Thursday January 31, 2008 - Vol. VII Issue 1
Runaway Industrial Chemical Reactions
[Download PDF for Printing]
The March 2006 issue of the PEAC Newsletter was devoted to
industrial accidents involving chemical reactors and looked at some
examples. We will look at two more
examples, both of which are or have been investigated by the U.S. Chemical
Safety and Hazard Investigation Board (CSB).
The results of the investigations have been or are being made public by
CSB with the objective of “emphasizing the importance of implementing comprehensive
safety management practices to control reactive hazards”, borrowing words from
one of their reports. The PEAC tool can
help industry identify risks and possible consequences during the manufacture
and storage of chemicals.
U.S. Chemical Safety and Hazard
Investigation Board
|
The U.S. Chemical Safety and Hazard Investigation Board
(CSB) is an independent federal agency charged with investigating industrial
chemical accidents at fixed facilities. The agency does not issue fines or
citations but does make recommendations to the industry involved and to
regulatory agencies and labor groups. It is designed to conduct scientific
investigations as to the root cause of chemical accidents and is not an
enforcement or regulatory body. Most of the Board members and staff have
degrees in chemical or
|
Mechanical or other engineering disciplines, have PE
licenses, have chemical process industry experience, or are health or safety
professionals. Congress in establishing CSB specifically stated (see 42 U.S.C.
section 7412(r)(6)(G)): “No part of the
conclusions, findings, or recommendations of CSB relating to any chemical
incident may be admitted as evidence or used in any action or suit for damages
arising out of any matter mentioned in an investigation report”.
CSB was authorized by the Clean Air Act Amendments of 1990,
but did not become operational until 1998.
A thorough CSB investigation of an industrial accident can take several
months, even sometimes over a year because of the complexity of the situation. First responders coming on scene of an accident
only have limited information as to what is happening.
CSB Report “Improving Reactive Hazard Management”
In 2002, CSB issued a report titled
Improving Reactive
Hazard Management. This 150-page
report can be downloaded by going to the CSB website,
http://www.csb.gov, and entering the report
name. The report documented 167 serious
reactive chemical incidents in the United States between January 1980 and June
2001 that resulted in 108 deaths, hundreds of injuries, and significant public
impacts. About 35% of the incidents
resulted from runaway chemical reactions.
Of the 167 incidents, 42% resulted in fire and or explosion, 37% resulted
in toxic gas releases, 16% resulted in both toxic gas and fire/explosion, and
the remainder (5%) involved a hazardous liquid spill only. Many of the runaway chemical reactions
occurred in reaction tanks that failed or even exploded because of thermal
runaway. The temperature of the
reaction increased rapidly resulting in increased pressure as liquids
evaporated, and the tank failed because of the increased pressure. Other incidents occurred because of
inadvertent mixing of incompatible materials, or chemicals exploded because of
instability. More than half of the 167
incidents involved chemicals not covered by OSHA regulations (20 CFR part
1910.119) or the EPA Risk Management Program regulations (40 CFR Part 68) at the time the CSB report was
issued in 2002.
Example Incident: T2
Laboratories, Inc. , Jacksonville FL, 19 Dec. 2007
Coast
Guard surveillance video watching Jacksonville harbor freezes the time of the
blast. Stack at left and two cooling
towers at center right are incidental.
From CSB website.
|
Coast
Guard surveillance video watching Jacksonville harbor freezes the moment the
blast is first seen from a distance, one second later, at left.
|
Coast
Guard surveillance video captures the scene five seconds later when the
explosion is the most brilliant (a vapor cloud explosion).
|
Arial view of site after blast from CSB bulletin
|
On 19 December 2007, at about 1:30 PM an explosion occurred
at T2 Laboratories in Jacksonville, Florida.
The explosion killed four T2 workers and resulted in hospitalizing 14
other people. Injuries requiring
medical attention occurred as far away as 750 feet from the site (35 people
according to a Jan. 25 report). The
blast was felt several miles away. Over
100 firefighters fought the ensuing blaze, which was described as a hellish
inferno.
Google
Earth image of T2 Laboratories site at 3043 Faye Rd before blast
T2 Laboratories is a small company employing about 12
people, and their facility in Jacksonville is their only production site. T2 Laboratories manufactured
methylcyclopentadienyl manganese tricarbonyl (MMT) under the trade name Ecotane
®. This chemical is used as a gasoline additive
to boost octane rating of gasoline and to help lower tailpipe emissions of NO
x. It is also used in refinery processing to
reduce emissions of nitrous oxide and increase the output of gasoline from
crude oil. Over one million pounds per
year are produced annually in the United States. More information on Ecotane
® produced by T2
Laboratories is in a paper written by R.S. Gallagher and M.F. Wyatt available
at
http://www.t2labs.com/ecotane/e118-T2%20Labs%20Report%20R041000T%20Introduction%20to%20Ecotane%20MCMT.pdf. Robert Scott Gallagher was one of the
people killed in the blast.
A
description of a procedure for manufacture of MMT is in a December 2006 report
prepared by the American Chemistry Council Petroleum Additives Panel as part of
an Environmental Protection Agency program and is available from the EPA at
http://www.epa.gov/HPV/pubs/summaries/mthmntri/c14889rt.pdf. The manufacturing process entails
the following: Under a nitrogen
atmosphere, methylcyclopentadienyl dimer is added to a dispersion of sodium
metal in diethylene glycol dimethyl ether.
A constant elevated reaction temperature is maintained to yield
sodium-methylcyclopentadienyl, which is an intermediate in the reaction
process. Manganese chloride is then
added to the stirred mixture containing the sodium – methylcyclopendienyl
intermediate. An elevated temperature
is maintained during the addition. Upon
completion, the reaction gives bis - (methylcyclopentadienyl)manganese, the
second intermediate of the reaction process.
The reaction vessel is then pressurized with carbon monoxide. The
addition of carbon monoxide results in MMT which is separated from the reaction
mixture via vacuum distillation.
The final product MMT is fairly safe
to handle, but very energetic chemicals are used in its manufacture. We (AristaTek) were not able to confirm that
the procedure summarized above was the one used at T2 Laboratories, but the CSB
statement issued January 3 seemed to indicate that it was similar.
On 3 January 2008 the CSB investigator
Robert Hall in charge of the investigation issued a public statement (available
at the CSB website). The facility
started producing MMT commercially several years ago using a batch reactor. The reactor blast occurred during the step
involving heating and reacting methylcyclopentadienyl dimer with
sodium metal. Prior to the rupture,
eyewitnesses reported hearing loud hissing and seeing vapor venting, which
indicated the development of excess temperature and pressure inside the
reactor. The reactor with its flammable
contents ruptured and the chemicals ignited releasing large amounts of thermal
energy (see Coast Guard Video pictures shown above). The blast also ignited various solvents at the facility creating
secondary fires and explosions. CBS
estimated that the pressure inside the reactor vessel must have reached several
thousand pounds per square inch. The
reactor vessel steel walls were 3 inches thick. The vessel’s head weighing several hundred pounds was located
about one-quarter mile away after the blast.
The explosion force was approximately equivalent to a ton of TNT. Debris was spread up to one mile from the
plant. At the time of the CSB statement
(3 Jan 07), the accident site still remained too hazardous for investigators to
enter, and that a plan for safe entry needed to be developed.
A visit to the T2 Laboratories
website reveals that the company was involved in producing a variety of
flammable specialty solvents which accounted for the major fire and hazardous
chemicals stored at the site..
CSB indicated that they plan to conduct chemistry testing
using T2’s recipe to better understand exactly what went wrong inside the
reactor on December 19. Their final
report is expected this summer.
Example Incident: Synthron, LLC, Morganton
NC, 31 January 2006
Illustration
of Synthron site after the blast from CSB Report 2006-04-1. Morganton Dept. of Public Safety.
Google
Earth image of the Synthron facility
before the blast
On 31 January 2006, a runaway chemical reaction and
subsequent vapor cloud explosion killed one worker and injured 14 people at the
Synthron, LLC facility in Morganton, NC.
The explosion destroyed the facility and damaged structures in the
nearby community. CSB investigated the
accident and issued a final report (No. 2006-04-I-NC) on 31 July 2007, which is
available from the CSB website, http://www.csb.gov.
Reactor M1 (from CSB final report)
|
The runaway chemical reaction at Synthron occurred in
their reactor M1 sketch at left. The
reactor had a capacity of 1500 gallons and was rated at 75 psig (pounds per
square inch gage) maximum. The
reactor is used to produce acrylic polymers.
In a typical operation, an acrylic monomer (purchased from a chemical
supplier) is mixed with various flammable solvents in the reactor, and then
steam is injected to heat the reaction mixture to a specified temperature
(usually near the mixture boiling point).
Then the steam is turned off, and a polymer initiating solution
metered into the reactor. The heat
given off by the reaction boils off the solvent which is condensed in the
overhead water-cooled condenser.
Liquid solvent from the condenser is drained back to the reactor. The system operates near atmospheric
pressure controlled by a vent on the condenser.
|
The acrylic polymer products produced by Synthron are used
for various coatings and paints.
According to CSB, the company had received an order for their product,
Modarez MFP-BH, which is a liquid acrylic polymer, and the order was for a
slightly greater amount of product than what the reactor was designed to
produce in a single batch. Operators
began preparing for the 6080 pound acrylic polymer batch the previous day which
was 12 percent greater than normal. The
chemical ingredients were scaled up to take care of the increased polymer
product, but because there was insufficient aliphatic solvent on hand in
storage the operator actually scaled back on the on the aliphatic solvent.
On the day of the explosion, operations appeared normal
until after the steam was turned off and the polymer initiating solution was
pumped into the reactor. The operator
in charge noted that initially the reaction did not proceed as vigorously as
expected, but later the solvent evaporated and the condensed solvent flow
returning to the reactor appeared within normal range. A few minutes later, the operator heard a
loud hissing and saw vapor venting from the reactor manway. The irritating vapor forced him out of the
building. Three other employees also
left the building because of the vapors.
The operator then reentered the building wearing a respirator and was
able to start emergency cooling water flow to the reactor. The building exploded less than 30 seconds
after he exited the second time. The
blast injured the operator and five employees who had exited the building
including two seriously. The
maintenance supervisor who was on a lower level by the laboratory near the
manufacturing area was killed. The
other injuries occurred to employees in a nearby trailer and to two citizens
driving by the site.
The blast damaged buildings nearby. Two church buildings and one house were
condemned. Glass was broken up to
one-third a mile distant from the site.
The Morganton Department of Public Safety responded rapidly calling
mutual aid from the county and surrounding municipalities, to assist injured
employees and extinguish the resulting fire.
Local residents were asked to shelter-in-place for several hours.
The CSB final report blamed the explosion on the following
combination of circumstances:
- Because
there was a shortage of the aliphatic solvent in storage, the operator
actually decreased the amount charged to the reactor by 12% compared with
the standard recipe, and increased the acrylic monomer by 12%. With the adjustments made to the
reactants to manufacture everything in one batch but with different
proportions of chemicals, the heat release was at least 2.3 times that of
the standard recipe.
- The
waterside of the condenser had apparently never been cleaned and was
fouled and could not remove the excess heat release as the solvents
boiled. Once the heating rate
exceeded the condenser cooling capacity, control of the reaction was lost
resulting in a runaway reaction.
- Only 4
of the 18 clamps specified by the manufacturer were tightened for the
manway cover. This was a
labor-saving step as it was a long-standing practice to open and clean
reactor tank after every batch.
The manway began to leak vapors (the hissing sound reported) when
the pressure reached approximately 23 psig. The flammable vapors filled the room and ignited.
The CSB report also stated that Synthron had no chemical
engineers or other engineers on staff, and none had been contacted to evaluate
the hazards associated with the reactive operations at the site. There was no comprehensive process hazard
analysis review.
Our Analysis
Both the T2 Laboratories and Synthron accidents involved
vapor cloud explosions that were initiated when flammable chemicals escaped
from batch reactors where runaway chemical reactions took place. In the T2 laboratory situation, a runaway
reaction caused the temperature and pressure to increase inside the reactor,
which resulted in reactor failure when a very high pressure was reached. The heated flammable contents gasified and
exploded in a vapor cloud explosion equivalent (according to the CSB) to a ton
of TNT. The root cause of the reaction
runaway has not been determined as of January 2008, but several scenarios can
be postulated including contact of sodium with accidental water in the system,
or failure of a temperature control. In
the Synthron accident, the flammable vapors vented into the room at the manhole
cover of the reactor, and exploded in a vapor cloud explosion a few minutes
later.
AristaTek does not have available the amounts of chemicals
or the reactor arrangement for the T2 Laboratory situation, but the CSB
indicated that the explosive power based on damage to the surroundings was
approximately equivalent to a ton of TNT.
Several chemicals were present which could have been under pressure at
the time the reactor top blew off, including methylcyclopentadienyl, sodium, and diethylene glycol dimethyl
ether (or a similar material).
Diethylene glycol dimethyl ether has both oxygen and hydrocarbon fuel in
the same molecule and would be expected to have a fairly high yield factor (the
fraction of the energy in a chemical which participates in a vapor cloud
explosion, the rest is released as heat in a fireball). This means that less chemical would be
require to produce the same punch as one ton of TNT compared if say an
aliphatic hydrocarbon such as heptane had been involved. Additionally, the reactor chemicals were
under great pressure which was suddenly released, which means that the entire
batch probably participated in the explosion and fireball.
The CSB report did not identify the chemicals or the recipe
used in the Synthron situation, only their boiling points. The chemicals consisted of (1) an acrylic
monomer with a normal boiling point temperature of 297oF, (2) an
aliphatic solvent with a normal boiling point temperature of 178oF,
and (3) an aromatic solvent with a normal boiling point temperature of 234oF. The monomer amount was increased 12% over
the standard recipe. The recipe called
for an equal amounts of aliphatic and aromatic solvents, but because there was
insufficient aliphatic solvent in storage, the operator actually cut back on
the aliphatic solvent and only increased the aromatic solvent by 6% according
to the CSB report, which increased the maximum heat output.
The CSB report did not state the TNT equivalent for the
Synthron vapor cloud explosion, but the remark that glass was broken up to
one-third mile away roughly is equivalent to several hundred pounds of TNT based
on glass breakage.
The PEAC Tool
The PEAC tool is designed for use of a first responder or
for use by industry for estimating the consequences of a potential dangerous
situation. The user enters name of the
chemical and amount or a tank or vessel size and the PEAC tool calculates the
consequences if the chemical is released or participates in a vapor cloud
explosion. The PEAC tool menu is set up
for the user to enter information this way.
The PEAC tool menu is not set up to do a “reverse engineering” analysis,
where the user enters blast damage observations and different distances from
the source and the PEAC tool calculates the amount of chemical or a container
size which could potentially produce the damage observed. Additionally, because of uncertainties of
the vapor cloud shape at the time of detonation, the PEAC tool incorporates a
factor of two on the distance. The PEAC
tool also assumes that the entire tank contents or amount specified
participates in the explosion and fireball using the same yield factors as in
the ARCHIE model. These assumptions are
explained in the PEAC tool manual and on the PEAC tool disclaimer statement.
Let us look at a hypothetical example where 3000 lbs of
cyclohexane is vaporized and participates in a vapor cloud explosion. We began by pulling up “cyclohexane” in the
PEAC tool. Cyclohexane is an aliphatic
solvent with a boiling point of 178oF. We will make sure “mass” is selected under “options” so we don’t
have to fuss with entering container sizes.
To initiate the calculation we
select the fireball icon which appears on the PEAC tool. A statement then appears on the screen,
under “Important Information”.
|
Several screens then appear. The user may select the coordinates of the
accident to have an overlay of the blast damage on a map. The user may select a container size or
simply state the amount of chemical involved. We will dispense with these extra steps and simply enter 3000
lbs of cyclohexane and ask the PEAC tool to compute the damage distance to a
0.5 psig overpressure, and the distance to second degree burns for an
unprotected human. The screen for the
overpressure selection and fireball distance for the second degree burns
appear at left. Because of
uncertaintities on the vapor cloud shape, a factor of two is built in the
distance display.
|
The final PEAC tool display (without
an overlay on a map) is as follows:
|
|
The display is followed by
distance estimates to varying blast damage for the 3000 lbs of cyclohexane.
|
|
|
|
|
The PEAC tool is designed such that
the user can enter different numbers rapidly.
This makes the tool useful for doing a potential consequence analysis
during a walk through of an industrial facility.