OCR error (C:\Conversion\JobRoot\00000AOH\tiff\2000W9PM.tif): Unspecified error
-------
Regional Center for Environmental Information
US EPA Region III
1650 Arch St.
Philadelphia, PA 19103
-------
Federal-State Task Force Report
Review of Methyl Isocyanate (MIC) Production at the
Union Carbide Corporation Facility
Institute, West Virginia
TABLE OF CONTENTS
Executive Summary and Conclusions .............................. page i
Introduction [[[ page iv
Chapter One
Manufacturing and Reactivity of MIC ............................ page 1-1
Chapter Two
Mechanical Review of MIC Production ............................ page 2-1
Chapter Three
Emission Control Systems ....................................... page 3-1
Chapter Four
Safety Program ................................................. page 4-1
Chapter Five
Detection and Notification of Emissions, Spills, and Releases ..page 5-1
Chapter Six
Contingency Plans ............................ . ................. page 6-1
Chapter Seven
Handling and Storage for Transportation ............. ..... . ..... page 7-1
Table 1: Engineering/Instruments Associated with Safety ....... page 4-2
Systems
Documents Reviewed by Task Gruov .......... . ............ .. ...... page 8-1
Attachments
Attachment One: Bhopal Methyl Isocyanate Incident Investigation Team
Report, March 1985 - Union Carbide Corporation, Danbury, Connecticut
Attachment Two: Summary of Regional Response Team Comments Regarding
West Virginia Contingency Plans
Attachment Three: Membership of Federal-State Task Croup on the Review of
-------
Executive Summary and Conclusions
/ SUMMARY /
December 3, 1984^more than 2,000 people were killed in Bhopal, India.\
Catastrophic release of methyl isocyanate (MIC). Union Carbide's Institute,
slant is the only facility in the United States that produces MIC. As a
of the Bhopal accident, production of MIC at Institute was suspended
in December. On February 12, 1985, Union Carbide announced its intention to
resume production of MIC.
Prior to resumption of MIC production at Institute, West Virginia, various
Federal and State regulatory agencies reviewed the facility and associated
manufacturing process to insure that production of MIC would not pose an
imminent and substantial endangerment to the health and welfare of the
residents of the Kanawha Valley.
f 5 £.*««-.
To accomplish this, a ffederal/^tate task gc«tfp^*was formed in February 1985
to review and evaluate the design and operating practices of. the MIC production
and storage units at Institute. This report was prepared by EPA staff with
review and comments by task force members. The agencies and members of the
task group are listed in attachment three.
The task-force based its review primarily on information provided by
Union Carbide in the form of documents, interviews and plant inspections.
The task-force independently evaluated information using basic principals-
of engineering, physics and chemistry, standard industrial practices f-e.^
and technical judgtaent. Although the task-force evaluated the entire
manufacturing process, including formulation, storage, handling and
transport of MIC, the primary focus of the evaluation concentrated on
storage, since this was considered the area of greatest potential for
catastrophic releases.
The following information was considered critical for determining
if any imminent and substantial endangerment to public health exists:
1. Hazards of MIC;
2. MIC process design and materials used for construction of equipment;
3. Emergency vent gas scrubber and flare control equipment design;
4. Safety systems; malfunction warning systems; and release detection
systems;
5. Documented operating procedures and maintenance programs;
6. Emergency contingency planning; and
7. Handling and storage for transportation.
-------
safety improvements that are scheduled to be in place at the Institute,
facility prior to resumption of MIC production are:
1. Replacement of brine coolant with chloroform for the unit storage tanks;
2. Installation of redundant storage tank monitoring instrumentation;
3. Modification of the design of the safety valves and vent lines for the
storage tanks;
4. Increased caustic storage capacity for the emergency vent gas scrubber;
5. An additional tank which increases residence time for liquid MIC to be
in contact with the caustic in the scrubber during emergency operation !
6. Provision for steam addition to maintain temperature in the emission
vent scrubber to achieve better reaction rates '
7. Improved caustic concentration control for the scrubber '
8. Modifications to the flare ignition system to increase its reliability -
x1
9. Installation of an air sampling leak detection system for MIC •
10. Installation of a computerized emission warning system (SAFER®)
that would assist emergency response officials in deciding corrective
actions that should be taken if a significant release of MIC were to
occur •
X
11. Improved internal emergency contingency plans and early warning
notification procedures establishing specific criteria for initiating
notification of local emergency response organizations
/
12. Additional alarm systems for the storage tank monitoring instruments,
are being directly wired to the shift administrator's office.
Records reviews and inspections will be periodically conducted by various
regulatory agencies to assure that all critical equipment has been properly
maintained, all operating procedures are being followed, and all necessary
operator training has been provided.
In addition, Union Carbide Corporation([willjwork with the local
emergency response planning authorities tfb develop coordinated emergency
response plans.
ii
-------
CONCLUSIONS :
Major findings of the task-force review of the MIC units at
Institute, West Virginia are:
0 Major equipment and mechanical features of the plant are
state-of-the-art or equal to the chemical Industrie*
accepted practice.
Oh*
0 Standard opening procedures for inspection and maintenance
of critical equipment are in place and acceptable. Union
Carbide maintains an acceptable program for training MIC
operating personnel.
0 The design capacity ofjthe^ emergency vent gas scrubber and
flare are adequate ggnutralize.\nd/or destroy an "extreme
case" accidental release of MIC.^v^ _
0 Instrumentation, alarms and control device's are adequate
to indicate abnormalities in production, storage and- handling
of MIC.
0 Internal and external contingency plans for accident
mitigation are in place. These plans are continually being
reviewed and improved. Important changes to the local
emergency response contingency plans, as recommended by
the Regional Response Team, have been addressed in verbal
agreements between Union Carbide, Kanawaha County and the
State.
fe> -«."« —
Based on these findings, the task .gjwflp concludes that resumption of
production of MIC at the Institute facility does not present an imminent
and substantial endangerment to public In ilrhjir flir rmrlrrmmrnti
iii
-------
Introduction
On December 3, 1984, methyl isocyanate (MIC) escaped from a Union
Carbide Company (UCC) plant in Bhopal, India, killing 2,000 people and
injuring thousands more. As a result of this tragic accident, Union Carbide
shut down MIC production facilities at its plant in Institute, in the Kanawha
Valley of West Virginia, the only place in the United States where MIC was
being produced. Remaining inventory of MIC at Institute was converted to
product or destroyed.
On February 12, 1985, Union Carbi
production of MIC at Institute
jnced its intention to resume
April
"*N^-- - F" _ '"
Three days later, oarTFebruary 15, 1985, Stanley Laskowski, Acting
EPA Region III AdiWnfep^tor at the time, indicated by letter to Union Carbide
that EPA would review—the company's start-up procedures, and requested
information relevant to that review.
An intergovernmental task force was formed^e-'earry out the review. "/ *\e_-c_
The agencies and members of the task gpectpare listed in attachment Jtwor*
Task force members interviewed Union Carbide officials and reviewed technical
material that the company submitted (see Documents Reviewed by Task Group) „
The purpose of the review was to determine if resumption of MIC
production in Institute would pose an imminent and substantial endangerment
to public health or the envigeaaant in the Kanawha Valley, and to determine
the potential for a catastrophic release of MIC during its manufacture,
handling, or storage. The task force based its review primarily on information
provided by Union Carbide in the form of documents, interviews and plant inspec- i
tions. The task force independently evaluated information using basic principals"""
of engineering, physics and chemistry, standard industrial practices and technical
j udgement.
This report presents the findings of the task force.
iv
-------
Chapter One
Manufacturing and Reactivity of MIC
PURPOSE;
To describe the basic chemical and physical characteristics of MIC, the
process used in its manufacture, and potential health effects of human
exposure to MIC.
DISCUSSION:
MIC is a stable but reactive toxic, volatile, and flammable compound. It
is used as a chemical intermediate in the manufacture of agricultural
pesticides such as SEVIN® and LARVIN®.
1. Reactivity. Because MIC is particularly reactive with water,
acidic materials, alkalies, amines, and their salts, contact with
these substances is to be avoided. However, MIC's reactivity with
water and alkaires^ also allows it to be destroyed rapidly using
a caustic scruber system. In the reaction with water, MIC produces
:arbon dioferde^and dimethyl urea. In the presence of catalyzing agents
such as iron, MIC may combine with itself to produce its triraer, a
larger molecule made up of three MIC molecules.
Both trimerization and reaction with water are highly exothermic,
or heat-producing, yielding respectively 540 and 585 Btu per pound of
MIC reacted (see Methyl Isocyanate F-41443A-7/76, Union Carbide).
In such a reaction this heat, if not removed, would tend to vaporize
liquid MIC which, at ambient conditions, is a colorless liquid with
a boiling point of 39°C. If this vapor were to be released by a
pressure relief valve it would need to be treated with a scrubber
or burned in a flare to prevent a release to the environment.
Reactivity is controlled in storage through rigid quality assurance
practices and through refrigeration, which keeps the bulk temperature
of the MIC at or below 0°C. Equipment is also kept closed and the
MIC is blanketed with dry nitrogen to prevent contamination, which
could lead to undesirable reactions.
Production. MIC is produced by reacting anhydrous monomethylamine
(MMA) with phosgene in a high temperature vapor phase reaction.
Phosgene is produced by reacting chlorine gas with carbon monoxide.
Phosgene and carbon monoxide are produced on-site at the Institute
plant; the MMA and chlorine are transported to the plant by rail.
1-1
-------
The MIC manufacturing process yields hydrogen chloride (HC1) as a
by-product. This is removed from the MIC by subsequent refining
processes using conventional distillation techniques. Chloroform
is a system solvent. A simplified explanation of the overall
chemical reaction can be expressed this way:
CH3NH2 + COC12 y CHaNCO + 2HC1
Monomethylamine + Phosgene yields MIC + Hydrogen Chloride
A simplified block flow diagram of the process looks like this:
MIC PROCESS
-
PHOS
MOMOME1
ST—
rHYLAM4N£
PHOSGENE
STRIPPING
t
REACTION
CVSTFll
A
•iTAtl C^^ta.
1
1
1
1
RESIDUES
r
T
CRUDE
MC
MIC
DEFINING
PHODUCT
MC
•
CHLOROFORM
HYDROGEN I DESTRUCTION
CHLORIDE I VGS/FLARE
hSbT%
t [
MIC
DERIVATIVES
UNIT
VENTS
At Institute, MIC is stored and handled in corrosive-resistjtat equipment
made of stainless steel or containing appropriate linings such as
HASTELLOY®, INCONEL®, or TEFLON®.
1-2
-------
Once manufacturing and refining are complete, MIC is temporarily stored
as a liquid in a refrigerated tank, where it is sampled to assure compliance
with product specifications. These^require the MIC to be essentially free
of water, chloroform, hydrogen chloride, and other impurities which might
lead to later undesirable reactions. If the MIC meets specifications, it
is pumped to larger underground storage tanks; if it does not meet specifi-
cations, it is reprocessed or destroyed.
3. Health effects. MIC is considered a hazardous material by any
means of contact. Because of its high vapor pressure, the potential for
inhalation exists if MIC reaches the environment. Exposure via inhalation
can cause chest pains, coughing, choking, asthma-like breathing, and fatal
pulmonary edema. Skin contact can cause severe burns. Even at very low
concentration MIC can irritate or injure the eyes.
Exposure to phosgene, used to produce MIC, can cause eye irritation,
vomiting, coughing, dyspnea, chest pains, cyanosis, skin burns, and dry,
burning throat, and can be lethal.
CONCLUSION:
MIC is a reactive, volatile, and toxic material that can be safely handled
if appropriate corrosion-resistant material, refrigeration, and dry nitrogen
blanketing are used in processing and storage.
1-3
-------
Chapter Two
Mechanical Review cf MIC Production
PURPOSE:
To assess the design and materials of equipment at the Institute, West
Virginia Plant; to contrast them to the industry standard; and to determine
their relative safety in MIC production, storage and distribution.
DISCUSSION;
The extremely reactive nature of MIC requires that storage and process equpiment
receive special design consideration.
1. Reaction equipment. As described in the previous chapter, MIC is
made by combining monomethylamine and phosgene. This reaction takes
place at high temperature with excess by-products being directed to a
a flare for destruction before venting to the atmosphere. Reaction
equipment includes several unit operations such as reaction vessels,
distillation columns, ancillary reboilers, condensers, and scrubbers
to destroy the hydrogen chloride gas generated in the reaction. T
process equipment which contacts the HCL gas are of graphite, Hastolloy®
and other highly corrosion resistant materials of construction. All
storage and transfer material of construction are of corrosion-resistant
material such as stainless steel. Piping is welded stainless steel with
gaskets of TEFLON®, or better material, to prevent leakage. Such
equipment is the accepted norm for corrosive materials. If checked
routinely for stress corrosion, it represents acceptable equipment.
2. Storage equipment. The following is a list of the MIC storage,
process and relief tanks at Institute:
Quantity
3
Type of Tank
Process tanks
(make tanks)
Underground
storage tanks
Relief tank for
MIC
Maximum Designed Storage Capacity*
14,000 gallons each (one always
kept empty)
30,000 gallons each (one always
kept empty)
30,000 gallons
2-1
-------
Quantity
1
1
Type of Tank
Process tank for
LARVIN® production
Relief tank for
LARVIN® production
Process tank for
SEVIN®
Transportation/
Distribution tanks
Relief tank for
transportation
distribution
Maximum Designed Storage Capacity*
7,000 gallons
8,500 gallons
\
/
7,000 gallons (will no longer be
used)
15,000 gallons each^ (will^no
longer be used; huwavei , way be
trp*! "fired withJB.OOO 'gallons ..
shipping containers)
10,000 gallons .s*dr (will no
longer be used)
* Standard operating procedures call for total loadng/not to exceed 85-90
percent of storage capacity. >^
MIC is piped to pressurized process tanks. The contents of each process make
tank are pumped through a heat exchanger, which is cooled by a refrigerant.
Prior to the Bhopal disaster,Union Carbide used brine as the refrigerant^X&ut
has agreed to replace the brine with chloroform before resuming production.
Chloroform is inert to MIC at normal operating temperatures. The cooling system
is a key to controlling potential runaway reactions of MIC in the tanks since^—-*
MIC is less reactive at reduced temperatures. For example, the reaction rate
of MIC and water is 3.45 times faster at 25°C when compared to the reaction
rate at 0°C. '
Duplicate pumps (one operational and one spare) circulate the MIC through
the cooling system. For added safety, one of these pumps is electrically
powered; the other steam driven. Back-up electrical substations are provided
for each pump to increase the reliability of the power supply. The pumps are
internally sealed to prevent low level leakage to the environment. This
type of pump is the accepted industry norm for hazardous liquids which
cannot be released safely to the environment.
MIC that meets specifications is transferred from the process tanks to
30,000 gallon ca
The process an
that are set s
inch (psi) pre!
These relief valves
.underground storage tanks with similar safety equipment.
tanks are now equipped with new pressure relief valves
enting to the scrubber will occur at 20 pounds per square
the make tanks and 30 psi pressure on the storage tanks.
rere previously set at 50 psi.
tit
2-2
-------
EmgxgaBcy Vyft Gas Scrubber. The capacity of the scrubber to
emergency releases has been increased by the following
fations:
a. Caustic storage/capcitv/has been increased by 70,000 gallons
b. Liquid residence-^CTme has been increased at the base of the
scrubber to assure for more complete neutralization of liquid MIC.
c. The ability to maintain the operating temperature of the scrubber
has been assured by the use of a steam supply. This assures a rapid
rate of reaction with MIC.
d. An instrumentation system has been installed to control the
caustic concentration at 10% during emergency operation.
e. The feed nozzle configuration into the scrubber has been modified
to prevent backflow of caustic into the storage tank.
Additional Reaction Equipment. Under normal operations, after MIC
leaves the storage tanks, it is fed to process operations where it is used
to manufacture SEVIN®. Equipment in these operations is also of corrosion-
resistant stainless steel, with vent gas scrubbers and flares similar to
those in the storage tanks.
CONCLUSION:
Equipment used in the MIC process at Institute represents, in many cases, the
state of the art in the chemical industry and, in all cases, accepted industrial
equipment for hazardous materials. The possibility of a catastrophic release
due to poor design or materials of construction at Institute is remote.
2-3
-------
Chapter Three
Emission Control Systems
PURPOSE:
To review the reaction kinetics of MIC in order to determine the volume of
gas discharged under an "extreme case" event; and, by reviewing the design
calculations for the flare and scrubber, to assess their adequacy for
controlling emissions from a runaway reaction of MIC under "extreme case"
conditions.
DISCUSSION;
Union Carbide describes an "extreme case" event as a situation in which
1) the MIC in one of the underground storage tanks is contaminated with
water at the limit of solubility at ambient conditions,2) the refrigera-
tion and back-up refrigeration systems of the tank have failed, and 3)
it is not possible for some mechanical reason to transfer the liquid MIC
to the scrubber for destruction.
Union Carbide chose as their "extreme case" the contamination of their
largest storage tank. This is because, all else being equal, the larger
the tank the greater the potential hazard. Within the Institute plant
there is a second MIC storage location, a derivative production site,
with one (smaller) tank and a smaller emergency scrubber followed by a
flare. The same design philosophy was followed at this second scrubber
as was followed at the larger scrubber analyzed in detail in this chapter
and the conclusions reached in the examined "extreme case" apply to the
smaller derivative storage tank and scrubber.
It should be noted that this "extreme case" event, though highly unlikely,
is still not the worst case that could occur. For example, it would be
possible, although even much more unlikely, that more than one tank could
be contaminated at the same time, or that a tank could be contaminated
by caustic or other material, which would react faster than water.
A caustic scrubber and a flare are used to control relief valve_dischar%es.^
The scrubber neutralizes the discharges by breaking them^W^arbon dioxide~~^> ^-
and urea compounds and renders them harmless; the flare bUrrisT any discharge °
that might escape neutralization in the scrubber. i
The first step in determining whether the scrubber and flare are adequately
designed is to calculate the maximum rate at which gas could be discharged
3-1
-------
from the storage tanks during an "extreme case" event. This calculation is
done by computer model developed by Union Carbide which performs a heat
balance and calculates the temperature and pressure resulting from simultaneous
trimerzation and MIC-water reactions. Two kinetic parameters — the activation
energy and the frequency factor — and the heats of reaction are required to
determine the rate for each of these reactions. Union Carbide has determined
the kinetic parameters experimentally.
1. Scrubber design. The answers to four questions help determine the
adequacy of the scrubber design:
a. Will the scrubber flood at the maximum off-gas rate expected?
Flooding occurs when the gas flow rate to the scrubber is so high
that it prevents the caustic scrubbing solution from flowing down
through the packing and reacting with the MIC. The scrubber flood
point is calculated—££pm the dimensions and shape of the scrubber
packing and th^VationJof liquid to gas flow rates through Q
scrubjjejc~---~r£~~ref lects the maximum gas flow at which the^lfqluid
no longer pass downward against the rising gas.
b. Is the caustic rate (in pounds per hour) and the amount of caustic
available (in pounds) enough to control the peak MIC emission rate
and the total MIC emission? This calculation is based on the
stoichiometry of the reaction and a material balance.
c. Can the caustic-MIC reaction take place without an excessive
temperature rise which would decrease scrubber efficiency? If
the scrubber gets too hot, the scrubbing solution can boil and
the chances of flooding are increased. This calculation is based
on heats of reaction and dilution and heat balance.
d. Will the MIC vapor and the caustic solution stay in contact
with each other long enough for the neutralization reaction to be
completed? This depends on the number of transfer units in the
scrubber. (A transfer unit expresses the relative difficulty of
absorbing a gas with a given liquid. It is based on measurements
of absorption efficiences at varying liquid flow rates and varying
inlet gas concentrations with the given scrubber packing. The
transfer units can be related to removal efficiency.)
calculation requires experimental data which Union Carbide
has developed by testing another scrubber at its plant in
Woodbine, Georgia.
An examination of all of the above factors shows that the scrubber is larger
than required to control the calculated maximum storage tank off-gas.
Estimated control efficiency is calculated at over 99.9 percent.
3-2
-------
2. Flare design. Steam assisted flare combustion efficiencies depend
primarily on the heat content of the combusted gas and the exit
velocity from the flare. In "extreme case" conditions, the maximum
off-gas would be controlled by the scrubber, with less than one tenth
of one percent going to the flare for final cleanup. In conditions
even worse than this, if the scrubber were out of commission, the
entire maximum off-gas release could go to the flare. /
Even under the latter conditions, the gas mixture^in the flare conforms
to the EPA guideline's, for minimum heat content and maximum exit velocity
to meet 90xperscent^combustion efficiency.
CONCLUSION:
The scrubber and flare are each capable of controlling the described "extreme
case" discharge from the storage tanks. The scrubber alone could reduce
emissions by 99.9 percent; the flare alone, by more than 98 percent. As
a result, the total estimated emissions of MIC with both the flare and
scrubber operating, under the "extreme case" is less than 10 pounds per hour.
These reductions will be adequate to protect public health and the environment.
During normal operation, both devices operate in series, with the flare combust-
t/ng any gas not neutralized in the scrubber.
9?,?
c. vt/iH-i*^-
«*
,01.
3-3
-------
Chapter Four
Safety Program
PURPOSE:
To determine if Union Carbide's safety programs are adequately designed to
1) prevent a release of MIC into the atmosphere, and 2) minimize any chance
of public harm should such a release occur.
DISCUSSION:
Engineering/instrument design. The task force felt that equipment
associated with preventing and/or controlling MIC emissions into the
atmosphere should: a) provide ample warning of any malfunction or
runaway reaction, b) be redundant, i.e., have back-up, and c) initiate
corrective actions through alarms or other systems. The following
information addresses these criteria for all critical instruments
and equipment.
4-1
-------
CO
S
Ol
4-1
CO
CO
CO
cu
Si
B CJ
CO eg
0)
U-l P
U-l
O O)
p
CO 3
4J 4-1
01 CO
CO M
01
01 O.
rH S
s
Ol
4-1
CO
^
co
60
B
•rt
B
M
CO
3
O. CU
3 JJ
O
CJ CU
2 ^
e u
CU U-l
Si iH
4J
e
-E P
O cfl
CO -H
Cd CO
g
p
CO CO
i-( S
cfl t-i
CO
I-l i-l
01 cfl
CU Si
3 0 si
4-> E H
CO CO
p
0) * •
G. 01 O
S CO o
cu IH m
4J P
4J
Ol O CO
jE J— *
4J U-l
CO U-«
u-i U) O
H-l 3
B 4J
•rt 0)
. 4-1 CO
CJ E
o O CO
CM U iH
•
Q
o
o
M
1-1
o
1-1
4J
E
O
CJ
cu
B
•H
0>
M
CO
B
CO
U-l
U-l
o
CO
4J
CU U-l
co o
Ol 01
1-1 4-1
a. co
3 P
o
CJ 01
O si
CU U-l
-C IH
4-1
S
CJ CO
CO i-H
Ed CO
•
^4
D
CO O
•f^ J3
^•SS.
cu cj
600
B in
CO •
CJ
B
0) CO
^ Si
3 4J
eg P
P Ol
01 4J
a. cfl
E 0)
0) P
4-1 QO
I— 1
o
p
4J
B
O
CJ
01
si
4-1
S
CO
•rt
S
p
cfl
cfl .
e
CU O
•E O
E™1 M
4J
01
CO Jrf
Ol CO
P 4J
cfl
cu
CO Si
U 4J
o
CO U-l
B •rt
cu
CO CO
cu G
M a
3 <-!
CO CO
CO
CU W
p p
a. co
4J
J2 CO
4-1
o o
pa u
o
1— 1
•o
B 01
CO Si
4J
•rt
to E
CU i-<
CM CO
•rt
CO
CU B
js C
O cfl
Cfl t-l
01 CO
p
O)
CU JS
p El
3
CO
CO •
01 -rt
P CO
a a.
•
s
o
o
1— 1
o
4J
B
o
CJ
J=
t u
co -a 4-1 co
P B 3 CU 0)
O CO JS Si
4J Cfl Si 4-1
Cfl CO 4J
CJ S O -rt B
•rt P 1-1 -O 3 iH
T3 Cfl 4J E
E i-l CO CO CO CO
•rt Cfl B 0) iH
O -O E
i-H O *J Ol 1-1 B
Ol 4J 3 Ol iH P
> CO u-i cfl
CU T3 C i-l •
I-H oi e oi o co B
4-1 co si •** o
T) O 4J 4J 01 O
•rt 01 Ol Cfl £ P
3 E > P i-l H
CT E cfl o 3 •-»
•H o .e u-i u o
i-t CJ p . p
O B i-l £t 4J
Ol Ol CO 3 CJ E E
j: P 1-1 o oi cfl o
H cfl cfl -o p 4J 0
(0
CO
.fl
CO
H
4J
*_4
[J
TJ
01
4J
CO
•rt
CJ
o
CO
CO
^
CO
4J
E
s
3
p
4-1
CO
E
I-H
^^
60
B
•rt
p
Ol
01
B
•rt
60
Cd
f>»
CJ
B
Cfl
•a
B
3
•a
01
05
CO
0)
CO
p
o
4J
CO
o
•rt
4J
B
3!
s
a
•rt
3
cr
w
U-l
o
01
to
o
a.
p
3
0-
T3
E
4-1
CO
•rt
U-l
Ol
4-1
UH
o
0)
E
o
T3
1-1
3
o
CJ
JE
CJ
•rt
JS
3
B
o
•rt
4-1
CJ
cfl
Ol
p
CO
'•W
o
B
o
•rt
4-1
CJ
Cfl
p
^
cfl
3
cfl
3
p
cfl
O)
E
o
u
Ol
JD
0)
J=
4-1
B
•rt
Ol
09
•rt
p
01
3
4-1
CO
Ol
a.
B
cu
4-1
Cfl
CO
•rt
p
a>
4-1
cfl
3
4)
CJ
B
•rt
CO
Itf*
E
CO
cu
60
Cfl
o
4-1
CO
B
CO
CO
cu
4-1
co
•rt
4-1
•rt
B
•rt
B
o
•rt
4-1
CO
B
•rt
E
CO
4J
B
0
CJ
.
B
o
•rt
4-1
CJ
CO
Ol
CJ
•rt
s
01
4-1
0 '
X
01
to
Ol
^4
CO
•rt
c
o
•rt
4J
CO
3
4J
CJ
3
•4-1
0)
3
4-1
Cfl
Ol
0.
B
Ol
4J
Ol
Q
o
CO
1
B
•g
CO
4-1
B
o
CJ
p
01
4J
CO
•»
^4
Ol
Ol
3
0
Si
•
1— (
o
01
4-1
CO
p
p
cu
4-1
CO
01
p «
60--!
CO S
p
cu o
CO E
3
cfl E
u ca
13 4-1
1— 1
3 CU
O 60
3 B
Cfl
e js
o u
•rt
4-1 U-l
CO O
i
a.
•rt
3
cr
Cd
Ol P 01
P O 60
3 U-l CO
4-1 p
CO CO O
P P 4J
CU O CO
a. co ^
B E CJ E
Ol Ol M CO
H to 2 4J
CO
3
4-1
U-l
O
CO
p
0)
4J
3
Cu
E
0
CJ
01
4-1
cfl
p
O
4-1
•rt
B
O
s
o
4-1
CU
p
3
4-1
CO
^4
0)
a.
B
0)
4J
U-l
0
Ol
60
CO
P
O
4-1
ca
E
•rt
CU
CO
•rt
CO
^
E
CO
4-1
O
4J
•rt
E
O
B
CU
P
3
to
CO
cu
p
OH
^
B
CO
4-1
Ol
60
CO
p
O
4J
CO
B
•rt
to
0)
8
CU
J3
i— <
3
O
CJ
p?
o
•H
pj
|3E
B
o
•rt
4J
CJ
cfl
0>
cfl
U-l
h-l
CO
o
4J
•rt
E
o
B
CU
P
3
to
CO
cu
p
OH
B
•rt
60
01
JQ
O
4J
01
p
01
5
E
O
•rt
4-1
CJ
CO
CU
p
^1
CO
3
CO
E
3
CO
CO
*j
C
CO
4-1
Ol
60
CO
p
O
4J
CO
B
•rt
Cfl
CU
TJ
3
O
3
P
O
cfl
CJ
•rt
•a
B
•rt
p
cu
si
4J
0
B
CO
O
4J
0)
3
•53
O)
p
3
to
CO
cu
p
a
*J
E
CO
4-1
E
•rt
Ol
CO
•rt
^
d reaction
E
CO
Ol '
I-l
3
4-1 .
Cfl
p
0)
a.
S
01 •
4-1 CO
4-1
TJ CJ
O) 3
Cfl T3
cfl O
Ol P.
P a
CJ 1
E S*i
•rt J3
c
to
Ol
4J
B
Ol
0)
a.
CO
p
0
4J
•rt
E
O
B
i— i
0)
Ol
1— 1
•^
•rt
3
tr
•rt
r-l
Ol
JS
H
i— I
Ol
— 4>
H~b
•^1 1
•o
•rt
3
cr
•rt
-j
tanks.
o>
60
CO
p
O
4J
CD
CU
4J
60
E
•rt
i— 1
r— 1
•rt
U-l
|^
Ol
o
^.J
o
UH I
(
CO
o
u
•rt
E
Q.
£\
CO
*s
E
60
CO
0
*Jk
&
-------
S
o
o
C
o
u
CO
S
0)
co
CA
01
4J
CO
w
%
5
0)
to
y
0.
a.
4J
o
z
CO
>,
CO
E
to
43
tO
H
01
<4-4
tQ
w
4J
•3
^3
ai
4J
to
•H
U
0
ca
ca
^2
CO
4J
C
i
i*
4J
ca
c
M
^^
00
•H
01
01
c
•H
bo
c
w
u
CO
•a
c
3
•O
0)
pi
CO
01
0
4J
>>
4J
C
£3
a.
•rt
3
cr
U
IH
O
01
ca
0
a
2
4-1
O
4J
4-1
C
4J
bl
O
a.
g
•H
CO
•H
£
3
,_4
4J
CO
O
s
^
0)
bl
o
4J
ca
a.
o>
.3
01
u O
IH
oi s
o
C <4-l
•H 0
CD
0)
CO 4J
0) CQ
U bi
3
CO O
bl -H
8.3
e «
T>
01
CO
CO
0)
u
c
C M
0
4J •
C 01
01 bl
•0 3
O) CO
a. bi
CU 01
•o a.
S
.2 2
<4-l
•o
01
c
CO
•H
c
0
•H
4J
A
^4
01
M
•H
U-l
U
1
to
c
"s
to
0) C
> 0
4J
U <4H
0) n
bi C
O 0)
CJ £&
CO
4-1 4-1
ca
bl CO
•H C
«-i O
•H
0) 4J
•£ a
•
T3
CU
4-1
U
01
cu
CO
3
ca
ca
•H
C
o
71
•a co
c to
CQ bO ^^
bl C CO
o o co a. 01
•4H -H B 3 1-1
^- 4J 01 1 I-H
•H 4J ^ a
ca c ca o a.
01 bO >> to 3
>< -H CO 43 CO
CO
bl 0) 4J
O 43 to
a. • 4-> j=
ii TO co -a «
tO CJ > J bl 01 S 43
U CU bO O 4J CO
CO "H •£• "O bl bl *H CO CU
•H ^ 4J 01 to IH x 0
01 iH i-H * tO -H
cu*OOfH cacu^>
bl 4JQOIOlbllHOI*
CO r-( bl-UcaOtOTJO)
iHOt04JCOtO<4-l CO
4jcooo) bibicaa.
aiccnyc-oojcoaibi
H o-H-3 bocotH»H4j a.
Devices to
assure that
CO
X
CO
i— i
CO
co
•H
01
iH
(0
rH
Mil
operating:
-dual pilot
CO
4-1
00
•H
iH
-dual thermo-
couple flame
detectors
-automatic
4-1
4J £
-C 00
60 -H
•H rH
rH CD
0) b
M
i— 1
4J CO
0 3
i-t C
Q* S
1
system
-temperature
interlock with
scrubber
automatically
actuates
emergency gas
-------
S
cu
IH
3
4-1 B
CO TH
g. u
S co
IH eu cu
CU B 4J TJ TJ
42 O CO O TJ
4J U rH B Q
O
TJ O 3
rH O
3 C H
O O TH
a TH cj
4J
sou
U B IH
CO 3 •u
rH UH CO
CO rH 3
CO to
o
4-J CU I-l
§44 CU •
CO 43 6
U 43 CU
Q.TH 3 4J
B TJ M CO
3 C CJ S,
CVi TH CO
tN
CO
CO
s
O O CU
0) - 4-1 -H 43
B IH B 4J4-I
O CU O rH CJ
CX o TH CO CU UH •
>, IH 44 UH CO TJ CX O CO
rHCUCUtOOOBCO Ol
C > > »H CXCOB X >
O rH CU CU >s CO TH CU 44 ,H
CO » O. 44 TH « 42 TH «
CO>OOlHT3lH4J4-lrH>
iH S3 rH CU C TH
>, rHTHTJ43CUO43>,
(U 4J •0043-H4234JC04-I
lHCUA!3cfl33cr iHCU
CUUH B B CX CT M CU TJ rH UH
42tOtOCOtOTHCJlHTJCUCO
He044BCJrHCOUHCOlHCO
CO
CX
B
ex
§5
CU TH 43
4-14-13
CO 10 IH
O rH O
TH 3 CO
rH CJ
CX b IH
3 TH O
Q CJ UH
cu
bo
CO
to co
C >s
O O 4-1
i-l TH TH
4J- 4J CJ
TH CO tO
•030.
TJ CO to
< cj cj
(U O
CO > i-l
CO »H 4-1
cu cu co
U CO
CJ
3
(0
IH CJ
cu cu
42 42
4J ^J
B
O)
CO iH
CU rH
> 0>
rH I-l
to
> Jrf
B
I*, «
4-1 4-1
01
UH Ol
tO 60
a> to
CU O
42 44
H CO
>. IH
43 CU
43
CU 43
IH 3
3 ^4
CO CJ
CO CO
eu
IH CU
CX 42
4-1
UH
0 0
4-1
a, «
3 bO
1 B
rH 4-1
TH B
-3 2
o
IH •
UH B
O
CO -H
4-1 4-1
rH CJ
3 CO
to cu
CU rH
IH
TJ
4J 0)
CO rH
42 rH
4J O
IH
01 4-1
IH B
itO O
C B
; 3
E C
« CO
B
eu
eu
43
cu
CO
CO
eu
rH
CO
>>
4-1
0)
UH
tfl
CO
Ol
H
CO
cu
to
O CO
CO 60
T> CJ
CU M
U s;
tO
rH UH
a o
01
IH CU
CO
TJ tO
B CU
CO rH
cu
TJ IH
eu
B 01
604=
TH 4-1
CO
Ol 4J
TJ CO
25
TJ .
Ol 0)
•H IH
rH tO
O rH
IH UH
4-1
B TJ
O B
cj co
IH M
0) Ol
4-1 43
4-1 43
.25 2
u
Ol CO
43
0)
TJ 43
rH 4-1
3
0 >%
S 43
e
CO
0 4=
4-J
CO TH
CU »
rH TJ
CO O)
> 4J
tO
01 IH
-C 0>
4-1 ex
o
s" ^
O i-H
IH. IH
4J CO
TH 3
T3 CO
CO S
B 01
M 43
B
CO
cu cj
4-1
O 42
B U
CU iH
h 42
3
"S S
4-1 0)
CO 4-1
M CO
cu >*
CX CO
o
rH
U O
TH IH
CO 4-J
B
B 0
CO U
60
TH
4J
CO
CU
ex
o
rH
g
U
o
B
0)
TJ
TH
Id
1
IH
Ol
>
0
CO
TH B
0
4J TH
•H 4-1
CO
eu IH
co eu
CO CX
CJ 0
B to
TH TH
42
co H
B
O
TH •
4J T3
TH CU
B eu
O 0)
CJ B
S
o cu
U 42
UH H
TJ
eu •
S B
IH 0
O O
UH IH
IH
Ol rH
a. o
IH
CU 44
43 B
O
TJ CJ
rH
3 01
0 42
S 4J
4J
B
Ol 0>
> 4=
0) 44
TJ H
oi ex o
CX 44
ex o
TH 44 CJ
3 TH
CT CO 44
CU .Jad to •
CO 3 CO
O TH CO J^
co TJ a B
rH CO
CO CU UH 44
U O
eu 3 TJ
IH 44 3 B
eo ex o to
3 rH
CO IH UH CO
CU 1 42 .* >
rH 44 CJ rH
CO TH CO CO
> S 43 >
TH IH
O3O
"47 TJ
TJ rH CU
CU CO
B IH CO
M o eu
TH rH
co en eu
cu IH IH
TJ O
CX 01
CO CO 43
TH >
o
4-1 O 4-1
B I-H
cu S cu
ex cu 01
TH N 3
3 TH
CT rH >,
cu to cu
IH 4=
•H 3
43 0) UH
H B TH
CO
•a
2 o
44
eu
bo co
to eu
i- bO
O C
4-1 tQ
"•g
5 c
4.) 60
TH
B co
o eu
li TJ
^ e
cu to
ex 4-1
B IH
3 CU
ex u
fi
cu
.2
0)
u >
o to
UH 42
CO
,U TH
o
S.-3
to
CJ B
O
CU TH
42 4-1
4J CO
N
CU TH
CO rH
CO co
CU M
IH 4-1
U 3
B 0)
TH B
IH
CU
43
3
I-i
CJ
CO
cu
H
B
B
IH
O
o
c
cu
a.
TH
3
cr
eo
-------
CO
CA
>,
4-1
Ol
M-l
CO
co
2 4g
to
H
bO
C
•H
Id
>
40 I H l-t IH
i-t J>i 1-44-1
i-d CJ CO T) CO CO
CO CU C Ol
B Id CJ
01
Id
3
CO 14-1
Id U
a. to o
B ^H
co cfl -o
CO
Ol 73 O
> C -I
CO CO U
£ (U
CO >
OQ u o
a)
a.
S
O
M
4J
C
CO O
C U
o
•H C
CO T)
O 01
a. 4J
co
01 o
> -H
•H T3
CO C
co
u
3 B
,1 CO
cu
Ol _- -
i-l > CJ 4J . _
i—t Cfl Cfl CO CO CO
<; j= oa i t
id
•H
Id cfl
O
>4d 4J
d
d 41
oi B
bo 3
O »-i
Id 4J
4-J CO
•H d
d -H
CJ
0>
r-l CO
0> d
O
Ol "H
4-1 4-1
cd eg
lJ 4J
cd co
O.X)
Ol 3
CO CO
I
CO
Id
o
4J
CO
Id
Ol
bO
JO
CO
U
D.
a.
o
d
01
e
s
id
•U
CO
d
M
•-^
bO
i-i
Id
01
0)
d
•^4
bO
d
Cd
CO
CJ
d -H
M ,J
^J
d
IV
B
0.
•H
3
cr
>4d
O
01
CO
o
O.
M
&
u 1-1
O CJ
4J
TJ
id d
O) CO
O 4J.
o. d •
0) CO
a) e 4-1
•a o. d
T> "3 1
o cr ?
I-. 41 id
a. 4-1
Ol CO
0 A d
H 4-1 -rt
01 O)
l-i 4-1
CO CO
. a. b
co s 01
CO 3 -<4d
cu d o. co co
d IH s d
•H cfl 4J O efl
d
01
Id
Id 01
01 O 01
Ud M
en .c a.
d 4J
Cfl 1H O
^ cu
o js
4-1 CO CO
T3 CO T3 S
CJ Ol O) CO O
M d ^1 O rH
r bO 3 •-. iw
•H 4J Id
01 co cfl 01 S
41 01 > O
CO
41
i-l 0)
cfl S.
> 4J
O
t ca
4-J S
3 ^O
CB Ud
CJ CJ
•H ^0
o d
4,1 Ol
3 >
CO Ol
Id
01
cfl O
m
I
0)
d id
•H O)
4J <4d
d ca
o d
U CO
CU 4J
O
4J t4d
CO O
O 01 0)
4J CO P
cfl 3
T3 CJ 4J
d a.
CO d 3
•H (J
ca
: oi
d o d
CO
o
CO 4-1 4J
Ol CO i-l
> O CO
iH E O
H"O<4d Oi-ljS4J4J«dr-l
CO
> 01
JC Ol
1-1 4J >
O TH
id e 4-1
4J TH
O C
41
CO
41 O
Id CJ
01
CO
cfl
O
IH O)
Id
>, 3
4J
01
O, 01 -H
co
CO
Cfl U-l
CO
l-t
M CO
d o
O -rt
Id -O Id
lJ 4J
CO U
bO CU
01 i-H
Id 41
• *
4J
d
a>
S
a
•H
3
CT
Cd
ca
0)
•H
4J
•H
i— 1
TH
4J
5
CJ
•H
^mt
4-1
U
Ol
1-1
O)
1
d
*D
8
3
Id
4J
CO
d
•H
1
Id
Id.
CO
to
d
1-1
1— 1
O
o
u
1
K-l
41
4J
<0
^
CO
CO
IJ "
H co
01
cj d
CO
>
Cfl
wB
I 01
i-t 4J
-H CO
CO >»
4-1 CO
-------
2. Management accountability system. Proper equipment is not enough.
It must also be operated and maintained properly. This is ultimately
a management function. The following presents some highlights of the
management accountability system in place at Institute.
0 A computerized reliability maintenance (RM) program provides for
scheduled inspection and replacement of critical equipment. The system
establishes accountability for assuring that the program is followed.
0 All storage tank temperatures, pressure, and liquid levels are
displayed in the contiol—room and simultaneously logged by a computer
system, providiag''Socumenttion)of read-outs of critical instruments.
The operation is- puil(Mll5lTyNfield checked.
0 Various management reports are required to document preventive
maintenance. All maintenance on instrumentation for the MIC unit has
just been completed as an additional safety precaution to ensure the
integrity of instrumentation. Critical instruments are checked, at a
minimum, every 12 months and critical alarms every 6 months for the
storage tanks.
0 As part of a preventive maintenance program, all major equipment is
inspected on a one to three year cycle, depending on usage. Specific
testing and inspection programs for process vessels (storage tanks)
and safety relief valves are in place.
3. Documented operating procedures. Documented operating procedures,
usually referred to as standard operating procedures (SOP), detail
specific instructions for operating personnel. They define corrective
actions in case certain instruments indicate the possibility of a
reaction which, if unattended, could cause a runaway reaction.
Standard operating procedures for the storage tanks require certain
corrective actions if the temperature or the pressure in the tanks
reach a certain level. These actions include additional refrigeration
pumping of liquid MIC to the base of the scrubber to neutralize the
MIC with caustic, and notification of plant management.
After use of process equipment and transfer lines, standard operating
idures call for isolating equipment to be cleaned with water and
nitorgenljSmrging to remove any water vapor before putting into service.
lis reduces potential for water contamination.
4-6
-------
Standard Operating Procedures call for strict administrative controls
(i.e., plant management approval) for removing the scrubber, flare, or
refrigeration system from service for maintenance, either planned or
unplanned. In a planned shutdown, no MIC is to be stored. Standard
operating procedures also require that, if the flare or scrubber
malfunction for more than 30 minutes, MIC production must cease and
efforts must be taken to reduce MIC inventory.
4. Training. In the first stage of training, operators must take a two-
week (80 hour) training program in hazards of contamination, ways to clean
equipment and ways to return equipment to service. They are tested at the
completion of the program. ,
The next stage of training is a maximum of seven days of on-the-job training'
under the supervision of an experienced operator working exclusively/>wit>xX^
the trainee. Additional on-the-job training follows, the length ofion^time
depending on the system involved. Trainees are given several exams (luring
this period, culminating in a system walk-through exam with the trainee
supervisor, system supervisor, and training coordinator. Trainees who do
not qualify may retrain for half the original period of time. If they still
fail to qualify, they are terminated or transferred from the MIC unit.
CONCLUSION;
Equipment maintenance procedures, instrument maintenance procedures, training,
documented operating procedures, and management accountability systems are in
place. Success of these procedures and systems depends on close company management
overview to assure that these procedures are followed and retraining is given
as necessary.
Various regulatory agencies will conduct periodic inspections to assure that
critical equipment is maintained and that all procedures and systems are followed.
All alarms are located in the control room. The added alarm located in the shift
administrator's office provides an additional alert system.
4-7
-------
Chapter Five
Detection and Notification of Emissions, Spill/ an/Releases
PURPOSE:
To evaluate the instrumentation and systems in place at the Institute, West Virginia
plant to detect leaks of MIC or the presence of a condition which might result in a
leak, and for providing notification of such a leak or condition.
4.o ^"re^idQ
DISCUSSION;
1. Detection of potential runaway reactions or malfunctions of control
equipment.
a. Temperature and pressure monitor. The process control and monitoring
system on the MIC production and storage units serves as the "first line
of defense" for detection of conditions, such as contamination, that
could result in possible leakage, spillage, or emission of MIC. A
recent modification in the process monitoring system now allows the rate
of temperature change to be tracked. In addition, continuous liquid
level sensors, each with a preset alarm level, detect unexplained
changes in volume in each of the tanks and prevent overfilling.
A series of redundant temperature and pressure sensors in the system
give both visual and computer readouts on conditions within the
storage and handling system. If Standard Operating Procedures for
the maintenance, calibration, and use of these monitors are followed
they will successfully indicate the potential for a major release of
MIC. Further, they should give sufficient warning to allow mitigating
action to be taken and to allow company notification of Federal, State
and local authorities.
b. Flare Monitoring. Dual thermocouple flame detectors on the MIC flare
serve two functions. They are wired into an automatic pilot relight
system to assure that if the pilot or flame goes out, immediate steps
are taken to relight it. They are also wired into alarm systems to
notify the MIC operators and the control room that the flare is out.
c. Scrubber Monitoring. The continued operation of the scrubber is
monitored by instrumentation that indicates the level, the flow,
pressure, and the temperature of caustic in the scrubber and whether
or not the pumps are operating. In the event of a failure in caustic
flow, which would be detected by the caustic flow monitor, MIC will
be directed from the scrubber to the flare. If the packed column
floods and channeling of vapor occurs, the temperature as detected
by the thermocouples, will not rise as much as expectedf Therefore,
this is considered another indication of scrubber malfunction or
failure.
5-1
-------
2. Det-r .on of MIC releases.
a. Operator Attention. During normal operating conditions, respirators
or gas-tight goggles are not required because no MIC is released to
the air. The operators' eyes and olfactory systems are, therefore,
sensitive indicators of even a trace of MIC release to the air.
If standard operating procedures are followed, then, in the event that
MIC is detected, the area is cleared and the foreman is notified. He
will notify plant management and initiate mitigating action.
b. Air Monitoring System. After evaluating several air monitors that
could serve as back-up release detectors of low levels of MIC, Union
Carbide selected a flame ionization detector-process gas chromatograph
(GC) capable of analyzing up to 16 sampling points in a sequence. Each
analysis of MIC in air at a 0.1 parts per million (ppm) detection limit
takes one and one-half minutes. The full sequence at 16 points therefore
takes approximately 24 minutes.
Although the system is in the first stages of utilization with many
aspects not yet finalized, discussions with Union Carbide have revealed
that the system might be modified to add more sample points, or that
other detector systems might be used, if necessary.
The system does have shortcomings, the most important of which are
that Union Carbide has limited operating experience using it for MIC and
there is no assurance, at present, that the system can achieve the claimed
detection level of 0.1 ppm. However, even with these limitations, it
it should be viewed as an attempt by Union Carbide to advance the state-of-
the-art in leak detection.
c. Air Dispersion Modeling System. Union Carbide has recently installed
a site specific air plume dispersion modeling system known as SAFER®.
The system is a minicomputer-based mathematical model with real time
data input of wind direction, wind speed, and other weather data from
the plant's meteorological tower. This data is used by the computer
to predict concentration and location of a major air release.
An operator enters such data as the location of the leak, size of the leak,
and other parameters. The computer calculates the direction of the plume,
rate of travel of the plume, and displays the plume location on a map of
travel of the plume, and displays the plume location on a map of the area
around the site by using a color video display. Critical impacted areas
in the community (schools, hospitals, etc.) are highlighted. Notification
and response actions can therefore be arranged in advance of the plume
impact with
5-2
-------
Other scenarios of major releases, including worst case scenarios, can be
simulated and appropriate actions arranged in advance with local
authorities.
Union Carbide has not yet finalized standard operating procedures for use
of the system during an emergency, nor, aside from some small smoke bomb
observations, has it performed calibration of the air dispersion model,
but these will be completed by May. Based on EPA experience with air
dispersion models, several other actions should be taken as soon as
practical including: model verification to assure that predictions are
within the desired accuracy range; training of response personnel to
assure that they understand the limits of the model's prediction;
tvelopment^pf an integrated contingency plan to assure that predictions
triggergj£ju?oper response actions; and a community information effort to
Serresponse actions; and a community information effort to assure that
the public understands the system. Even before these actions are completed
the SAFER® System will still be useful in determining worst case scenarios
that can be used to aid in response decisions.
3. Notification.
a. Notification of Local Authorities. If standard operating procedures
are followed, the foreman would be informed of any abnormal conditions
in the process and storage areas, and an in-plant toxic gas alarm
would be sounded if there is a possibility, no matter how uncertain,
that a leak has occurred. At that alarm, the shift administrator is
informed, and decides whether the incident is serious enough to
warrant notification of off-site authorities. In-^addition, Union
Carbide has agreed to initiate the following drf*sjke notification
procedures: ^—~-
0 Within a period of two (2) hours, when and if all available
refrigeration to MIC storage tanks is inoperable (but only if
MIC is in storage) * ~~
0 Immediately, when and if the temperature of MIC in a storage tank
reaches 5°C and temperature cannot be reduced by 1°C within a
period of fifteen (15) minutes •
0 Anytime the scrubber and flare are concurrently out of service for
more than thirty (30) minutes.
Union Carbide plans to continue to meet with state, county, and Kanawha
Valley Emergency Planning personnel to revise and update applicable
contingency plans.
5-3
-------
b. Notification of the Federal Government.
reportable quantity of
a hazardous substanceQ.eaves the plant site^Union Carbide must notify
either the National Response Center or 'the EPA Regional Office.
CONCLUSION;
x^~—\
Instrumentation, detailed standard^opeating^prrocedures, and well-trained operators
and supervisors make it unlikely that any but a minute release of MIC could occur
without being detected. Improvements made in notification procedures will save
valuable time and help eliminate confusion in the event of an MIC release.
5-4
-------
Chapter Six
Contingency Plans
PURPOSE:
To determine if contingency plans that exist or axe being developed provide for
an adequate response by local, county,Cap.ate, an£federal communities to an MIC
release in Institute, West Virginia. ^~
DISCUSSION;
EPA involvement in reviewing existing contingency plans was initiated through
agency chairmanship of the Regional Response Team (RRT). Attachment two
presents the results of the review.
To implement the recommendations of the RRT, EPA met with officials of the Federal
Emergency Management Agency, the West Virginia Office of Emergency Services, and
the West Virginia Department of Natural Resources. Participants agreed that:
0 The County emergency management agency should receive first notice of a
release and should assume initial command of a response.
T •f
The responsible party should notify, in^additon^to the Federal Government,
the County emergency management agency as soon as it knows of the release
into the environment of a hazardous substance in an amount equal to or
greater than the reportable quantity defined by federal regulation.
0 These two recommendations are being implemented expeditiously.
These issues and others have been discussed and agreed to by Union Carbide
officials during recent task force meetings.
A number of events are now taking place which will serve to improve the
Kanawha Valley's preparedness in the event of a hazardous substance incident.
Some of these are mentioned in the preceding chapter. Other significant events
include: ,
0 Meetings betweei(_azate and county officials to resolve problems in
the areas of notification and responsibility. Existing State code requirements
are being evaluated to determine if modifications are necessary to further
clarify these issues.
0 Incorporation of additions to the West Virginia Hazardous Materials
Emergency Response Plan to address RRT comments, particularly in the areas
of county responsibility/authority andCg^zate/EPA coordination regarding
Superfund.
6-1
-------
Agreement between the West Virginia Office of Emergency Services and
the Kanawha Valley Emergency Planning Council that notification will be
to the county, which will be responsible for public messages and evacuation
decisions. Discussion of appropriate stages of notificat(lo/by industrial
managers is planned for future meetings. «., &
Revision of the Kanawha.Valley Emergency Planning Council's contingency
plan to insure that it is compatible with local, county, andfarate
requirements.
Installation by Union Carbide of a more effective warning siren to alert
surrounding communities that an announcement is being made by county
officials on the emergency broadcast system.
CONCLUSION:
As the RRT has indicated, contingency planning is necessarily a dynamic process
requiring continuous revision and updating. The RRT comments -are being actively
considered, and significant progress is being made in preparing changes to all
relevant contingency plans. The major immediate issues needing resolution have
been adequately addressed in verbal agreements between the State, County and
industry. Other issues raised by the RRT are being addressed in a draft update
to the West Virginia Hazardous Materials Emergency Response Plan currently under
review by the State. Changes to local government and industry contingency plans
are scheduled to follow.
6-2
-------
Chapter Seven
Handling and Storage for Transportation
PURPOSE:
To determine if steps are being taken to reduce the danger of/a/MlC release
during transport. V^-"
DISCUSSION;
Prior to the Bhopal incident, Union Carbide shipped MIC to customers in rail
tank cars, portable tanks, and drums. The MIC was transferred from one
container to another via stainless steel hoses. All operators working with
MIC were required to wear self-contained_b*e*thing apparatus and to avoid all
contact with MIC vapor and liquid^^Jpecsif i^xLoading and handling instructions
are detailed in their SOP.
Since the Bhopal incident, Union Carbide says it has no intentions of shipping
MIC in any form at the present time, but that it may do so in the future if
required to contractually. Should that be the case, the company says it will
ship MIC only in portable tanks which can be hauled on a flatbed truck and that
it will no longer ship MIC in rail tank cars nor store MIC in tanks at trans-
portation/distribution points. Thus a major storage area which held much of
the plant inventory of MIC before the Bhopal incident is now eliminated. As
a safety procedure, the»tanks are sampled before filling and kept on site for
24 hours to assure thatlno adverse reactions are occuring. ___ p
Since Bhopal, the U.S. Department of Transportation (DOT) has issued stricter
regulations for the shipment of compounds which exhibit high acute air
toxicity such as MIC.
CONCLUSION:
The U.S. DOT will be further reviewing the shipping and handling procedures
submitted to the task force by Union Carbide Corporation.
7-1
-------
Documents Reviewed by the Task Gffottp"
1. Methyl Isocyanate: Institute Plant Start-up Review (UCC 1)
2. MIC Safety Valve Design Considerations, March 18, 1985 (UCC 2)
3. Reaction and Refining System, Safety Valves (UCC 3)
4. MIC Vent Scrubber Design Considerations, March 18, 1985 (UCC 4)
5. MIC II Emergency Vent Scrubber Calculations, March 19, 1985 (UCC 5)
6. Flare for MIC II Unit (UCC 6)
7. Flare (UCC 7)
8. MIC Storage Tank Process Dynamics (UCC 8)
9. MIC Storage Tank Instrumentation (UCC 9)
10. Prevention of Backflow from Emergency Vent Scrubber to Emergency Vent
Header (UCC 10)
11. Back-up Utilities (UCC 11)
12. MIC Air Monitor (UCC 12)
13. Additional Questions (UCC 13)
14. Governmental Agency Questions, March 14, 1985 (UCC 14)
15. Results of Methyl Isocyanate Scrubber Performance Tests Performed on
March 16, 1978 and April 12, 1978 at the Woodbine Carbamoylation
Unit (UCC 15)
16. Bhopal Methyl Isocyanate Incident Investigation Team Report, March 1985
(UCC 16)
17. In-Plant Methyl Isocyanate Network (UCC 17)
18. Letter from H.J. Karawan to Greene Jones, March 25, 1985
19. Letter from H. J. Karawan to Greene Jones, March 26, 1985
8-1
-------
20. Testing and Inspection of Instruments, SOP
21. Testing and Inspection of Process Vessels, SOP
22. Pressure Testing of Process Equipment, SOP
23. Safety Relief Device Procedure, SOP
24. MIC Operator Training, SOP
25. RM Program, SOP
26. Administrative Controls, SOP
27. Local Toxic Gas Alarm Procedure, SOP
28. MIC No. 2 Unit Safety, Reactive Chemical Data, SOP
29. MKymstrubution) Procedures, September, 1984
30. Letter from Robert Oldford, President, U.C. Agricultural Products, Co.,
Inc. to Greene Jones, EPA dated April 10, 1985
31. MIC Storage at Methomyl/LARVIN® Unit
-------
ATTACHMENT ONE
-------
-------
BHOPAL METHYL ISOCYANATE INCIDENT
INVESTIGATION TEAM
REPORT
MARCH, 1985
UNION CARBIDE CORPORATION
DANBURY, CONNECTICUT
-------
TABLE OF CONTENTS
Section Page
Summary i
1.0 Introduction 1
2.0 Background 2
3.0 The MIC Process 4
4.0 The Event 11
5.0 Chemistry of the Event .14
6.0 A Hypothesis for the Event 23
-------
SUMMARY
The Bhopal Methyl Isocyanate Incident Investigation Team included seven
engineers and scientists who were charged with assisting in the safe disposal of the
remaining methyl isocyanate (MIC) at the Bhopal Plant, investigating the incident
and determining its probable cause. The team spent 24 days in India and continued
its work for a period of more than two months thereafter.
The team was able to inspect many relevant written documents in India, but
was restricted by the Indian Central Bureau of Investigation from interviewing the
Union Carbide India Limited (UCIL) employees directly involved. Interviews were
held with the Works Manager and the MIC Production Manager and informal
discussions were held with a few witnesses to the event.
The team was able to obtain core samples of the residue in Tank 610, the MIC
storage tank involved in the incident, as well as samples of the methyl isocyanate
(MIC) in some other tanks and associated lines. These samples were analyzed by
UCIL and Union Carbide Corporation (UCC) research groups. Although the team
was not permitted to conduct a detailed examination of Tank 610 and associated
piping, through extensive experimentation, the UCC research groups succeeded in
replicating reactions which yielded residues with the same major components found
in residue from Tank 610. The team examined many possibilities, including those
deemed remote or unlikely, and arrived at a hypothesis for the event in which it
has a high degree of confidence.
Early on December 3, 1984, the safety valve on Tank 610 opened as a result of
a chemical reaction in the tank. At the time, the tank contained approximately
90,000 pounds of stored MIC. The team believes that the safety valve remained
open for approximately two hours before it reseated and during that period in
excess of 50,000 pounds of MIC in vapor and liquid form was discharged through
the safety valve.
This incident was the result of a unique combination of unusual events. The
team's hypothesis is that the reaction in the tank occurred when a substantial
amount of water was introduced into Tank 610. Through an extensive amount of
experimentation, it appears that the substances found in the Tank 610 residue were
produced by the reaction of MIC with large amounts of water, higher than normal
amounts of chloroform and an iron catalyst at a high reaction temperature. It was
concluded that the reaction which occurred in Tank 610 had to involve all of these
factors. The exothermic reaction between the water and the MIC raised the
temperature in the tank. Since the MIC in the storage tank was at ambient
temperature, the rates of reaction and temperature rise were rapid. A concurrent
exothermic trimerization of MIC was catalyzed by iron resulting from corrosion of
the tank walls due to the high temperatures produced by the reacting mixture. The
laboratory work indicates that 1,000 to 2,000 pounds (120 to 240 gallons) of water
would have been required to account for the chemistry of the residue. The source
of the water is unknown, but the report discusses possible means by which water
could have entered the tank.
• i -
-------
1.0 INTRODUCTION
This report provides the knowledge and understanding that the investigation
team has at this time of the Bhopal Methyl Isocyanate (MIC) incident that occurred
on December 3. 1984. The report is organized into five sections following this
introduction.
Section 2.0, "Background," presents how the investigation team was organized,
the investigative process it used, the types of information it gathered and the
restrictions it encountered in doing its work.
Section 3.0, "The MIC Process," provides the reader with an overview of the
process and details of items of equipment relevant to the incident
Section 4.0. "The Event," describes the incident, tank residue sampling and
quantity, and disposal of the MIC remaining after the incident.
Section 5.0. "Chemistry of the Event," presents the composition of the tank
residue, the chemistry by which the residue components could be formed, the
experiments used to replicate the chemistry and the probable sources of the
components which caused the chemical reactions.
Section 6.0, "A Hypothesis for the Event," describes the team's hypothesis for
the event
-------
2.0 BACKGROUND
The Bhopal Methyl Isocyanate (MIC) Incident Investigation Team convened in
India at 10:30 AM on Thursday, December 6. 1984, for its first meeting. The team
included seven engineering and scientific specialists who worked directly on the
task. In addition, the team was assisted by many other UCIL and UCC technical
personnel. The team had access to the UCIL Bhopal plant starting at 3:30 PM on
December 7, 1984. The team worked in India 24 days and then continued its work
in the United States from January 2, 1985, until the completion of this report.
The objectives of the team were to:
— Assist in the safe disposal of the remaining MIC.
— Investigate the incident and determine its probable cause.
The first objective required 15 days and was completed on December 22, 1984. The
team then concentrated on investigating the incident which took more than two
months thereafter.
The team was organized into two subteams:
(1) The Process Information Subteam had accountability for tracking down and
reviewing written records, inspecting equipment and gathering information from
interviews. The team was permitted to inspect many of the relevant documents.
To date, the team has not been permitted to interview the UCIL employees
directly involved in the incident. This restriction has been established by the
Government of India Central Bureau of Investigation which is also investigating
the incident. Consequently, the description of the event in this report is constructed
from interviews with the Works Manager and the MIC Production Manager. They
arrived at the plant after the event and developed their information from informal
discussions with personnel directly involved. Various team members also had
informal discussions with a few first-hand witnesses as they worked together
during the processing of the MIC remaining in the plant after the event.
Information gathered from these informal discussions is also included in the
description of the event.
(2) The Chemical/Physical Information Subteam had the task of defining the
physical and chemical nature of the event. This was done by analyzing samples of
the process streams and residues to determine their composition. This subteam
inspected relevant plant quality assurance laboratory records and while in India
worked with the UCIL research group. The team was permitted to obtain core
samples of the residue in Tank 610. the tank involved in the incident. Since
returning to the United States, they have been working with UCC research groups
to replicate the reaction/event to give a residue with the same major components as
that obtained from Tank €10. The team also sampled and analyzed the MIC in
Tanks 611 and 619. in the MIC refining still product line, in the transfer line to the
Derivatives Unit, in the charge tank at the Derivatives Unit and in drums.
However, various restrictions by Indian government authorities and courts have -
resulted in the team not being able to:
— Open and inspect Tank 610 and its piping.
— Do further sampling and analysis on the contents of the tank, other
equipment and piping.
-------
The investigation methodology used by the team was a carefully disciplined,
"no-stones-unturned." approach. During the initial phase, the team gathered
available information that might be relevant to the incident. Because of the need to
identify all factors, every effort was made to avoid premature dismissal of any
possible explanation or contributor. This general approach is illustrated in the
following diagram:
Problem Solving Approach
Start
•»• Gather Information •*-
I
Evaluate Information
List Possible Causes or Scenarios
(Brainstorming)
Evaluate Each Cause or Scenario for Elimination
Or Acceptance
(Based on Available Information)
Develop Information
Needs For Further *
Evaluation
Reach Conclusion
A key element of problem solving was to identify many possibilities, including
those deemed remote or unlikely. Each possibility or scenario was examined in the
light of information available to the team. Some of the possibilities were ruled out
at that point, but in other cases additional information was required. The required.
information was defined and appropriate assignments were made to obtain it. This
led to an iterative process of examining, gathering information and re-examining
that was repealed many times. Consequently, each possibility or scenario was
rigorously studied, and eliminated only after exhaustive checking against a slowly
growing numlier of accepted facts. In addition, the team periodically stopped its
deductive process to explore any new possibilities that may have evolved.
-------
3.0 THE MIC PROCESS
This section contains an overview of the MIC process and a description of four
items of equipment associated with this event: the MIC refining still (MRS), the
MIC storage system, the vent gas scrubber (VGS) and the flare tower.
3.1 Overview
This overview contains three parts: (1) the two basic chemical equations used in
producing MIC, (2) a simplified block diagram of the MIC process, and (3) a word
description of the MIC process.
Chemical equations for MIC production are given below:
(1) COCI2
Phosgene
CHNH,
CHNHCOCI
3, 3
Monomethylamine Methylcarbamoyl
(MM A) Chloride
(MCC)
HC1
Hydrogen
Chloride
Heat
(2)
CH,NHCOCI
fMCC)
CH,NCO
Methyl Isocyanate
(MIC)
Simplified block diagram is shown below:
MIC PROCESS
HC1
•TAILS'
PHOS
MONOME1
GENE
rHYLAMINE
PHOSGENE
STRIPPING
I
REACTION
SYSTEM
1
CHLOROFORM
•TAtt ?BHBB»I
'
1 1
1
HYDROGEN
CHLORIDE
I
AcetntHTC
r
T
MIC
DeSTRLK
VR&/FI
t
UNT1
VENT
csuoi
MIC
mow
ARE
r
5
t
MIC
REFINING
PRODUCT
MIC
STORAGE
PRODUCT
MIC
DERIVATIVES
Description of the MIC process follows:
The raw materials used to make MIC are monomethylamine (MMA) and
phosgene. Phosgene is produced on-site by reacting chlorine and carbon monoxide
(CO). CO is produced by an adjacent production facility within the plant. The MMA
and chlorine are brought in by tank truck from other parts of India, stored in
tanks, and used as needed. Chloroform is used as a solvent throughout the process.
-------
A vapor phase reaction system converts the phosgene and MM A to
methylcarbamoyl chloride (MCC) and hydrogen chloride (HCI). An additional
amount of phosgene is used to assure conversion of all the MMA fed.
The reaction products are then quenched with chloroform and fed to the
phosgene stripping still (PSS) where the unreacted phosgene is removed and
recycled. The tails or bottoms from the PSS are then fed to the pyrolyzer to -
achieve separation of the MIC from the HCi.
The MIC refining still is fed from the pyrolyzer condenser. This feed stream
contains mostly MIC, MCC, chloroform and small amounts of residues. MIC is
separated from the chloroform in the upper part of the still. MCC, chloroform,
residues and some MIC go out the bottom of the still and are recycled back to the
process. This distillation operation is discussed in more detail in Section 3.2. The
refined MIC is taken from the overhead of the still to the mounded, type 304
stainless steel storage tanks. MIC is transferred from these tanks to the MIC
derivatives units as needed.
The materials of construction throughout the process are chosen to resist
corrosion. These include materials such as Hastelloy •, nickel, Inconel •, stainless
steel, glass, Teflon •, Karbate • and Haveg •.
3.2 Equipment Description
This section details four items of equipment: the MRS, MIC storage system,
VGS and flare tower.
MIC Refining Still (MRS)
The MRS is a 45-tray column used in the final step of the production of MIC.
The purpose of the MRS is to purify MIC by separating it from chloroform, MCC
and residues. The key operating conditions that control the purity of the MIC are
temperature, pressure, reflux ratio and heat input. Instrumentation is provided to
properly control the column operating conditions.
Maintaining the temperature and pressure is critical to the quality of the
material produced. An increase in temperature above normal operating limits can
lead to an increase in impurities beyond specification limits.
A high reflux ratio of approximately 20 to 1 is required for this separation.
Low reflux ratio and high product rate can result in impure material being
produced.
Excess heat input causes an overload or flooding condition in the still which
allows impurities from the lower aections of the column to be driven overhead and
possibly enter the storage tanks.
MIC Storage System
Refined MIC can be stored in two of three horizontal, mounded, 15.000 gallon
tanks. The third tank is used for emergency storage of MIC and for temporarily
holding off-specification MIC prior to reprocessing. The contents of tanks are
circulated through heat exchangers cooled by a 30-ton refrigeration system to
maintain the MIC at a temperature of about 0°C. Refined MIC .is then transferred
to Derivatives Units as needed.
-------
MIC that does not meet specification can be returned to one of two locations in
the system for reprocessing. If necessary, the MIC can be transferred to the VGS
and destroyed.
Instrumentation of the MIC storage tanks includes the following:
— A temperature indicator/alarm that is activated by high temperature.
— A pressure indicator/controller which regulates the pressure in the tank
by adding nitrogen or venting vapors to the VGS and/or the flare.
— A liquid level indicator/alarm with high and low level alarm set points.
All the above instruments have indicators at the tank and in the control room.
Alarms are audible and have indicating lights in the control room when activated.
The tanks were designed as storage vessels with the following specifications:
— Nominal capacity - 15.000 gallons
— Diameter • 8 feet
- Length - 40 feet
— Material of construction • type 304 stainless steel
— Design pressure • full vacuum to 40 psig at 121°C
— Hydrostatic test pressure • 60 psig
— Cathodic protection to prevent external corrosion
— Mounding of the tanks for puncture protection from outside sources,
protection from external fire and as insulation
— Construction in accordance with the ASME code for lethal service
vessels
— Additional bolting strength for leak protection provided by using 300
psig rated flanges on process lines and equipment Screwed piping
connections were minimized.
A diagram of the MIC storage tank is shown on the following page.
The MIC storage system includes provisions, procedures and programs for
(1) the prevention of contamination, (2) early detection when contamination occurs,
and (3) the handling of contaminated material. The provisions, procedures and
programs are as follows:
(1) Prevention of Contamination
— Storage tanks and associated process lines are dedicated to MIC service
and no other materials are permitted in this equipment.
— The material of construction of the tanks and related process piping is
type 304 stainless steel, which resists corrosion in this service.
— Each tank has a rupture disc installed to provide a positive seal between
the safety valve and the tank. This prevents any materials from backing
into the tank from the relief valve vent header. A pressure gauge
between the rupture disc and safety valve is used to determine that the
disc is intact.
— Nitrogen UM-d for tank instruments, pressure control and purging is dry,
high-purity nitrogen. The nitrogen supply is protected with a bank of 450
-------
111
8
OC
(D
CO
O
flC
<
zee
I I
oc
UJ
o
oc
8
oc'oT
ii
go
QQ
UJUJ
croc
UJUJ
OCQC
o-a
I I
_o
0.0.
-------
— A storage tank temperature alarm is activated when the temperature is
higher than the set point
(3) Handling Contaminated Material
— A refrigeration system is provided to maintain the stored material at low
temperatures which retard reaction rates and allows time for
reprocessing or destruction in the case of contamination.
— A reject line is provided so that contaminated material can either be
returned to the system at one of two locations for reprocessing, or be
destroyed in the VGS.
— An emergency tank is provided to receive off-specification or
contaminated material for recycle or disposal, to receive material from
faulty equipment, and to provide additional vapor space and cooling
capacity in the event of a reaction.
— Versatile arrangements of piping and valves, as well as spare pumps, are
provided to transfer contaminated material.
— A safety relief system protects each tank by relieving pressure above 40
psig to the relief valve vent header.
Vent Gas Scrubber (VGS)
The vent gas scrubber (VGS) is a 5'6" diameter packed tower in which all gases
entering are contacted with circulating caustic soda solution. The MIC facility has
two vent headers which go into the VGS. These are the process vent header (PVH)
through which MIC system vents are routed and the relief valve vent header
(RWH) which collects discharges from the MIC facility safety valves. The PVH
and the RWH are connected to the VGS and the flare line with an option to be
routed either way. The vent from the VGS can either go to the atmospheric vent or
to the flare line.
The VGS system has several functions. First, it handles process vents from the
PVH. Second. H has the ability to destroy contaminated MIC in a controlled
manner in liquid or vapor form. The VGS is also one element of an integrated
system for handling contaminated MIC. Contaminated material can be sent to the
MIC process, VGS, flare or moved to the spare storage tank to permit additional
expansion and coding capacity or reprocessing.
Instrumentation for the VGS includes:
— Caustic flow indicator and alarm in the control room. This system will
automatically start the spare circulating pump in the case of low flow.
— Remote and local switches to manually start the caustic circulating
pumps.
— A pressure indicator in the control room.
— A high pressure alarm audible in the control room.
— Local temperature indicator at the VGS base.
— Local liquid level indicator at the VGS base.
-------
Operating procedures for the VGS include:
— Log and continually maintain the circulating flow of caustic.
— Analyze the caustic strength daily.
— Maintain the caustic strength by the addition of 50 percent caustic from
the caustic storage tanks.
— Monitor the VGS accumulator temperature for an indication of a reaction
in the VGS.
— Log the VGS pressure every 2 hours and verify correct operation.
Flare Tower
The primary purpose of the flare tower is to burn vent gases from the carbon
monoxide unit and the MMA vaporizer safety valve. The flare also burns normal
vent gases from the MIC storage tanks, the MRS and the VGS. Vents from the
MIC storage tanks can be routed to the VGS or directly to the flare. The flare
tower is equipped with a flame-front generator for ease of lighting in the event the
flame goes out. A wind shield is in place to ensure the pilot flame is protected from
high winds. At the base a seal tank is installed to prevent backfiring in the vent
lines.
10
-------
4.0 THE EVENT
4.1 Description
The following description of the event begins on Sunday, December 2, 1984, at
10:20 PM near the end of the second shift. The description extends through the
disposal of MIC inventory at Bhopal on December 22, 1984. It is developed
primarily from interviews with the Works Manager and the MIC Production
Manager and informal discussions with others, as indicated. Since both the Works
Manager and the MIC Production Manager arrived at the plant after the release
had stopped, their accounts of the event are based upon their discussions with
operating personnel on duty during the event.
Prior to the event, MIC Storage Tank 610 was reported to contain 41 metric
tons as determined from plant inventory records. This amount equates to 90,400
pounds or 11,290 gallons. The liquid level in Tank 610 was near 70 percent of
capacity, which is below the maximum operating level. This level results in a
headspace of 2 feet. 9 inches in the eight-foot diameter tank. It was reported that
all valves to and from the tank were closed except for the relief valve line
containing the rupture disc and safety valve. The pressure in Tank 610 was later
reported by the operator via the Works Manager to be 2 psig on Sunday, December
2, at 10:20 PM. Earlier readings had been recorded to be 2 psig since the second
shift on December 1.
Shift change took place at 10:45 PM. At 11:00 PM the control room operator
noticed the pressure in Tank 610 was at 10 psig. This pressure was not thought to
be unusual because the tank was normally operated at a pressure between 2 and 25
psig. It is not known whether the 2 psig pressure reading observed 40 minutes
earlier by the operator on the previous shift had been communicated to the new
-operator. Corresponding tank temperatures were not available since no tank
temperatures were logged.
At 11:00 PM the field operator reported a MIC leak in the structure near the
VGS and process filters, but the source of the MIC was not discovered by the
operating personnel.
At 12:15 AM on Monday, December 3, the field operator reported a MIC
release in the MIC process area. The control room operator looked at the tank
pressure again. The reading was 30 psig and rapidly rising. Within moments the
pressure was beyond 55 psig (top of scale). He called his supervisor and ran outside
to the tank. He heard rumbling sounds from Tank 610, a screeching noise from the
safety valve, and felt heat radiating. As he ran back to the control room, he heard
the cracking of the concrete over the tank. As soon as he returned to the control
room, he turned the switch to activate the VGS. The VGS had been removed from
an operating mode to a standby mode on October 23, 1984, after the MIC unit was
shut down with a total MIC inventory of 183,000 pounds in Tanks 610 and 611. The
return to an operating mode was dependent upon the operator being alerted to a
problem and taking prompt action to activate the circulating pump. The flow meter
did not indicate that caustic circulation had been started. The operator did not go
( into the unit to check the pump and verify a flow. Prior to the incident, the flare |
I had been removed from service for maintenance work and was not operating at the 1
L. time of the incident.
At 12:20 AM. the MIC Production Supervisor notified the Plant Superintendent
of the release. The Plant Superintendent, who was in the formulations area,
arrived in the MIC Unit around 12:25 AM and found much MIC in the atmosphere.
11
-------
The timing of the following events are the best recollections of various people
during a very active period. At 12:45 AM, the Supervisors Log records that
Derivatives Unit operations were suspended because of the high concentration of
MIC in the area. About 1:00 AM. a Derivatives Unit operator turned on the Toxic
Gas Alarm. Also at about this time, the Plant Superintendent and the MIC
operator verified that MIC from Tank 610 was being emitted from the VGS stack
to the atmosphere. They turned on the fixed firewater monitors and directed them
to the stack and the MIC process area to knock down the MIC vapor as much as
possible. In a test subsequent to the event with two monitors in service, water
reached beyond the top of the stack. Water streams were also directed on the MIC
tank mound and on the relief valve line to the VGS for cooling. Steam came from
the cracks in the concrete indicating the MIC tank was hot.
Sometime between 1:30 AM and 2:30 AM, the safety valve reseated, indicating
a tank pressure below the 40 psig safety valve setting, and the emission of MIC
stopped. The MIC tank and associated piping remained intact throughout the
event. This is confirmed by the sub-atmospheric pressure that developed in the
tank prior to noon on December 3 as the tank cooled. Nitrogen was later used to
raise the pressure in the tank to 7 psig.
The MIC Production Manager reported that Tank 610 was hot to the touch
(45-60pC) around 5:30 AM on December 3. The MIC Production Superintendent
reported that the thermometer on the VGS caustic accumulator read 60°C about
6:00 AM on December 3, indicating that a MIC reaction had taken place. This
would indicate that caustic circulation had occurred during the event. The caustic
concentration at the start of the event was not known since no analysis had been
made since October 23. Subsequent to the event, the caustic circulating pump
remote start capability was tested and found to be working properly. According to
verbal reports, the flow indicator was cleaned after the event and thereafter
functioned properly.
4.2 Tank 610 Residue
On December 20, six core samples of solids in Tank 610 were taken through the
1 1/2-inch thermowell nozzle using a section of one-inch stainless steel pipe. The
pipe was 14 feet, 8 inches long and sharpened at one end to allow it to be driven
through the solids to the bottom of the tank. The core of solids collected inside the
pipe was extracted using a close-fitting pusher rod. From measurements taken
during the core sampling operation, Tank 610 appears to have about ten inches of
solids lying in the bottom, with some evidence of a crust four inches above the
solids with a void in between. A uniform 10-inch bed of solids would represent
10,000 pounds of material at the measured bulk density of 1.5 grams per cubic
centimeter.
It is quite possible that either more or less solids are in the tank. For instance,
more solids may be in the tank if what appears to be a crust with a 4-inch void
above the 10-inch bed of solids were really a 14-inch bed of solids throughout most
of the tank. In this case there would be 21,000 pounds of solids in the tank. In
addition, it is reasonable to expect crystals attached to aD of the interior surfaces
of the tank since the thermowell removed from the tank had a thin coating of
crystals.
12
-------
Alternatively, the tank may contain less than 10,000 pounds of solids. The core
sample was taken near the tank sump. Even prior to the event, this end of the tank
was slightly lower to achieve complete draining. From the appearance of the crack
in the concrete above the tank, the far end of the tank may have been raised even
more as a result of the event. In this case, the solids may be at a maximum depth
where the core sample was taken with less solids at the far end.
4.3 MIC Inventory Disposal
The disposal of Bhopal MIC inventory remaining after the incident was
completed without difficulty on December 22. The MIC was processed in the
normal fashion to carbaryl insecticide after a safety review of the facilities and
operating procedures. The Bhopal Operating Staff, the Indian Government Officials
and the UCC Team participated in the safety review.
The 45,000 pounds of MIC in Tank 611 (remaining after use of 47,600 pounds
for derivatives production since November 24, 1984) and the 3.300 pounds in drums
met product specifications. The 2,400 pounds in Tank 619 was out of specification,
but this is not unexpected since one purpose of this tank is to collect off-
specification material prior to reprocessing or disposal. Under the circumstances
prevailing at the plant at the time, conversion of the off-specification MIC to
carbaryl was judged to be the safest approach. The safety of this approach was
verified by a laboratory test prior to conversion in the plant.
13
-------
5.0. CHEMISTRY OF THE EVENT
I
5.1 Residue Samples from Tank 610
Two of the six core samples obtained on December 20, 1984. as described in
Section 4.2, were given to the team for analysis. These are the only samples of
Tank 610 contents the authorities have permitted the team to examine, and the
compositions presented are based on these samples. The plant retained one sample,
and three were taken by Indian Government authorities.
5.2 Composition of the Samples
The two core samples were divided into three portions each • top, middle and
bottom - for analysis. The six portions were similar but distinctly heterogeneous
and represent only a small fraction of the residue in the tank. The significant
components and their levels were:
0
II
C 40 - 55% MIC trimer -
/ \ 1,3,5-Trimethylisocyanurate «
CH3-N N-CH3 l,3,5-Trimethyl-l,3.5-triazine-
I I 2,4.6(lHI3H,5H)-trione
o-c c-o
\ / (Trimer)
N
I
CH3
O
II
C 13 • 20% 1,3-Dimethylisocyanurate -
/ \ l,3-Dimethyl-l,3,5-triazine-
CH3-N N-CH3 2,4.6(1 H.3H,5H)-trione
O-C C-O (DMI)
N
H
(CH^N.HCl 3 • 4% Trimethylamine hydrochloride
fTMA-HCl)
(CH3X,NH.HCI 2 • 2.5% Dimethylamine hydrochloride
(DMA.HCI)
CH3NHrHCI 1 -1.5% Monomethylamine hydrochloride
(MMA.HCI)
CP 2-3% Additional hydrolyzable chloride
14
-------
CH--N
I
H
o
II
c
\
N
N-CH,
I
C-0
H CH,
4-8%
1,3,5-Trimethylbiuret
(TMB)
O
II
C
/ \
CH3-N N-CH3
H2C C-0
V
I
CH,
5-7%
Dihydro-1,3,5-trimethyl-
1,3,5-triazine-2.4
-------
o
II
A
CH,-N N-CH3 1 • 2% 1,3-Dime thy] urea
H H (DMU)
Fe, Cr, Ni 0.18 • 0.26% Iron, chromium and nickel salts
CH3NCO 0.2 -1.0% Methyl isocyanate
CHCIj 0.4 -1.5% Chloroform
Up to 2% Water
The water analysis may not reflect actual water content in the residue
immediately following the event, because of the tendency of the water sample to
absorb water from the atmosphere (hygroscopic).
The foregoing analyses were made using several analytical methods. The team
has a high degree of confidence in the accuracy of the results, which have been
reviewed by other scientists within Union Carbide Corporation.
The team faced the task of:
— proposing the chemistry which could lead to these components
— experimentally replicating the reactions to form the identified
components
— discovering the route by which the initiating substances entered Tank
610.
5.3 Chemistry by Which the Components Could be Formed
Three significant findings about the residue were:
— The presence of a relatively large proportion of chloride, approximately
6 percent, requires that a chlorine-containing compound was in the tank,
and that before and/or during the event H was converted to hydrolyzable
chloride. About 87 percent of the total chloride present in the tank
residue samples was hydrolyzable chloride.
— Iron, chromium and nickel were present in approximately their ratio in
type 804 stainless steel, suggesting that their origin was from the
corrosion of stainless steel. Under certain conditions, chlorides are
known to corrode stainless steel, particularly in the presence of water.
The amount of residue in the tank has been estimated to be 10,000
pounds, on which basis the metal content in the residue would have been
about 20 pounds.
16
-------
— The mixture of organic components found in the residue requires
complex chemistry to explain.
If iron-catalyzed trimerization of MIC had been the sole reaction, the only
significant product expected would have been MIC trimer. Though trimer was the
most abundant component, it comprised only about half the product Trimerization
certainly occurred, but there were other reactions also.
(3) SCHSNCO
iron catalyst
0
II
c
/\
8 1 1
o-c c-o
N
CH3
MIC Trimer
Linear polymerization of MIC is a known reaction. However, no linear polymer
was found in the residue.
If hydrolysis of MIC by water had been the sole reaction, the only significant
products expected would have been a mixture of 1,3-dimethylurea and
1,3,5-trimethyIbiuret. Although both were found, they comprised only a small
percentage of the total residue.
(4) CHjNCO + HgO - *• CHjNHj * C02
MIC Monomethylamlne
(5) CHjNI^ + CH3NCO * CHjNHCONHCH,
1,8-Dimethylure* (DMU)
(6) CHgNHCONHCH, * CH^CO *• CHjNHCONCONHCH,
CH,
l,S,5-TrimethyIbiuret (TMB)
Sodium hydroxide (caustic aoda). the solution in the vent gas scrubber,
accelerates a water/MIC reaction; but the fact that the residue contained less than
0.001 percent sodium indicates that sodium hydroxide did not play a role.
It is known from the literature (J. Pkys Cfom. C9 791 (1965)) that liquid
moDomethylamine hydrochloride disproportionates at high temperatures to an
17
-------
equilibrium mixture of ammonium chloride and other more highly methyl-
substituted amine hydrochlorides, as illustrated in the following equation. The
equilibrium mixture is temperature dependent
(7) CHSNH2.HC1 iZ£ NH4CI + (CH^NH.HCI * (CH^N.HCI * (CH,)4NCI
Such disproportionation could explain the presence of DMA.HCI and TMAIHCI.
Subsequently the reaction of DMA with MIC could form TMU and TRMB.
+ CH.NCO
(8) (CHJUNH * CH.NCO—*• (CH^CONHCH, —*• (CHj^NCONCONHCH,
DMA MIC TMU r*u
OH,
TRMB
It is also known from the literature that methylated ureas will undergo exchange of
amine groups at elevated temperatures (US Patent 3,633,614 (1972) and Japanese
Patent 72 07004 (1972)). And in the process for preparing DMU from MM A and
carbon dioxide, monomethylurea and TMU are also formed (Japanese Patent 79
30116 (1979)). These references suggest possible routes to TMU and TRMB under
conditions in the tank during the event.
The second most abundant compound in the core sample of the residue was
DMI. This compound is readily formed by the reaction of MIC with isocyanic acid
or its salts (J. Org. Chem. 35 (7) 2253 (1970) and 44 (22) 3769 (1979)). DMI and
other derivatives of isocyanic acid have occasionally been isolated as a residue from
process streams in the MIC unit, notably in the phosgene stripping still, which
operates above 100°C.
0
II
s\
(9) 2CH3NCO + HNCO • >• CHg-N N-CH3
Isocyanicacid ^ {,_Q
\ /
N
I
H
DMI
In the MIC process the HNCO can be produced from reaction of phosgene with
ammonia formed from dissociation of ammonium chloride produced in equation 7,
or with the small amount of ammonia contained in the methyiainine.
(10) COClj «• NH3 * HC1 * HjNCOCl »• HNCO + HC1
Isocyanic acid can also be produced by thermal decomposition of
dimethylallophanyl chloride (DM AC) in chloroform (UCC unpublished results).
DMAC can be formed from MIC and methylcarbamoyl chloride (UCC). In equation
12 the DMAC configuration shows how methyl chloride might be formed, leaving
the remainder of the molecule to become HNCO and If 1C.
-------
(II) CHSNCO * CHSNHCOCI
MCC
COCI
DMAC
0
II
c
(12) H-N
\
N-CH
'c£\ c-o
x X
\ n >
CH3C1 + HNCO + CH3NCO
DMAC
The dione can be readily formed by reaction of TMB with formaldehyde or its
acetaJ (Butf. Soc. Chim. FT., 5-6 Pi. 2, 1419 (1975)).
(13)
CH3-N
I
O
II
C
/\
N-CH
H C-O
N
/ \
CH-0
CH3-N
0
II
C
\
N-CH
H
CH
TMB
I I
H,C C-O
2 X /
N
CH3
Dione
H»O
Whether formaldehyde could have been formed during the event is uncertain.
but it is known that chloroform and water do react under certain conditions to
produce formic acid and hydrochloric acid (J. Chem. Soc. 398 (1959) and J. Research
InsL Catalysis Hokkaido Univ. 3 147 (1955)). Also, it is likely that dichloromethane
(Cr^Clo) was formed during the event, and served as an intermediate to produce
formaldehyde. Chloroform can be reduced to dichloromethane under a variety of
conditions (METU J. Pure Appi Set. C (3) 255 (1973), Izv. Akad. Navk SSSR, Ser.
Khim. (2) 354 (1982) and Ind. Eng. Chen. Prod. Res. Rev. 17 (3) 236 (1978)). The
hydrolysis of dichloromethane to formaldehyde is also well documented (J. Chen.
Soc. 1326 (1958); U. S. Patents 1.616.533 (1927) and 1.679.673 (1927), Japanese
Patent 72 02608 (1972)).
In summary, all of the significant components of the residue could have
resulted from recognized, albeit complex, chemical reactions.
5.4 Replication of the Chemistry
The team decided that i'n order to substantiate the chemistry, it would have to
establish reaction conditions which would produce the major components in the
residue. Elevated reaction temperatures. 120-275'C, were used for two reasons: to
19
-------
•void waiting through an induction period which might be days long, and to achieve
temperatures in the range likely reached during the event. The surface to volume
ratio was much greater in the laboratory experiments than in Tank 610, so external
heat had to be supplied to replace heat losses through the walls. Reactions were
conducted in closed systems, whereas during the reaction in the tank, MIC vapor
escaped through the safety valve. Accordingly, reactants were mixed in proportions
likely to react to complete conversion to give the residue components found in the
tank.
The presence of so much chloride in the residue, the fact that chloroform is the
process solvent, and the conclusion that an excessive amount of chloroform
probably was in the tank as discussed in Section 5.5, led the team to include
chloroform in the reaction simulations. Experiments in which MIC and chloroform
were the only reactants did not yield the products observed in the residue.
However, when water was present, at high temperature the chloroform was
consumed, chloride ion was formed, and the major components found in the tank
residue were produced. Water was included in the experiments because of the
presence of ureas and biurets in the core samples. These materials are produced
when MIC reacts with water.
Because of the high level of metals in the residues, direct addition of iron
chloride as a reaction catalyst was included in some reactions. This produced about
the same product mix as exposure of the hot reaction to types 304 and 316
stainless steel experimental reactors. Stainless steel was also found to corrode
under these hot reaction conditions. At temperatures in excess of 200°C, the
reaction mixture is sufficiently corrosive to type 304 stainless steel so as to account
for the metals found in the tank residue.
Also investigated were various levels of hydrogen chloride in the form of MCC,
DMAC, phosgene and hydrochloric acid. Experimentation most recently concluded
indicates that their presence had a less significant effect on the residue composition
than water, chloroform and iron. Other reactive chemicals added to MJC in the
experiments to determine their effect on residue composition included materials
which were available within the plant, e.g., monomethylanu'ne, chlorine,
trimethylamine, carbon tetrachloride, dichloromethane, and process residues. None
of these additions was as effective in replicating the desired residue composition as
a combination of water, chloroform and iron.
Through the above experimentation, involving over 500 experiments, H was
demonstrated that the major components found in Tank 610 residue can be
produced by materials and conditions whose presence in the tank can be accounted
for. MIC, chloroform, water, iron and high temperatures. In general, it appears
that reaction temperatures in excess of 200°C are needed to produce the mixture
of products, especially to form large amounts of DMI and significant quantities of
TMU. In particular, there was a strong positive correlation between the amount of
chloroform added and the amount of DMI produced. Chloroform also influenced the
formation of Dione and of the amine hydrochlorides, probably through Ha
conversion to dichloromethane. which reacted with some of the DMU and TMB
already formed.
The one material found in significant proportions in the tank but not in the
experiments is TRMB. Its absence can be explained by rapid depletion of the MIC
during the experiments, so that none was left to react with TMU after that
compound was formed. In Tank 610 there must have been MIC present during the
20
-------
cooling period which followed reseating of the safety valve, so that DMU and TMU
would have reacted with it to form TMB and TRMB. respectively. Both reactions
have been demonstrated in the laboratory.
Considering the product distribution in the tank, the experimental results, and
the probability that solids and liquids were vented with the gas during the incident
(Section 6.0). the amounts of water and chloroform originally present in Tank 610
are estimated to have been 1,000 to 2,000 pounds of water and 1,500 to 3,000
pounds of chloroform.
5.5 Sources of Reaction Components
Given the technical data and information available to date, it seems likely that
the reaction was caused by the entry of a large quantity of water into Tank 610,
which contained MIC with an abnormally high level of chloroform. The presence of
chloroform alone in MIC would not have caused the incident.
Water
The exact source of the water is not known, but laboratory work demonstrated
that 1,000 to 2,000 pounds of water would have accounted for the chemistry of the
residue. Water could have been introduced inadvertently or deliberately directly
into the tank through the process vent line, nitrogen line or other piping. Records
indicate that the safety valve discharge piping to the RWH from four MIC process
filters was being washed shortly before the incident. Oral discussions indicate that
a slip blind was not used to isolate the piping being washed. However, entry of
water into Tank 610 from this washing in the MIC unit would have required
simultaneous leaks through several reportedly dosed valves, which is highly
improbable.
Tank 610 could not be pressurized on November 30 and December 1, although
nitrogen was reported to be flowing into the tank (per records and discussions). It
is possible that a vent valve was leaking and/or the rupture disc was not intact and
the safety valve was leaking. Oral discussions indicate that the rupture disc was
tested and found to be intact on November 30. If nitrogen was indeed escaping, the
escape route could also have provided a route of entry for water.
Chloroform
Records for the operation of the MIC refining «t£D for the period October 18-22,
1984, indicate that the column was being operated during part of that time at
higher than normal temperatures. Computer simulation of the distillation column
indicates that under those conditions, a higher than normal amount of chloroform
would have distilled along with the MIC.
Also, there is no record of analyses of MIC in Tank 610 after October 19.
However. MIC Unit logs show that in preparation for shut down, MIC containing a
high concentration of chloroform was sent to Tank 610 instead of to the empty
Tank 619. The chloroform content of the MIC left in the MRS product line was
between 12 and 16 percent, as reflected in samples taken and analyzed on
December 16. This indicates that the last product from the MRS to Tank 610
contained well over the specification maximum of 0.5 percent chloroform. This
would in turn have raised the chloroform content of Tank 610.
In addition, material sampled on December 5 and 9 from the MIC transfer line
from the storage tanks to the Derivatives Unit contained 2.5 and 1.9 percent
chloroform, respectively. Two MIC samples from the Derivatives Unit charge tank
on December 7 contained 1.5 and 0.8 percent chloroform. All these results were
21
-------
higher than normal. Samples from the other possible sources, Tanks 611 and 619.
showed lower amounts of chloroform, i.e., less than 0.3 percent chloroform in Tank
611; and less than 0.6 percent in Tank 619. Therefore, it appears that the high-
chloroform MIC transferred to the charge tank came from Tank 610.
Iron
As discussed in Section 5.4, high corrosion rates of stainless steel at high
temperatures would be required to generate the metals found in the tank residue.
These kinds of corrosion rates have been demonstrated in laboratory experiments.
The corrosion rates were found to be 10 to 20 mils per year at 100°C and 1,700 to
3,500 mils per year at 200°C using MIC containing 18 percent chloroform and 2 to
7 percent water. (One mil equals one-thousandth of one inch.)
The rates, dependent on chloroform level, are far lower at room temperature.
Corrosion rate tests showed the rates to be 0.18 mils per year with refined MIC
containing 16 percent chloroform, compared with 0.04 mils per year with MIC
containing 2.5 percent chloroform. These rates were measured over a 15-day
exposure period. The corrosion rate with MIC containing a normal amount of
chloroform (0.5 percent maximum) is negligible (nil) at room temperature.
22
-------
6.0 A HYPOTHESIS FOR THE EVENT
Based on the information gathered by the team, several hypotheses and
scenarios were proposed and examined for compatibility with known facts. They
were tested experimentally where necessary for compatibility with pertinent
analytical and chemical data. The investigative process outlined in Section 2.0 was
used to choose the most probable scenarios. A single scenario to fit the event could
not be proposed with complete certainty, since sufficient critical information was
missing in each case due to restrictions placed on the team. However, a high
probability can be given to the following scenario.
One to two thousand pounds of water entered the tank. Although entry from
vent headers (RWH, PVH) cannot be ruled out at this time, direct introduction of
water through the vent or other piping has a higher probability for occurrence
because it does not require the simultaneous leakage through a series of reportedly
closed valves. The chemistry, described in Section 5.0, indicates that the large
amount of water and higher-than-normal levels of chloroform were necessary to
obtain the particular mix of significant residue components found in the core
samples. The large amount of water was necessary to generate the heat needed to
initiate and accelerate the subsequent reactions. The temperature of MIC in Tank
610 before the incident was at 15 to 20°C as compared to the requirement of about
0°C. The lower temperature would have retarded the reaction rates and
considerably extended the time available for corrective action. The refrigeration
system provided to cool the MIC in the storage tanks had been made non-
operational in June, 1984.
The reaction of water with MIC led to an increase in pressure due to evolution
of carbon dioxide as well as an increase in temperature due to the exothermicity.
The higher pressure, 10 psig, noticed at 11:00 PM on December 2 is believed to be
due to this phenomenon. The increase in temperature was not signaled by the tank
high-temperature alarm since it had not been reset to a temperature above the
storage temperature. The nitrogen escape route discussed in Section 5.5 would
have allowed carbon dioxide and MIC vapor to flow to the VGS. Since the VGS was
not in service prior to the incident, this would explain the untraceable MIC leak
reported in the MIC unit at 11:00 PM. Although the pressure in the tank could not
be increased by nitrogen addition in earlier attempts, it increased at this time due
to the far higher rates of carbon dioxide evolution.
A transfer of 2,300 pounds of MIC was made to the Derivatives Unit charge
tank at about 11 PM as shown in transfer logs. As discussed in Section 5.5, this
probably was from Tank 610. Since it may have come from Tank 610 after the
introduction of water, this material would be expected to contain water. This water
would have eventually been consumed by reacting with MIC. Alternatively, there is
the possibility that the transferred material may not have contained water because
the transfer point is the tank bottom sump. The last MIC to enter the tank through
the make line to the sump on October 22 contained between 12 and 16 percent
chloroform, and would have had higher density than water. Therefore, the water
introduced at the top of the tank may not have reached the lower portion of the
tank. The tank contents were not mixed after October 19.
As the temperature in Tank 610 increased due to the exothermic reaction of
MIC with water, the corrosion rate increased markedly because of the presence of
an abnormally high level of chloroform'. The iron thus produced catalyzed a
concurrent exothermic trimerization of MIC. Both reactions accelerated as
temperature increased and the mixing of water and iron with the bulk of the MIC
further increased the violence of the reaction. As temperature and pressure
23
-------
increased rapidly, the rupture disc in the line to the safety valve burst (if not
already broken) and the safety valve opened at 40 psig. The chemistry of the
residue shows that 1,000 to 2,000 pounds of water and 1,500 to 3,000 pounds of
chloroform were required. Calculations have shown that the presence of 2,000 or
more pounds of water would have caused the incident without chloroform being
present.
The amount of material forced out of the tank cannot be determined exactly.
However, based on the heats of reaction, about 40 percent of the MIC reacting
would release enough heat to raise the temperature of the tank and vaporize the
remaining 60 percent of the MIC. From this calculation, 36,000 pounds of solids
would be expected in the tank instead of the estimated 10,000 pounds. One
explanation for the lower amount of solids is the loss through the safety valve of
some liquid and solids along with the vapor, due to foaming of the reaction mass
during the most active period of reaction.
An initial discharge rate of 10,000 pounds per hour of MIC vapor would occur
when the pressure in the tank reached 40 psig, the rupture disc burst and the
safety valve opened. The exact time period the safety valve remained open is not
known, but a two-hour period is reasonable based upon the available information.
In order to discharge most of the contents of the tank within two hours, the
pressure had to average 180 psig. At these conditions, material would be
discharged at a rate of 40,000 pounds per hour: 29,000 pounds per hour of vapor
and 11,000 pounds per hour of solids/liquid mixture. Approximately 54,000 pounds
of unreacted MIC left Tank 610 together with approximately 26,000 pounds of
reaction products. As discussed in Section 5.4, the maximum temperature reached
was probably greater than 200°C.
A better estimate of the maximum pressure and temperature reached in the
tank can be made when it is inspected and measurements are made to determine
whether the metal started to yield. The start of permanent head deformation of
Tank 610 would occur at about 100 psig. The stainless steel would start to yield
above 200 psig. The tank would not be expected to rupture until the pressure
exceeded 300 psig. The tank did remain intact
The sequence of reactions required to duplicate the composition of the core
sample cannot be determined exactly. These are discussed in Section 5.0. However,
recognizing the limitations imposed upon the investigation team, the overwhelming
experimental data indicates that a large amount of water must have entered the
tank containing MIC with an abnormally high amount of chloroform, and initiated
the chemical reactions.
24
-------
ATTACHMENT TWO
-------
SUMMARY OF REGIONAL RESPONSE TEAM COMMENTS
REGARDING WEST VIRGINIA CONTINGENCY PLANS
As provided for in Subpart D of the National Contingency Plan, the
Regional Response Team (RRT) has met several times since the incident in
Bhopal for the purpose of reviewing existing contingency plans relating
to the Union Carbide plant in Institute, West Virginia. Member agencies
and their representatives participating in this review included the
following:
1. EPA Tom Voltaggio (Co-Chair)
2. USCG Captain Pete Lauridsen (Co-Chair)
3. FEMA Karen Larsen
A. HHS Frank Piecuch
5. BOJ Dave Buente
6. OSHA Ken Gerecke
7. DOD Jack Dempsey
8. WVDNR Ron Shipley
9. WVOES Manny Griffith
10. EPA (HQ) Jim Makris
11. EPA (ERT) Dr. Joe Lafornara
12. CDC Kent Gray
13. DOI Anita Miller
14. NOAA Frank Rossi
15. USD A Bob Adams
Our review sought to determine what improvements can be made to
Contingency Plans relevant to Union Carbide's Institute Plant. We believe
the need for detailed pre-planning and contingency plan development is
crucial in order to provide for efficient, coordinated and effective
response to a release of hazardous substances, thus minimizing the potential
impact on human health, welfare and the environment.
The various Plans that were reviewed include the following:
1. W.V. Hazardous Materials Emergency Response Plant (draft,
dated 12/84) prepared by Office of Emergency Services (final version
anticipated in March 85).
2. Kanawha County and local jurisdictions - Emergency Response
Operations Plan (updated 1982; in print now; may be subject to change
based on current events).
3. West Virginia Emergency/Disaster Plan (dated 12/83; prepared
by OES).
-------
4. Kanawha Valley Industrial Emergency Planning Council Manual
(revised 5/81)
5. Union Carbide Institute Plant Emergency Procedures Manual (updated
12/84).
We realize that some of these plans are in draft stage, and that
because of all the concern presently focused on Union Carbide's Institute
facility, there are events underway which may result in supplemental
plans or significant changes to existing ones. Given this dynamic situation,
our comments are generally broad in scope, attempting to identify major
issues of concern, and as such should not be construed as delineating in
detail each specific item which requires attention.
Based on our review, we believe that some of the plans, as presently
organized, need to be improved to effectively deal with the wide range
of hazardous situations that can occur. In particular, the County Plan
appears to need significant improvements to fulfill the need for a com-
prehensive operations plan that could be employed in a variety of emergency/
disaster situations.
Our comments have been organized into six major categories, as
presented in the following paragraphs. These categories are somewhat
interrelated, and should be read in that context.
1. Responsibility
A. Assignment of the responsibility and authority for overall
direction and control during an accident needs to be established in a
single location at each governmental level and within each plan.
The clear and specific designation of a lead Agency/ individual for
response organizations is needed to help avoid disputes between Agencies
at a critical time in the response operation.
B. The assignment of authority to order an evacuation of the general .
public is vital, but appears inconsistent throughout -the plans. References
are made to "senior law enforcement official," "senior elected official,"
"first responder," "field CP group," etc. Clarification may be needed with
regard to KVIEPC's ability to initiate protective actions by the general
public via activation of the Emergency Broadcast System.
C. While it is quite logical that KVIEPC and its constituent companies
have a major coordinating role in response to incidents within their
plants, it is unclear how and when the local, County and State authorities
will assume control regarding offsite effects of an incident
occurring. This does not reflect the preeminent authority and responsibility
of local government to protect the health and safety of its citizens.
-------
2. Coordination
A. Coordination between the plans needs to be strenghtened. Questions
such as who has authority, what are its limits (both legal and geographical),
when does transition of authority take place, etc., need to be clarified.
Formal, written agreements between the various response organizations
should be considered. Such agreements could identify availability of
resources, methods of gaining access to them during an emergency, uniform
protective action guides, and provide a mechanism to limit liability for
assisting personnel. They would help to provide a synchronized response
while avoiding duplicated effort.
B. Air modelling and meteorological projections from the National
Weather Service Forecast office to ascertain the area of Impact, and more
attention to the location and characteristics of the population at
risk should be incorporated into the plans. Response patterns should be
predicatated on such information and be predictable. Without a map of
prevailing wind patterns, it is difficult to determine the potential
people and resources at risk. In the case of an air release there is a
possibility that areas that seem remote could be affected. If this were
the case, response personnel and residents in remote areas must receive
adequate notification of a need to evacuate or take other protective
action.
C. Relationships with medical authorities should be clarified
regarding health effects consultation, including decisions to evacuate
(who makes the decision, based on what information, etc.) and personal
protection equipment to be worn. Based on the types and numbers of
injuries that could be expected for various possible hazardous materials
incidents, the capabilities of available medical facilities should be
evaluated. Where helpful, agreements should be established between local
governments, industry and hospitals; hospital plans for dealing with
foreseeable emergencies should be incorporated or referenced in the plan.
Joint government/hospital drills or exercises should test response plans
for casualties resulting from the most probable occurrences.
3. Operations
A. An overall "concept of operations" is needed, particularly in
terms of determining 1) the severity of an incident and the type of
response actions necessary; 2) the adequacy of cleanup or response by
the responsible party; 3) compliance with all applicable local, State
and Federal requirements.
-------
B. The various plans need to be focused more directly towards
hazardous material incidents, and to incorporate appropriate private,
State, and Federal resources and capabilities available to respond to such
incidents. In other words, who will do what, when, and at whose direction?
C. Criteria should be considered for relinquishing controlled access
to an area which has been cleaned up and is able to be returned to
unrestricted use.
D. A mechanism for funding and performing necessary cleaning operations
in the absence of the spiller voluntarily undertaking this responsibility
should be developed.
E. Procedures for conducting monitoring of air, water, or food-
chain that may have been contaminated during a release, and continued
monitoring after the release to determine "how clean is clean" must be
addressed.
4. Notification
A. Clarification is needed regarding the notification system to be
followed between industry, local, County, and State agencies, as well as
the procedure to alert surrounding communities of an incident. There
must be mutually agreed upon bases for notification and exchange of
information between the various governments and the chemical industries
in the area. For this purpose, telephone listings should be incorporated
into all the plans, and periodic updating of notification lists should be
required.
B. It is not immediately apparent how and when the community awareness
alarm becomes activated, who's decision it is to sound it, or what the
consequences of an alarm are. In discussions with the community, more
information regarding this process should be made available. Tests of
the alarm systems and other notification procedures should be regularly
scheduled and conducted, and the effectiveness of the systems in reaching
the desired areas should be determined. (Guidance for the evaluation of
off-site warning systems is available through FEMA).
C. Contact mechanisms with the Federal response commuity need to be
identified and emphasized. Through the National Response Center (NRC)
(800-424-8802) there is a well established system in place to notify
Federal agencies of potential or actual emergency incidents. A list of
Federal contacts should be presented in a single section of all plans,
as well as notification/action sequences to be followed in the event of
a hazardous material incident.
-------
5. Public Information
A. The degree to which the Public has been involved in the planning
process and its awareness of the existence and significance of the various
notification systems, alarms, etc. is unclear. An ongoing community
relations program should be established to inform residents of the
community awareness alarm and its significance. Protective action guides,
potential limits, and types of recommended protective actions for the
general public should be pre-established and clearly understood by the
local population.
B. Methods to coordinate public information activities so as to
clearly and effectively communicate specific information and thereby
avoid confusion and rumors should be strengthened. Public information
releases during an emergency response should be prepared by a central
coordinating source. A system should be considered for releasing
information contained in the initial and follow-up messages received for
the accident site, including verification procedures* Standard news
release formats may be developed which only require the application of
site-specific information. Concise messages that are easily understood
by persons unfamiliar with the technical aspects of an accident should be
stressed.
6. Regional Inventory Needs
A. A comprehensive data base including the types, quantities, and
locations of hazardous materials, as well as toxicological information
related to these materials is needed. This should be provided to the
appropriate response personnel on an ong-oing basis. The companies could
provide periodic briefings to local response staff, identifying specific
chemical hazards that may be encountered in any sort of response effort.
Due to the rapid nature of a toxic release, the role of the companies
cannot be underestimated. While they are aware of the substance at
their facilities and the associated toxicology, it must be assumed that .
the majority of the standing response team will have to acquire and
assimilate the information before action be taken.
B. Availability and locations of personal protective equipment and
other equipment suitable for response needs should be identified. This
equipment should be tested, and training for first responders in its use
should be provided. Types of equipment needed for specific response
activities may vary with the type of emergency. Guidance on the proper
selection of the available personal protective equipment should be made
available. OSHA and NIOSH both have recommended programs for users of
-------
personal protective equipment. These include:
1) material identification
2) environmental surveillance
3) medical surveillance
A) selection of equipment
5) training in and fitting of equipment
6) decontamination and cleaning
7) inspection, maintenance and storage
8) written program
9) program review and evaluation
10) operational use
Our review of the various plans reveals that the documents, by
themselves, do not indicate the establishment of an effective personal
protective equipment program useful in the control of hazardous material
incidents. The elements which are lacking may be adequately addressed in
other documents, however.
C. Training for first responders and others in the response team is
needed regarding regulations pertaining to personal protective equipment. The
deployment of individuals to the site without proper training or equipment
can result in casualties. Periodic briefings for employees and response
personnel should be provided. Simulations of releases of hazardous
substances should be held to test response systems. Joint government/hospital
drills or exercises should test response plans for casualties resulting
from the most probable occurrences. Training and plan update should, of
course, be part of an ongoing process. The need for training of governmental
services personnel should be based to some extent on the agreements
spelled out between government and industry regarding responsibility for
various parts of the emergency response.
D. Medical support capabilities (location of hospitals, critical
care facilities, etc.) and special evacuation considerations (schools,
nursing homes, etc.) should be addressed. Based on the types and numbers
of injuries that could be expected for various possible hazardous material
incidents, the capabilities of available medical facilities should be
established between local governments, industry and hospitals; hospital
plans for dealing with foreseeable emergencies should be incorporated or
referenced within the plan. Evacuation plans including special consideration
such as schools, nursing homes, etc. should be formulated with the
involvement of these institutions. This should be followed by drills
supervised by response team members.
-------
We note that Congressman Bob Wise, in his 1/11/85 memorandum (copy
enclosed) has identified a number of specific proposals relating to
emergency response in the Kanawha Valley. These suggestions are well
thought-out and compatible with our comments, although in general they
go beyond the limits of the RRT's purview.
In summary, the member agencies of the RRT were requested to review,
based on their particular areas of expertise, each of the available
Contingency Plans, and provide relevant comments. There has not been an
attempt to develop totally new plans, or to substantially rewrite the
existing ones, but rather to point out areas which should be strengthened.
Our review was conducted with the understanding that it is impossible,
by definition, to totally prevent accidents and emergencies from occurring.
By insuring that contingency planning procedures are thorough and properly
implemented, the impact of such events can be minimized. The concept of
accident prevention, either through Regulatory procedures or voluntary
changes enacted by Union Carbide, was outside the scope of the RRT review.
Furthermore, we did not intend to determine the degree of risk to
the local population, or to quantify/qualify it in any way. We did not
believe it appropriate, at this time, to evaluate methods of providing
for, or implementing, mass evacuations from the surrounding communities.
We believe that Congressman Wise is certainly correct in his assessment
of a need for additional development and public communication regarding
this procedure.
By transmitting our comments to the WVDNR and (through FEMA) to OES, we
hope these Agencies will cooperatively insure that our comments are
implemented at the appropriate levels in the planning process. We suggest
that a task force consisting of representatives from FEMA, EPA, WVDNR and
WVOES be formed to oversee the implementation of these recommendations
for improvement. The RRT will be available as an advisory body if needed
by the task force.
The RRT members are commended for their valuable contributions to
this crucial review process. A great deal of time and thoughtful
consideration has been provided, frequently under extremely tight schedules.
These efforts will undoubtedly enhance the level of preparedness in the
Kanawha Valley, and may lead to an increased awareness nationally of the
importance of contingency planning and prevention methods.
-------
ATTACHMENT THREE
-------
Attachment Three
Membership of Federal-State Task (Jr-oopon the review of the Resumption of
MIC Production by Union Carbide Corporation, Institute, West Virginia.
U.S. Environmental Protection Agency (lead agency)
Greene Jones, Chairman - B.S. Civil Eng.
Robert Kramer - B.S. Env. Sci.
Rick Horner - B.S. Chem. Eng. - M.B.A. Bus. Adm.
Joseph Lafornara - Ph.D. (Srganic^Chem. |n di v
Peter Schaul - M.S. Chem. £ng~.
Les Evans - B.S. Chem. - M.S. Bus. Adm.
Michael Zickler - M.S. Env. Sci.
Robert Borgwardt - B.S. Chem. Eng.
Occupational Safety and Health Administration
Doug Ray - B.S. Biology
Rich Jeffrey - M.S. Safety and Health Management
Federal Emergency Management Administration
Provide technical assistance on emergency response planning.
James Asher - 35 years experience in Emergency Response
Karen Larson - B.A. Soc. and Dev. Phy.
West Virginia Department of Natural Resources
Ron Shipley - B.S. Chem. Eng. - MPA, J.D.
Sanja Kanth - B.S., M.S. Chem. Eng.
West Virginia Air Pollution Control Commission
Carl Beard - B.S. Chem.
Dave Fewell - B.S. Chem. Eng.
Steve Pierce - B.S. Chem. Eng.
------- |