended%20030907%20rpm-b.pdf.
On 23 April 2004, an explosion and fire occurred at a
polyvinyl chloride (PCV) manufacturing facility killing five and injuring three
employees. The resulting fire spread to PVC resins stored in a nearby
warehouse, resulting in smoke drifting over the nearby community of Illiopolis,
Illinois. The community within one mile of the site evacuated for two days.
The facility was shut down and not rebuilt.
Vinyl chloride monomer is used in the manufacture of PCV. The
chemical has a boiling point of 7
oF (-14
oC), a lower
explosive limit of 3.6%, toxic (OSHA 8-hr permissible exposure limit of 1 ppm,
and a potential occupational carcinogen. The CSB report determined that the
explosion occurred when an operator drained a full and pressurized PVC
reactor. Another operator cleaning a nearby reactor apparently opened the
bottom valve of an operating releasing approximately 15000 lbs vinyl chloride
monomer, PVC, and other ingredients. The two operators and the shift
supervisor attempted to manage the release and did not evacuate and
subsequently died when the released vinyl chloride contacted an ignition
source. Two other workers were also killed, including one who died in the
hospital two weeks later.
The root cause was confusion on the part of an operator or
operators as to which valve to open, and that the interlock protection for the
bottom valve had been bypassed. The company had an emergency transfer
procedure to transfer reactor contents to another reactor in case of excessive
pressure buildup in a reactor which required them to open the bottom valves and
transfer valves to connect the reactor to another reactor, but the tank
contents were released to the room. According to accounts, there were several
explosions as the fire contacted other tanks.
Rocketing Exploding Flammable Gas Cylinders, St.
Louis, MO, 2005; cylinders launched into community
On 24 June, 2005, an unusually hot day in Saint Louis,
Missouri, at a gas filling and distribution facility, a small fire from one
propylene cylinder spread to other cylinders containing flammable gases.
Exploding cylinders and parts rocketed up to 800 feet into the surrounding
facility damaging property and starting secondary fires in the community.
There was one death due to smoke inhalation. The initiating event was captured
on the company security camera, shown below:
Security camera view, One cylinder catches fire (upper
left), 3:20 PM, the start of the incident.
|
Security camera, fire spreads after three minutes;
cylinders are starting rocket at two minutes
|
According to the CSB Safety Bulletin (available at CSB
website), the likely root cause was a combination of hot ambient temperature
and direct sunlight heating which caused one of the partially-filled propylene
cylinder internal pressure to build up. The safety release valve on the
cylinder opened and vented the propylene. The vented propylene gas found an
ignition source (probably static electricity) and caught fire. The fire heated
an adjacent cylinder causing it to vent propylene which also caught fire. A
domino effect occurred, eventually affecting about 8000 cylinders. The safety
release valve on the initial cylinder may have been set to vent at too low a
pressure as the gas cylinders were designed to withstand much greater pressures
than what would normally be produced in a cylinder sitting in the hot sun. The
normal minimum relief setpoint for propylene cylinders is 390 psig, which
corresponds to a temperature of 149
oF. On a very hot day with
cylinders sitting out in the sun, temperatures could exceed maybe 130 or 140
oF,
which could result in venting if the relief setpoint is too low.
The CSB investigation also noted that this type of accident
has occurred before at other locations, where propylene cylinders caught fire
and rocketed causing damage to the facility and surrounding community, all
occurring on hot days. The government organization put together a safety
training video on this hazard which can be viewed at
http://www.chemsafety.gov/index.cfm?folder=completed_investigations&page=info&INV_ID=59#.
Complicating the issue was that some of the acetylene
cylinders may have contained asbestos which may have been in the toxic plume
cloud. Older acetylene cylinders are filled with a porous material containing
asbestos. Noxious smoke and fumes from the fire resulted in a plume cloud over
1/3 mile wide and 1 mile long, and was attributed by the St. Louis Chief
Medical Examiner to cause an asthma attack resulting in death of one resident.
Following the incident, the company contracted a cleanup of the asbestos which
the Missouri Department of Natural Resources monitored.
Burned out vacant commercial
building, 300 feet away, from rocketing burning fragments
|
Burned out car, 200 feet
away; several cars catch fire due to burning cylinder fragments
|
Another burned-out car220
feet away due to rocketing cylinder fragment
|
Rocketing gas cylinder
glances off car and comes to rest here, 600 feet away
|
3-foot hole in side of
residential building due to rocketing fragment, 530 feet away
|
Fragment, 500 feet away
|
The above photos, taken from the CSB report show the impact
on the community. Fragments were found as far as 800 feet away.
Employees evacuated the facility within minutes, as
cylinders began exploding and flying into other parts of the facility at two
minutes after the start of the fire. The St. Louis Fire Department arrived on
scene at about 3:35 PM, 15 minutes after the start of the fire. Firefighters
evacuated local residents, directed a water stream on the fire, and
extinguished secondary fires started by cylinders propelled off site. The
quick evacuation undoubtedly saved lives. The fire was under control at about
8:30 PM.
While not mentioned in the CSB reports, the St. Louis Fire
Department took a great deal of risk in this situation while fighting fires and
evacuating people to a safe distance as exploding cylinders rocketed in all
directions. No one was injured by flying debris.
Role of the PEAC tool
The examples illustrated show the complexity of real-world
fire and explosion events. The PEAC tool is designed to strip away the
complexity and allow responders to examine the bare essentials. What are the
hazards? What harm can the chemicals potentially do? How far should people be
evacuated?
The PEAC tool allows the user to rapidly examine a lot of
“what if” situations. The emergency responder often does not know what the
situation is at hand and can run through several “worst case” scenarios in the
PEAC tool, and estimate a safe evacuation distance. Similarly, an outside
inspector can do a “walk through” at an industrial facility and look at
potential “worst case” releases or accident situations depending upon what
he/she sees. Where are control rooms or other buildings housing people located
with respect to places where a potential vapor cloud of a flammable gas can
occur?. What are the best places to site sprinkler and deluge systems, gas
detection alarms, and other safety devices? The PEAC tool can be used as a
rapid first assessment to be followed by more detailed technical analysis
specific of the site under evaluation.