United States
Environmental Protection
Agency
United States
Occupational Safety and
Health Administration
EPA550-F-98-01
September 1998
&EFA
EPA/OSHA JOINT
CHEMICAL
ACCIDENT
INVESTIGATION
REPORT
Surpass Chemical Co., Inc.
Albany, NY
EPA and OSHA
i Printed on recycled paper
-------
-------
The EPA/OSHA Accident Investigation Program
Under a Memorandum of Understanding (MOU), EPA and OSHA are working together
to investigate certain chemical accidents. The fundamental objective of this joint effort is to
determine and report to the public the facts, conditions, circumstances, and causes or probable
causes of any chemical accident that results in a fatality, serious injury, substantial property
damage, or serious off-site impact, including a large scale evacuation of the general public. The
ultimate goal is to determine the root causes in order to reduce the likelihood of recurrence,
minimize the consequences associated with accidental releases, and to make chemical production,
processing, handling, and storage safer. Reports, such as this one, are issued by the agencies to
describe the accident, discuss the root causes and contributing factors, and summarize the findings
and recommendations.
Prior to releasing investigation reports, OSHA and EPA must ensure that the report
contains no confidential business information. The Freedom of Information Act (FOIA), the
Trade Secrets Act, and Executive Order 12600 require federal agencies to protect confidential
business information from public disclosure. To meet these provisions, OSHA and EPA have
established a clearance process in which the companies mentioned in the report are provided a
portion of the draft report. This portion contains only the factual details related to the
investigation (not the findings, the conclusions nor the recommendations). Companies are asked
to review this factual portion to confirm that the draft report contains no confidential business
information (CBI). As part of this clearance process, companies often will provide to OSHA and
EPA additional factual information. In preparing the final report, OSHA and EPA consider and
evaluate any such additional factual information for possible inclusion in the final report.
Chemical accidents investigated by EPA Headquarters are conducted by the Chemical
Accident Investigation Team (CAIT) located in the Chemical Emergency Preparedness and
Prevention Office (CEPPO) at 401 M Street SW, Washington, DC 20460, 202-260-8600. More
information about CEPPO and the CAIT may be found at the CEPPO Homepage on the Internet
at "www.epa.gov/ceppo". Accidents investigated by OSHA Headquarters are conducted by the
Chemical Accident Response Team (CART) located in the U.S. Department of Labor - OSHA,
Directorate of Compliance Programs, Washington, DC 20210, 202-219-8118. More information
about OSHA may be found at the OSHA Homepage on the Internet at "www.osha.gov".
At the time that EPA and OSHA decide to jointly investigate an accident under the MOU,
an investigation team is formed consisting of representatives of both EPA's CAIT and OSHA's
CART. This team is referred to as the Joint Chemical Accident Investigation Team (JCAIT).
-------
U.S. Chemical Safety and Hazard Investigation Board (CSB)
In 1990, the U.S. Chemical Safety and Hazard Investigation Board (CSB) was created as
an independent board in the amendments to the Clean Air Act. Modeled after the National
Transportation Safety Board (NTSB), the CSB was directed by Congress to conduct
investigations and report on findings regarding the causes of any accidental chemical releases
resulting in a fatality, serious injury, or substantial property damages. In October 1997, Congress
authorized initial funding for the CSB. The CSB started its operations in January 1998, and has
begun several chemical accident investigations. More information about CSB may be found at the
CSB Homepage on the Internet at "www.chemsafety.gov".
For those joint investigations begun by EPA and OSHA under the previously mentioned
MOU and prior to the initial funding of the CSB, the agencies have committed to completing their
ongoing investigations and issuing public reports. Under their existing authorities, both EPA and
OSHA will continue to have roles and responsibilities in responding to and investigating chemical
accidents. The CSB, EPA, and OSHA (as well as other agencies) are developing approaches for
coordinating efforts to support accident prevention programs and to minimize potential
duplication of activities.
Basis of Decision to Investigate
On Tuesday, April 8, 1997, a 5,700-gallon hydrochloric acid (HC1) storage tank ruptured
while being filled at the Surpass Chemical Co., Inc. The spill of HC1, a corrosive and toxic
chemical, resulted in injuries to employees and members of the public, as well as public
evacuations. EPA and OSHA considered the impacts of the tank failure with respect to the
MOU criteria and the potential for lessons-learned and decided to initiate a joint investigation.
The scope of the investigation was to determine the immediate and root causes of the tank failure
and to make recommendations that could assist Surpass and others to prevent similar accidents
from occurring in the future.
-------
Accident Investigation Report
Surpass Chemical Company, Inc., Albany, New York, April 8, 1997
Executive Summary
On Tuesday, April 8, 1997, at approximately 8:59 a.m., a 5,700-gallon hydrochloric acid
(HC1) storage tank ruptured during filling at the Surpass Chemical Co., Inc. (Surpass), in Albany,
New York. The failure of the HC1 tank caused a significant portion of its liquid contents (which
totaled about 4,800 gallons of 31% HC1) to suddenly surge over the secondary containment. The
force of the liquid also caused a break in the secondary containment wall. Witnesses described
seeing greenish-yellow fumes drifting offsite as well as liquid material running offsite and along
the street curb to the storm drains. As a consequence of the incident, 8 workers and 32 others
were taken to the hospital. A 10-block area, including nearby businesses and residences, was
evacuated.
Based on the impacts of the incident and the potential for lessons-learned, EPA and
OSHA decided to undertake a joint chemical accident investigation to determine the immediate
and root causes of the HC1 tank failure and to make recommendations to Surpass, government,
industry, and others that could assist in preventing similar incidents from occurring in the future.
The Joint Chemical Accident Investigation Team (JCAIT) determined that the immediate
cause of the incident was the overpressurization of the HC1 tank. The team identified the root
causes as:
> Modifications to the venting of the HC1 tank were not within the tank
manufacturer's specifications for emergency venting.
> No hazard analysis of the modifications to the venting of the HC1 tank was
performed.
*• Inadequate preventive maintenance of the scrubber system.
Additionally, the JCAIT identified the following contributing factor:
*• Lack of a written standard operating procedure (SOP) for air off-loading of
deliveries to the HC1 tank, including an inadequate method for determining that the
delivery was complete.
The JCAIT has developed recommendations that address the root causes and contributing
factors in order to prevent a similar event:
> Surpass and other facilities should ensure that modifications to their equipment, in
in
-------
this case for the purposes of environmental control, do not create new hazards or
compromise safety.
*• Surpass and other facilities should maintain environmental control systems to
ensure continuous reliability and effective operation.
> Surpass should develop written standard operating procedures (SOPs) related to
the use of air pressure for off-loading HC1 and maintenance of the scrubber system,
including consideration of human factors such as adequate measuring devices to
reduce the chances of errors in determining the completion of the delivery.
*• EPA and OSHA should develop an alert to raise awareness about the need for
thorough consideration of safety when designing equipment or processes for
environmental control.
In addition to the root causes and contributing factors associated with the HC1 tank
failure, the JCAIT identified other potential problem areas that may have contributed to the
consequences of the incident. These issues included the location of incompatible materials (HC1
and sodium hypochlorite) near each other and the need for periodic inspection of storage tanks.
As appropriate, these issues will be addressed in any alerts that EPA and OSHA develop.
Also, Surpass is a member of the National Association of Chemical Distributors (NACD)
and participates in the NACD Responsible Distribution Process program, which encourages
continuous improvement in the safe handling of chemicals. A timely and thorough
implementation of the Responsible Distribution Process program by Surpass may have led to
improvements in Surpass's system to manage health, safety, and environmental concerns.
Another issue identified by the JCAIT is the listing of HC1 solutions under the Risk
Management Program (RMP) Rule. Under a recent modification to the list of regulated
substances for the RMP Rule, only anhydrous hydrogen chloride and HC1 solutions of 37% or
greater will be covered (62 FR 45130, August 25, 1997). As this incident demonstrates, solutions
with HC1 concentrations below 37% may pose potential hazards to human health or the
environment. The circumstances of this incident should be considered in any future evaluation of
how to list HC1 solutions for the RMP Rule.
IV
-------
Table of Contents
1.0 Introduction 1
1.1 Description of the Event 1
1.2 Scope of Investigation 1
1.3 Structure of Report 1
2.0 Background 2
2.1 Facility Information 2
2.2 Process Information 2
2.3 Chemical Information 9
3.0 Description of the Incident 11
3.1 Chronology of Events 11
3.2 Consequences of the Incident 13
3.3 Description of the Emergency Response 16
4.0 Analysis and Significant Facts 16
4.1 Significant Facts 17
4.2 Analysis 20
5.0 Conclusions 25
5.1 Causes 25
5.2 Root Causes and Contributing Factors 25
6.0 Recommendations 27
7.0 Other Findings 28
References 30
Appendices
A Joint Chemical Accident Investigation Team (JCAIT) Members 31
B Industry Codes 32
C Other Accidents Involving Atmospheric Pressure FRP Tanks 33
D Modeling of Venting System 34
-------
List of Figures
Figure 1 Layout of Tank Area (Top view) 4
Figure 2 Representation of Offloading Operation 5
Figure 3 Diffuser Section of Vent System 6
Figure 4 Separation of HC1 Tank 14
Figure 5 Top of HC1 Tank 14
Figure 6 Closeup of Bottom of HC1 Tank 15
Figure 7 Closeup of Top of HC1 Tank 15
Figure 8 Scrubber Tank 19
Figure 9 Diffuser Section of Vent System 19
Figure 10 Modeled HC1 Storage Tank Pressure, Scenario: No Fouling of Diffuser
Openings 23
Figure 11 Modeled HC1 Storage Tank Pressure, Scenario: 84% Reduction in the Area of
Diffuser Openings 24
VI
-------
1.0 Introduction
1.1 Description of the Event
On Tuesday, April 8, 1997, at approximately 8:59 a.m., a 5,700-gallon hydrochloric acid
(HC1) storage tank ruptured during filling at the Surpass Chemical Co., Inc. (Surpass) in Albany,
New York. The failure of the HC1 tank caused its liquid contents (about 4,800 gallons of 31%
HC1) to suddenly surge over the secondary containment. The force of the liquid also caused a
break in the secondary containment wall. Witnesses described seeing greenish-yellow fumes
drifting offsite as well as liquid material running offsite and along the street curb to the storm
drains.
Local, state, and federal officials responded. As a consequence of the incident, 8 workers
and 32 members of the public were taken to the hospital, treated, and released. A 10-block area,
including nearby businesses and residences, was evacuated. Surpass and its contractor remediated
the spill in coordination with local, state, and federal officials.
1.2 Scope of Investigation
At the conclusion of the emergency response and remedial actions, EPA and OSHA
initiated an investigation by a Joint Chemical Accident Investigation Team (JCAIT). The JCAIT
was directed to determine the immediate and root causes of the HC1 tank failure and to make
recommendations to Surpass, government, and industry that could assist in preventing similar
incidents from occurring in the future. The investigation was to be concurrent with the OSHA
compliance investigation. This report represents the conclusion of the JCAIT's investigation.
1.3 Structure of Report
This report summarizes the findings, conclusions, and recommendations of the JCAIT.
Section 2 presents background information on the facility and the HC1 storage and filling
operations. Section 3 describes the incident, including the chronology of events, the
consequences of the failure of the tank, and the emergency response. Section 4 describes the
investigation and the technical and causal analyses of the facts. Section 5 describes the JCAIT's
conclusions about the immediate cause, the root causes, and the contributing factors that led to
the incident. Section 6 summarizes the JCAIT's recommendations to Surpass, government, and
industry for chemical accident prevention, and Section 7 covers other problem areas identified in
the course of the investigation of the HC1 tank failure.
-------
2.0 Background
2.1 Facility Information
Surpass is located at 1254 Broadway in Albany, New York, about 1.75 miles northeast of
the downtown area. The company manufactures pool chemicals and repackages chemicals and
detergents. The Broadway facility is located in a light business area at the edge of a residential
neighborhood.
2.2 Process Information
At the Broadway facility, Surpass repackaged 31% HC1 onsite in a bottling operation into
one-gallon bottles for sale as a treatment for swimming pools. In the spring, Surpass typically
started receiving HC1 shipments more frequently to meet demands for the swim season. Based on
production reports for April, 1997, Surpass repackaged up to 12,000 gallons of HC1 April 1
through April 7.
Based on purchase order records for 1995 through early 1997, Surpass received tank
truck deliveries of HC1 at an average rate of one to two shipments per month, with some
variability due to seasonal demand. During the same period, shipments generally ranged from
4,600 gallons to 5,200 gallons (nominally 5,000-gallon orders) and were ordered from either of
two suppliers, Reagent Chemical and Research, Inc. (Reagent), Middlesex, New Jersey, or PVS
Chemicals, Inc.(PVS), Buffalo, New York. In April, Surpass had received two deliveries prior to
the day of the incident- 5,060 gallons on April 2 from Reagent and 4,600 gallons on April 4 from
PVS. Reagent was making a delivery of 4,950 gallons on the day of the incident, April 8.
(a) HC1 Storage Tank
A 5,700-gallon (working capacity) fiberglass reinforced plastic (FRP)1 atmospheric
pressure storage tank was used for the bulk storage of 31% HC1 at ambient temperature. The
HC1 storage tank was ll/2 feet in diameter and 18 feet high. The tank was manufactured by
Owens-Corning (model 86 MACS) and purchased by Surpass in 1978. As originally designed,
the top of the HC1 tank had two 3-inch diameter nozzles, a 2-inch diameter nozzle, and a 22-inch
diameter manway. A 3-inch diameter nozzle was installed on the side of the tank, about 7 inches
from the bottom of the tank. The manufacturer's design specifications included a caution that, if
the tank was to be air loaded, it had to be vented with a minimum 22-inch diameter opening
during the filling period. The manufacturer also specified that the tank pressure was not to
1 Reinforced plastics are composites in which a resin (in this case, a phenolic resin) is combined with a
reinforcing agent (in this case, glass fiber) to improve one or more properties of the plastic matrix. FRP combines
the corrosion resistance of plastic with the strength of glass fiber. FRP tanks are widely used to store corrosive
materials (Lees, 1996).
-------
exceed 10 inches of water, equivalent to about 0.4 pounds per square inch gauge (psig).2
The tank was put into service in 1979 or 1980 and used for HC1 storage until about 1985.
It was originally located on the west side of the building and elevated about 2 to 3 feet above
grade, supported by a steel stand, to permit gravity discharge of its contents. The HC1 tank was
splash filled from the top through the 2-inch diameter nozzle. One nozzle on the top was fitted
with a vacuum breaker and the other was not in use. The tank did not have any gauge for
measuring volume; the method of measuring the liquid level at that time is not known by the
JCAIT. At that time, the tank had no controls for the HC1 vapors. Surpass reported that the
manway was loose-bolted and HC1 fumes could escape through the manway. In 1985, the tank
was taken out of service because fumes escaping from the tank were irritating to those downwind
of the tank, and there was corrosion around the manway.
In 1988, Surpass contracted with Empire Fiberglass Products, Inc., Little Falls, New
York, to make repairs, seal the manway closed (in anticipation of adding a system to control
fuming), and add a 2-inch diameter nozzle in the side wall about 2 inches from the bottom of this
HC1 tank (in anticipation of adding a gauge). In 1989, the tank was placed back in service and
installed on the southeastern side of the building, within a newly built secondary containment area.
(See Figure 1 for a schematic of the storage tank area.) The building provided two walls of
containment. A dike, 4 blocks high and reinforced with steel bar, provided the other two walls.
The HC1 tank was elevated 8 feet above ground level, supported by a steel platform, to permit
gravity discharge of its contents. Nearby, in a separate diked area, three other tanks were used
for bulk storage. At the time of the incident, two of these were used to store 13% sodium
hypochlorite (NaOCl). Further information on the working capacities or the inventories of these
tanks was not collected by the JCAIT.
At the time of re-installation, Surpass made two additions to the HC1 tank, a scrubber
system and a pressure gauge adapted to indicate volume. The scrubber system was intended to
reduce the quantity of HC1 fumes escaping into the environment. Acid vapors generated during
the filling operation were vented through two lines— 2 and 3 inches in diameter— that intersected
at a tee and continued as a single 3-inch diameter vent line. This vent line ran from above the top
of the tank and extended vertically below the HC1 tank into a scrubber tank. (See Figure 2 for a
schematic of the HC1 tank.) The end of the vent line was fitted with a diffuser section consisting
of a connection, a 90° elbow, and an 18-inch length of 3-inch diameter plastic pipe which had
been drilled with 36 holes, each %-inch in diameter, and fitted with an end cap with 3 holes drilled
into it (a total of 39 holes). (See Figure 3 for a schematic of the diffuser section.)
The diffuser sat in a 50-gallon, loosely-covered plastic drum referred to as the scrubber
tank. (The scrubber system is shown in Figure 2.) Initially, the tank was filled with sodium
2This design pressure is consistent with the design pressures commonly found for atmospheric tanks; for
example, "[m]ost storage tanks are designed to withstand a gauge pressure of only 8 inches of water (0.3 psi) and
will burst at about three times this pressure." (Kletz, p 91.)
-------
containment
wall
exhaust
vent
FRP Tank containing
31%HCI
PRODUCTION
AREA
•Tanks containing 13%
sodium hypochlorite
OFFICE
AREA
jr, ~
,,Frotit,' ;
Entrance
CURB-
BROADWAY
Figure 1: Layout of Tank Area
(Topview)
-------
c
g
'•4—*
CO
CD
Q.
O
D)
C
T3
(D
c
0)
Q.
CD
o:
(Si
CD
-------
Connect to 3" Acid
Vent Line
3" PP Quick
Connect
3" Cap
90° Elbow
Figure 3: Diffuser Section of Vent System
(not to scale)
-------
carbonate (Na2CO3) for the intended purpose of neutralizing the vented HC1 acid vapors. The
neutralization reaction between the NajCC^ and HC1 was expected to form sodium chloride
(NaCl), carbon dioxide (CO2), and water. To attain a lower freezing point in the scrubber
solution, the sodium carbonate was replaced with sodium hydroxide (NaOH). A similar
neutralization reaction between NaOH and HC1 was expected to produce NaCl and water. Other
chemical reactions in such a scrubber system are also possible. For example, carbon dioxide
(CO2) from air may react with NaOH to form Na2CO3.
At the time of the incident, a NaOH caustic solution was being used in the scrubber tank.
Based on interviews, the scrubber system had been last disassembled and emptied in November,
1996. At that time, Surpass reports that the scrubber solution was replaced with 15 to 20 gallons
of 18% concentration NaOH.
There were no written standard operating procedures for the maintenance of the scrubber
system. Maintenance required the periodic removal of NaCl, a by-product of the reaction
between the HC1 and NaOH, as well as monitoring of the pH of the solution to maintain basic
(high pH) conditions. The general procedure for monitoring the pH of the scrubber solution was
to test the solution using litmus paper following each acid delivery. If the pH was found to be
below 9, one-quart bottles of either 50% or 18% NaOH were added to raise the pH. No written
records of the pH monitoring were kept by Surpass. No written records were kept of the caustic
additions, and it is not known how many, if any, additions were made between November 1996
and April 1997.
Surpass had little documentation on the design of the venting and scrubber system.
According to interviews, the vent line was sized using a rule of thumb that the area of the
discharge (outlet) vent should be at least twice the area of the inlet vent.
At the time that the HC1 tank was re-installed, a pressure gauge also was installed on the
HC1 tank for the purpose of measuring the liquid level in the tank. The gauge was installed on the
2-inch diameter line near the bottom of the tank and was protected from corrosion by a diaphragm
system. The pressure gauge measured the pressure head of liquid above the tank bottom, using a
scale reportedly ranging from 0 to 15 psig. Surpass performed theoretical calculations relating the
pressure head to the height of liquid in the tank and the density of HC1 to develop a template
displaying volume in gallons that was overlaid on the dial face. The scale ranged from 0 to 6,120
gallons and was marked off in 360 gallon increments. Surpass made a final calibration of the
gauge with the first HC1 delivery. Surpass believed the gauge to accurately reflect delivery
amounts by plus or minus 100 gallons. Over time, the gauge was not recalibrated, as the volume
readings were generally in agreement with the expected quantities of the deliveries.
To supply the bottling operation, HC1 was gravity-fed from the HC1 tank to a float tank in
the production area that served as a reservoir for the bottling operation. The HC1 tank was
equipped with an air inlet check valve to allow air into the tank as it was emptied and thereby
-------
prevent a vacuum. The original design specified that the vacuum should not exceed 4 inches of
water.
(b) Off-loading Operation
A standard HC1 tanker delivered the HC1 shipment to Surpass. The working capacity of
the cargo tank was reported to be up to 52,000 pounds, equivalent to 5,380 gallons. The bill of
lading for the April 8 delivery showed that the truck contained 47,840 pounds, about 4,950
gallons of 31% HC1. To ensure that the truck is empty at the end of the off-loading operation, the
cargo tank is designed with a 4-inch diameter dip tube that goes down into a sump in the bottom,
rear of the cargo tank.
The tank truck used air pressure to unload the cargo tank. The use of air to off-load HC1
is relatively common; one chemical supplier estimated that air off-loading is used at about 90% of
its customer facilities. The cargo tank was designed for a maximum allowable working pressure
of 35 psig and equipped with a pressure relief valve set at 32 psig.3 The truck was equipped with
a compressor to pressurize the cargo tank. An air hose was used to connect the compressor to
the air line, which was connected to the trailer tank. The air line was equipped with a pressure
gauge to measure pressure on the cargo tank.
To make a delivery, the truck backed into an area on the northwest side of the building. A
2-inch diameter flexible hose was used to hook up the product discharge valve on the truck to the
facility's hook-up flange for the fill line. The 2-inch diameter fill line ran vertically to the rooftop
and across the roof to the top of the HC1 storage tank. (Figure 2 shows the delivery set-up.)
Surpass had no written operating procedure for the off-loading of the HC1 to this storage
tank. By tradition, the procedure was for the facility operator, known as the unloading
supervisor, to check that the HC1 tank was empty and ensure that all discharge valves on the HC1
tank were closed. The unloading supervisor would show the truck driver the correct hook-up
flange and instruct the driver to use between 20 and 25 psig of pressure to off-load. Once the
transfer began, the unloading supervisor would visually check the scrubber system for percolation,
an indication to him that air was flowing though the diffuser. The unloading supervisor would
periodically check the HC1 tank gauge, which was calibrated for volume, and monitor the off-
loading procedure. When the gauge read "5,040 gallons," the unloading supervisor would
instruct the driver to shut off the compressor. The bleed-off pressure from the cargo tank would
be used to push the remainder of HC1 from the truck to the HC1 storage tank. As the remaining
liquid HC1 was pushed out and replaced with air, the hose would surge or "kick," indicating that
3The U.S. Department of Transportation (DOT) regulates the transportation of hazardous materials,
including the specifications for design and construction of HC1 cargo tanks. Examples of these design
specifications include requirements for maximum allowable working pressure; material and thickness of material;
pumps, piping, hoses, and connections; and pressure relief. Title 49 of the Code of Federal Regulations (CFR)
details the requirements for hazardous materials transportation.
-------
all of the liquid had been transferred. The unloading supervisor relied solely on this hose kick as
an indication that the liquid delivery was completed.
Reagent had a written standard operating procedure (SOP) for driver unloading. This
procedure included having the customer identify the correct hook-up flange; hooking up the
flexible hose from the cargo tank to the hook-up flange; opening the facility's product valve;
pressurizing the cargo tank to about 10 psig; and opening the product discharge valve. During
the unloading, the driver is expected to monitor the tank pressure using the gauge. The SOP
warns that tank pressure is not to exceed 30 psig at any time during transfer. When the tanker is
empty, the driver is expected to ensure that the hose is clear of material. The driver is also
expected to check with the facility as to the procedure for bleeding off the pressure from the
tanker.
2.3 Chemical Information
The chemicals involved in the April 8th release were hydrochloric acid (HC1) in aqueous
solution, sodium hypochlorite (NaOCl), also in aqueous solution, and chlorine, generated by the
reaction between HC1 and NaOCl. Information on each of these substances is presented below.
Hydrochloric acid
Aqueous HC1 is a solution of hydrogen chloride (a gas under ambient conditions).
Aqueous HC1 is a strong acid. It is corrosive and can cause severe eye and skin burns. Hydrogen
chloride fumes can be released from aqueous HC1; the amount of fuming depends on the
concentration of the solution and conditions such as temperature. The fumes are irritating to the
skin, eyes, and respiratory system.
HC1 is a versatile chemical that has a number of different industrial uses, including
production of chlorides, ore refining, as a laboratory reagent, as a catalyst in chemical production,
and etching and cleaning metals.
The most generally shipped solutions of HC1 are 20 degrees Baume' (°Be')4, equivalent to
31.45% HC1; 22°Be' (35.21% HC1) and 23 °Be' (37.14% HC1) (Chlorine Institute, 1996). The
solution shipped to Surpass for repackaging was 20 °Be". The density of 20 °Be" HC1 is
approximately 9.671 pounds per gallon at 60°F.
Aqueous HC1 is reactive with a number of substances. It reacts with most metals to
4The Baume' hydrometer scale is a calibration scale for indicating the specific gravity at 60 °F (15.6 °C) of
some liquids in commerce. Baume' is abbreviated as Be', and the reading on the scale is degrees Be' (°Be'). For
liquids heavier than water, 0 °Be' corresponds to a specific gravity of 1.000 (i.e., the density is equal to the density
of water). Specific gravity is calculated as 1457(145 - °Be') at 15.6°C. 20 °Be' corresponds to a specific gravity of
1.16 (Handbook of Chemistry and Physics, 1989).
-------
release flammable hydrogen gas, and it reacts with strong oxidizers to release toxic chlorine.
Sodium hypochlorite
NaOCl is a solid in pure form, but is not very stable as a solid; it is generally produced and
used in water solution. Aqueous solutions of NaOCl are used as bleach or disinfectant. The
aqueous NaOCl solution stored onsite at Surpass was 13.25% concentration.
Aqueous solutions of NaOCl are fairly stable, but are subject to some decomposition,
depending on factors such as concentration, pH, temperature, light, and impurities. The major
decomposition products are oxygen and chlorate ion (C1O3"). If NaOCl is mixed with acid,
hypochlorous acid (HOC1) is formed. HOC1 is much less stable than NaOCl and will undergo
decomposition reactions forming oxygen, chloric acid (HC1O3), and chlorine. Decomposition to
chlorine involves a reversible reaction between HOC1 and HC1 (an intermediate decomposition
product). If HOC1 is mixed with large amounts of HC1, the reaction will proceed primarily in the
direction of chlorine formation, and chlorine will be generated (Kirk-Othmer, 1993).
NaOCl is a strong oxidizer. NaOCl solutions are corrosive, and exposure to solutions can
cause irritation to the eyes, mucous membranes, and skin.
Chlorine
Chlorine, which was produced in the reaction between HC1 and NaOCl, is greenish-yellow
gas with a suffocating odor. It is poisonous and corrosive. Exposure to relatively low
concentrations may cause stinging or burning of the eyes, nose, throat, and chest. Exposure to
high concentrations can result in death.
10
-------
3.0 Description of the Incident
3.1 Chronology of Events
+ 1978 HC1 tank was purchased by Surpass.
+ 1979 (estimated) HC1 tank was installed at original location.
+ 1985 (estimated) HC1 tank was taken out of service.
1988
Spring 1989
+ November 1996
>Nov. 1996 to
April 1997
+ April 7 1997
6:30 a.m.
(time not known)
(time not known)
4:50 p.m.
HC1 tank was repaired and modified under contract to Empire
Fiberglass Products, Inc. in order that Surpass could place it back
into service. At this time, the 22-inch diameter manway was
permanently bolted closed.
The HC1 tank was installed in the spring. A pressure gauge,
modified to read volume by affixing a template on the dial, was
installed on HC1 tank for the purpose of monitoring the liquid level
in the tank. (See Section 2 of this report for additional details.)
Additionally, a scrubber system was added to reduce HC1 fumes.
Based on interviews, the scrubber system was cleaned out in
November and the solution replaced with 15 to 20 gallons of 18%
NaOH.
5,000-gallon deliveries of 31% HC1 were received by Surpass at a
rate of 1-2 deliveries per month.
Started bottling HC1 from the storage tank. Time based on
production report.
The bottling operator drew off HC1 from the lowest nozzle on the
storage tank until no more product would gravity feed. The
volume gauge also read zero.
To continue with the bottling operation, Surpass ordered a 5,000-
gallon shipment of 31% HC1 from Reagent for delivery the next
day.
The Reagent tanker was loaded at Standard Chlorine of Delaware,
Inc. The bill of lading stated that 47,840 pounds of 31% HC1 were
11
-------
loaded. Time is based on bill of lading.
April 8 1997
7:15 a.m.
7:30 a.m.
(time not known)
7:40 a.m.
(time not known)
7:50 a.m.
8:55 a.m.
The tank truck arrived at Surpass to deliver 47, 840 pounds
(equivalent to 4,950 gallons). Time based on witness interview.
This was the truck driver's first delivery to Surpass. The driver
asked about unload air pressure. The unloading supervisor stated
that unload pressure should be 20 to 25 psig. Unloading supervisor
also told driver that it would take approximately 1 /^ hours to
unload, including hooking up and disconnecting the product hose.
Time is estimated.
The unloading supervisor reportedly walked to the HC1 tank and
checked that the discharge valves on the HC1 tank were closed.
The unloading supervisor reportedly walked back to the truck.
Tanker started off-loading HC1 to the storage tank using air
compressor. Time based on witness interview.
The unloading supervisor reported that he checked scrubber tank
and observed percolating in the scrubber, an indication to him that
the vent line and diffuser were open and operating.
The unloading supervisor told bottling operator that he could start
drawing off HC1. Per the bottling production report, the bottling
operator began drawing off HC1 from the HC1 storage tank to the
reservoir for the bottling operation. The bottling operation
continued until the HC1 tank failed. Time is based on production
report.
The unloading supervisor noted that the volume gauge read 5,040
gallons and reported this to the driver. According to the truck
driver, the compressor for the tank truck was turned off about 1
hour and 15 minutes into the delivery. At the time that the
compressor was stopped, the pressure gauge on the cargo tank was
reported to have read 20 psig. Both the unloading supervisor and
the driver reported picking up the product hose line after the
compressor was turned off and that the hose felt heavy, indicating
to each of them that liquid was still in the line. Within 1-2 minutes,
it was reported that the pressure in the cargo tank of the truck
drops to about 18 psig. Time is estimated based on witness
12
-------
recollections of how long after these events the rupture occurred.
8:59 a.m. Approximate time of vessel rupture.
The unloading supervisor instructed the truck driver to shut the
unloading valve on the tank truck. Truck driver reported that the
pressure gauge on the cargo tank read 16 psig after the rupture.
9:01 a.m. First of several emergency 911 phone calls was placed.
3.2 Consequences of the Incident
The HC1 tank head separated at a point about 5 feet from the top of the tank and flew
about 15 feet west to the roof of an adjacent building. Figure 4 shows the separated tank after
the incident, and Figure 5 shows the top of the tank. The bottom of tank failed in the knuckle
area where the cylindrical part of the HC1 tank meets the flat bottom, as shown in Figure 6. The
tank bottom remained on the platform. The cylindrical part of the fiberglass shell began to unwind
itself at the top edge (see Figures 4 and 5). The shell also remained on the platform but
collapsed, leaning toward the west building wall (see Figure 6). The scrubber tank was not
affected by the tank failure; it was found intact in its original position after the incident, as shown
in Figure 7.
At the time of rupture, the HC1 tank contained about 4,800 gallons of material. The
failure of the HC1 tank caused a sudden surge of liquid over the secondary containment wall. The
force of the liquid also caused a break in the masonry of the secondary containment wall. Surpass
has estimated that as a result of this release, about 150 gallons were absorbed by soil within the
property boundary and about 2,300 gallons entered storm drains located on Broadway. The
storm drains emptied into an underground stream, the Patroon Creek, a tributary of the Hudson
River. Surpass also estimated that 1,900 gallons were contained within the secondary
containment dike and that 400 gallons entered a nearby building through a window in the north
wall and through an exhaust fan in the west wall.
During the event, a 2-inch diameter NaOCl line that was located in proximity to the HC1
tank was broken, and an estimated 200 gallons of NaOCl was released into the secondary
containment. The JCAIT believes that a reaction occurred between the two chemicals, causing
the generation of chlorine gas. Witnesses described seeing greenish-yellow fumes (assumed to be
chlorine) drifting offsite, as well as liquid material running offsite and along the street curb to the
storm drains.
As a consequence of the incident, 8 workers and 32 members of the public were taken to
the hospital, treated, and released. A ten-block area, including nearby businesses and residences,
was evacuated.
13
-------
14
-------
6: of of
HCI
-. \ \
Failure Along
I
Mr X- ^
7: of Top of HC! Tank
Cover
15
-------
3.3 Description of the Emergency Response
The JCAIT did not collect in-depth details about or evaluate the emergency response
actions, but a brief overview is provided here.
Employees from Surpass's Broadway facility, where the incident occurred, and another
Surpass facility on Bridge Street, Albany, New York, responded to the incident. The Albany Fire
Department and other local and state officials responded to the emergency. The local sewer
authority tested the pH at the confluence of the stream and the Hudson River.
Federal officials, including representatives from OSHA's Albany, New York, Area Office
and U.S. EPA Region 2, responded by afternoon on the day of the incident. The OSHA
compliance officers entered areas of the facility where the HC1 had been released to collect
samples and gather preliminary information from the employees and managers. The EPA on-scene
coordinator also entered areas of the facility where chemicals had been released to assess the
extent of the release, to take photographs, and to monitor the response and recovery activities
performed by contractors hired by Surpass.
4.0 Analysis and Significant Facts
On April 21, 1997, the JCAIT formally met with the OSHA compliance team to begin
collaborating on the collection of evidence, the formal request for documents, interviews of
employees and managers, and other field work. Additionally, the JCAIT arranged for a
demonstration of the off-loading procedures by the chemical supplier as part of the field work to
support the investigation.
At this preliminary stage of the investigation, the main failure scenarios considered by the
JCAIT were (1) overfilling with liquid; (2) overpressurization due to a blockage in the vent line or
diffuser; and (3) overpressurization due to undersizing of the vent to handle the pressure bleed-
down of the tanker.
A material balance based on company records was consistent with testimony that the HC1
tank was essentially empty prior to the delivery and the empty tank had the volume capacity to
receive the delivery. Additionally, the failure mode of the tank, the force associated with the
damage, and witness accounts are consistent with a pneumatic failure. Thus, the JCAIT focused
on overpressurization of the HC1 tank and the role of the venting/scrubber system in the event.
The JCAIT did not consider material failure of the HC1 tank given the circumstances of the
incident and the force associated with the failure mode.
16
-------
4.1 Significant Facts
The facts considered by the JCAIT in determining the causes of the incident are listed
below:
The HC1 tank truck that arrived at Surpass to make a delivery had a shipment of 4,950
gallons. The bill of lading for the delivery stated that 47,840 pounds of 31% HC1,
equivalent to 4,950 gallons, were loaded onto the tank truck on April 7.
On the morning of the incident, before the off-loading of the tank truck began, the HC1
storage tank was empty, as indicated by the following:
The inventory at the end of March indicated that the HC1 tank contained
approximately 2,300 gallons. During April, prior to the day of the incident,
Surpass received 9,660 gallons of HC1 and bottled 11,680 gallons. The
accumulation in the storage tank based on these values is 280 gallons. Within the
accuracy of the gauge readings, this would indicate that the tank was empty.
The bottling operator reported drawing down the tank on April 7 to empty it by
opening both the 3-inch diameter and the 2-inch diameter discharge lines located
near the bottom of the tank.
The HC1 tank had a working capacity of 5,700 gallons.
The heel in the empty HC1 tank has been calculated by Surpass to be in the range of 75 to
100 gallons.5
According to the bottling production report, 288 gallons of HC1 were bottled on the
morning of the incident, during the period of time between the beginning of the delivery
and the HC1 tank rupture.
Prior to the rupture, pressurized air was entering the HC1 tank from the pressure bleed-off
of the tanker, as indicated by the following:
One witness reported that after the incident, while the truck was still at the site, he
opened the manway on the top of the truck to look inside and observed that the
cargo tank was "bone dry." Reagent also reported that the truck was empty. This
indicates that the entire delivery of 4,950 gallons was transferred out of the tanker
and into the HC1 storage tank.
5The heel is the amount of residual that cannot be withdrawn from the bottom of the tank by normal
emptying procedures. The estimate of the volume of the heel is based on the cross-sectional area of the HC1 tank
and the height of the lowest product discharge nozzle.
17
-------
The vertical hard piping from the tank truck hook-up to the storage tank was
inspected by OSHA after the incident and found to be essentially empty, indicating
that the last material through the hard piping was air and further supporting the
finding that all of the HC1 had been transferred into the storage tank prior to the
rupture.
The pressure gauge on the tanker was reading accurately. The tank truck was inspected
by New York State Department of Transportation (DOT) on April 9, 1997, and OSHA
verified that the reading on the truck's pressure gauge was reading accurately.
The HC1 tank ruptured into three pieces, as described in Section 3.2.
The manway on the tank top was sealed shut, and the only way for vapor to escape from
the tank was through the vent line. (See Figure 7.)
The scrubber tank was not affected by the tank failure; it was found intact in its original
position after the incident. (See Figure 8.)
After the incident, the diffuser in the scrubbing tank was removed by disconnecting the
vent line quick connect and removing the lid from the scrubber tank. The holes in the
diffuser were found to be clogged with a white crystalline substance, as shown in
Figure 9. This substance was sodium chloride, according to laboratory analysis.
A layer of white crystalline material was also found in the bottom of the scrubber tank.
This substance also was sodium chloride, according to laboratory analysis.
18
-------
19
-------
4.2 Analysis
Based on visual observations of the fragmentation of the HC1 tank and consideration of
the force that would be required to cause the observed damage, the JCAIT determined that the
tank failure was due to overpressurization with compressed air rather than to overfilling with
liquid. Failure from overpressurization involves a higher energy release than failure from
overfilling with liquid, and the damage resulting from the failure of the tank is consistent with
higher energy release. In addition, the tank was empty before the delivery of the HC1 began and
had sufficient capacity to contain the delivery, therefore, overfilling with liquid is unlikely. Given
the circumstances of the incident, the JCAIT believes that the failure of the HC1 tank was not due
to age, wear, or defective materials.
4.2 (a) Venting System Calculations
As discussed above, the JCAIT decided to focus on the overpressurization of the HC1 tank
and the role of the venting/scrubber system in the event. The JCAIT believed that
overpressurization was due to either blockage of the diffuser openings with NaCl or to
undersizing of the vent. In order to predict whether blockage in the diffuser or undersizing of the
vent was the more likely cause of the overpressurization, the JCAIT's contractor developed a
profile describing the change in pressure in the space above the liquid in the storage tank during
the HC1 delivery operation. Based on the analyses, the JCAIT found that the configuration of the
vent/scrubber system, including the sealing of the manway, led to the operation of the HC1 tank
above the manufacturer's design specifications during the normal air off-loading of deliveries. The
fouling of the diffuser over time led to the further increase in tank pressure and ultimately to the
failure of the tank. The analysis is summarized here; the consultant's report, describing the
analysis in detail, appears in Appendix D.
General Description
In general, when liquids are transferred into atmospheric storage tanks fitted with an open
vent, the volume of the head space above the liquid level is reduced, increasing the tank pressure
momentarily. The increased pressure causes the displacement or flow of vapor from the storage
tank to the atmosphere in order to equalize the tank pressure with atmospheric pressure. The
liquid fill rate and the vapor flow rate must be equal to ensure that negligible tank pressure builds
over time.
In this case, however, the storage tank was not fitted with an open vent to the atmosphere.
Instead, the vent line ran to a diffuser that was submerged in an NaOH solution in a scrubber tank.
The hydrostatic pressure of the solution in the scrubber tank created a backpressure, which
prevented displaced vapors from flowing through the diffuser until the tank pressure exceeded the
hydrostatic head. For a period of time while liquid was being transferred, pressure built up in the
HC1 tank. Once the tank pressure exceeded the hydrostatic head, vapor flow out of the diffuser
began. The flow rate of vapor was a function of the tank pressure, the backpressure from the
20
-------
scrubber solution, and any pressure losses (for example, friction losses in piping or resistance in
fittings).
The rate of liquid flow into the tank was a function of the pressure in the cargo tank of the
truck, the differential pressure head created by pumping liquid from the truck to the top of the
HC1 tank, the backpressure created by the pressure within the HC1 tank, and the dimensions
(length and inner diameter) of the fill line and hose.
At the end of the transfer, as the cargo tank was emptied of liquid, a point was reached at
which pressurized air from the cargo tank of the truck flowed into the HC1 storage tank.
Modeling
A computer model was created to analyze the pressure profile of the HC1 tank during the
off-loading operation and to evaluate the effects of the HC1 tank's design features on the pressure
within the tank. (See Appendix D for details.) The analysis of the pressure in the HC1 tank is
based on an unsteady state mass balance calculation routine.
To bracket the potential peak pressure in the HC1 tank during the off-loading, two
scenarios were modeled:
(1) Unrestricted flow, assuming no fouling of the %-inch diameter diffuser openings; and
(2) Restricted flow, assuming that all of the diffuser openings are reduced to a 1A- inch
diameter because of fouling (about 84% reduction in the cross-sectional area).
The HC1 tank pressure as a function of time, calculated by the model, is presented in
Figure 10 (for unrestricted flow) and in Figure 11 (for restricted flow).
The model assumes that the HC1 storage tank was at atmospheric pressure prior to the off-
loading of the HC1. The submersion of the diffuser in the scrubber solution resulted in a
backpressure equal to the hydrostatic pressure of the solution, estimated to be 0.6 psig. It was
assumed that no vapor flow occurred out of the diffuser until the pressure within the HC1 storage
tank exceeded this superimposed backpressure.
As the off-loading began, delivery of the liquid into the HC1 tank pressure decreased the
tank's void volume. The combined effect of no vapor flow out through the diffuser and the
decreased void volume was to increase the HC1 tank pressure. The model predicts that the HC1
tank pressure increased until it exceeded 0.6 psig. Note that this is above the design pressure of
0.4 psig for the tank. Once the backpressure was exceeded, vapor flow out of the diffuser would
have begun. This point is marked as point A on Figures 10 and 11. The rate of liquid flow into
the HC1 tank determined how quickly the tank pressure rose from atmospheric to the predicted
value of 0.6 psig. For both scenarios, the rate of liquid flow into the HC1 tank was the same since
21
-------
the parameters of the off-loading are fixed by the known facts and circumstances of the off-
loading operation on April 8, namely, that 4,950 gallons of HC1 were delivered; the HC1 delivery
took approximately 80 minutes; the fill line and the flexible hose were 2 inches in diameter
(nominal); and that the pressure on the cargo tank of the truck was up to 25 psig. (These
parameters have been previously discussed in sections 2.2(a) and 3.1 of this report.)
Assuming that both the cargo tank pressure and the diffuser backpressure remain relatively
constant, the pressure in the HC1 tank remains constant at a predicted value of 0.6 psig
throughout the delivery of liquid.
Throughout the delivery of the liquid to the HC1 tank, the pressure of the liquid discharged
to the top of the HC1 tank is reduced below the pressure of the cargo tank by both the line
pressure drop and, more significantly, the change in the liquid head. At the end of the delivery, as
all the liquid in the cargo tank is evacuated, the effect of the change in the liquid head is quickly
eliminated as the liquid in the line is evacuated and displaced with vapor. At this point (labeled B
on Figures 10 and 11), the HC1 tank pressure is predicted to increase rapidly as the pressure
within the cargo tank is relieved into the HC1 tank. The net pressure in the tank is a function of
the flow of pressurized air into the tank and the rate of vapor flow out of the diffuser.
The rate of vapor flow out of the tank through the diffuser was modeled as a function of
tank pressure and the diffuser backpressure and represented by flow calculations for compressible
fluids through an orifice. The pressure drop associated with the flow of the vapor through the 3-
inch diameter vent line would have further restricted flow, however, this factor was considered
negligible for the purposes of this modeling. The difference in the two scenarios becomes evident
in this portion of the pressure profile because the assumed available flow area through the diffuser
differs. The peak pressure in the HC1 tank is dependent on the flow area of the diffuser.
For the unrestricted scenario, of unobstructed flow through the diffuser, the HC1 tank
pressure is predicted to peak at 3.4 psig. For the restricted flow scenario, assuming fouling of the
diffuser openings, the pressure could have peaked as high as 12 psig. These peaks are
represented as point C on Figures 10 and 11.
4.2(b) Tank Failure Pressure
The exact pressure that caused the tank failure was not estimated. Because FRP
composite structures are not homogenous; the design and manufacture of tanks varies with the
manufacturer; and the original design calculations were not available, the exact pressure at which
the tank would have failed cannot be readily predicted from the known facts. The modeling of the
pressure profile in the HC1 tank predicted a peak pressure of 3.4 psig during the off-loading,
under normal operation of the scrubber. Although this predicted value is significantly above the
design specifications of the HC1 tank, the prior use of the tank in this service indicates that it did
not exceed the yield point for the tank. Additionally, the diffuser was found to be plugged,
potentially raising the pressure to 12 psig, well beyond the normal operation peak.
22
-------
I
CO
CO
CD
co
0
Q.
"co
3
CL
CO
D)
s-l
(D O
Q- CD
^ CO
2> O
CO _
i- O)
T3 Z
0 . .
0 .2
-§ fe
------ CD
CD
o
CO
- «
t — CO
< CD O
Bisd
>|uei
IOH P9ie|no|eo
D) ^
CO (S
«£
23
-------
CO
cn
co Q.
s
o> o
u. to
-o °
is
8
CO
•2
(A
CO
Q.
H
"c
(D
"o
CL
• O
• CO
s
CD
O
in
o
CO
M—
O XL
§1
IS
CD T
Q- o g
-y I «
° «
0 CO
°
CO if C
CO -t= 3
0 r- CO
fe-l S
-^ >t o
•*- lu CD
Q) T3 CD
D) 0) Q-
CD N —
o "^ •—
•EO
>, •
CD
< CD O
6isd '
IOH
CD
3
24
-------
5.0 Conclusions
For the purpose of maximizing the lessons learned, the JCAIT considered both the root
causes and contributing factors in developing the recommendations. Root causes as defined in the
EPA/OSHA memorandum of understanding (MOU) are the underlying prime reasons, such as
failure of particular management systems, that allow faulty design, inadequate training, or
deficiencies in maintenance, which in turn lead to an unsafe act or condition and result in an
incident. Contributing factors are reasons that, by themselves, do not lead to the conditions that
ultimately caused the event; however, these factors facilitated the occurrence of the event.
The JCAIT developed an events and causal factors chart (that is the basis of the
chronology presented in Section 2) and used a root cause tree approach that covered both the
equipment and human performance root causes. This type of methodology provides a standard
set of root causes for investigators to evaluate and provides for a consistent and methodical
approach to be used by all the investigators.
5.1 Causes
The configuration of the vent/scrubber system, including the sealing of the manway, led to
the operation of the HC1 tank above the manufacturer's design specifications during the normal air
off-loading of deliveries. The fouling of the diffuser over time led to the further increase in tank
pressure and ultimately to the failure of the tank.
5.2 Root Causes and Contributing Factors
The JCAIT concludes that the root causes of the incident are:
• Modifications (the sealing of the manway and the addition of the scrubber system)
to the venting of the HC1 tank were not within the tank manufacturer's
specifications for venting.
These modifications provided inadequate venting for the air off-loading according
to the tank manufacturer's original design specifications, and eliminated any
emergency relief of the vessel in the event that it was overpressurized.
• No hazard analysis of the modifications to the venting of the HC1 tank was
performed.
Surpass did not review the design of the modifications for the venting of the tank
(sealing the manway and adding the scrubber) to assess whether these changes
would lead to an overpressurization. An evaluation of the changes in the design of
the HC1 tank using tools such as management of change (MOC) would likely have
identified the hazard of overpressuring the tank during air off-loading of HC1
25
-------
deliveries under normal or expected conditions. The air pressure capability of the
tanker far exceeded the design pressure of the HC1 tank. A formal analysis of the
process hazards would have identified the need to ensure that the pressure from
the tanker was not directly delivered to the HC1 tank.
• Inadequate preventive maintenance of the scrubber system.
Inadequate maintenance permitted the diffuser section to become clogged with
solids, further reducing the scrubber's capacity to vent the pressure buildup in the
HC1 tank. Surpass had no written procedure for maintaining or inspecting the
scrubber system. Maintenance procedures would have been improved by
developing detailed procedures for testing and adjusting the scrubber solution--
including the frequency of tests, the parameters (pH, specific gravity, etc.) to be
measured, and the acceptable range of those parameters. Maintenance results
should be documented in order to provide a historical basis for revising the
procedures.
The JCAIT concludes that the contributing cause to the incident is:
• Lack of a written standard operating procedure (SOP) for air off-loading of
deliveries to the HC1 tank, including an inadequate method for determining that the
delivery was complete.
Surpass had no written procedure for off-loading material from the delivery truck
to the HC1 tank. While Surpass has procedures that have evolved over time based
on the experience of its employees, documenting those procedures in writing will
ensure that all employees perform similar tasks and procedures in a consistently
safe manner. Additionally, written procedures can be made available for ready
reference and can be used in the training of new employees.
The written procedure should include the elements of the non-written, traditional
procedure such as step-by-step descriptions of tasks, definitions of the safe
operating limits, and additional precautions. The SOP would be improved by
addressing certain elements in more detail including, the pressure bleed-off of the
cargo tank; checking the operation of the scrubber system; and the issue of
simultaneous filling and drawing off to the production area. As part of the SOP,
clear and definitive process displays must be used so that the operator can easily
recognize system errors.
Surpass's reliance on informal methods of determining that the HC1 delivery was
complete could permit errors by the operators . The operator relied on a pressure
gauge modified to read volume to monitor the end of the delivery. The device did
not readily permit the operator to detect a potential error, such as
26
-------
overpressurization of the tank and an inaccurate volume reading. The operator
also looked for the hose kick as an indicator that the liquid delivery was complete.
The hose kick is a transient occurrence (on the order of seconds) that could be
overlooked. Instruments or devices that give a clear, understandable indication to
the operator (for example, a sight glass) would reduce the possibilities of errors. A
written SOP for off-loading deliveries to the HC1 tank should include procedures
for determining when the delivery is complete.
6.0 Recommendations
Based on the root causes and contributing factors of the HC1 tank failure, the JCAIT has
developed the following recommendations:
Surpass and other facilities should ensure that modifications to their equipment, in this case
for the purposes of environmental control, do not create new hazards or compromise safety.
Before modifying equipment, Surpass and similar facilities should thoroughly review and approve
changes prior to implementation to ensure safe operation. Results of the review should be
documented One way to do this is by using formal management of change procedures for any
processes which involve the handling of hazardous materials.
Surpass and other facilities should maintain environmental control systems to ensure
continuous reliability and effective operation. Based on the system design, the known failure
history, and engineering judgement, Surpass and similar facilities should evaluate how long the
scrubber solution can be used before it needs to be replaced; develop detailed procedures for
inspecting, testing, and adjusting the scrubber solution— including the frequency of tests, the
parameters (pH, specific gravity, etc.) to be measured, and the acceptable range of those
parameters; and establish control mechanisms to ensure that preventive maintenance is performed
correctly. Maintenance procedures should be written. Maintenance results should be
documented in order to provide a historical basis for revising the procedures.
Surpass should develop written standard operating procedures (SOPs) related to the HCl off-
loading and maintenance of the scrubber system. SOPs should be written in simple and
understandable language, reviewed for safety issues, and validated for accuracy. Procedures
should include details of the task to be performed; the types and frequency of instrument readings
and samples to be taken; safety precautions; critical parameters and safe operating limits.
Additionally, human factors such as communication issues; operator/equipment interfaces for
displays; and adequate measuring devices should be incorporated along with the procedures to
reduce the chances of errors.
EPA and OSHA should develop an alert to raise awareness about the need for thorough
consideration of safety when designing equipment or processes for environmental control.
As part of their ongoing effort to prevent chemical accidents, EPA and OSHA jointly issue alerts
to increase awareness of potential hazards. In recent months, EPA and OSHA have investigated
27
-------
several accidents related to the design and/or operation of control devices for air pollution.
Because of these accidents, the agencies are considering developing an alert to highlight the need
to consider safety prior to implementing changes, such as addition of end of the pipe devices, to
ensure that the devices are designed, maintained, and operated safely and integrated with the rest
of the process to ensure that it is not adversely affected.
7.0 Other Findings
While investigating the HC1 tank failure, the JCAIT identified other potential problem
areas that may have contributed to the consequences of the incident. These issues are listed
below:
• Although the HC1 storage tank was located in a separate diked area from the NaOCl
storage tanks, the NaOCl storage tank discharge lines ran nearby to the HC1 tank,
contributing to the hazard created by the incident. Due to their proximity, the NaOCl lines
were broken when the HC1 tank ruptured. Incompatible substances (HC1 and NaOCl)
were mixed together when they were accidentally released, resulting in a reaction that
produced a hazardous substance (chlorine). The generation of chlorine added to the
hazard posed by the hydrogen chloride fumes that were generated from the spill of
aqueous HC1. Adequate separation distances for chemicals that are incompatible because
of reactivity are site-specific. Facilities should evaluate their site layout for potential
chemical incompatibilities. One way to do this is to perform a process hazard analysis and
an off-site consequence analysis (for example, dispersion modeling) to evaluate the
potential risks. The results of such analysis should be documented and specific actions
taken, such as relocating tanks or installing safety measures or barriers in situations where
there are incompatibility problems.
• The design of the secondary containment was not adequate to withstand the sudden surge
of liquid over the dike wall. Similar instances have been cited in the literature. For
example, Lees suggests that the tidal wave of liquid resulting from the catastrophic failure
of an FRP tank is capable of demolishing a dike wall, or, if the tank is indoors, a building
wall (Lees, 1996, p. 22/65).
• As part of a facility's mechanical integrity program, storage tanks should be periodically
inspected for parameters such as wall thickness, defects, surface hardness, and strain
measurement. The parameters to be tested, the type of testing, and the frequency schedule
should be determined as part of the facility's mechanical integrity program based on
known failure history, the manufacturer's recommendations, and engineering judgement.
In 1995, Owens-Corning sent a letter to all of its former customers recommending that
they have their FRP tanks inspected annually by qualified fiberglass chemical equipment
process experts. This type of information should have been incorporated by Surpass into a
mechanical integrity program.
28
-------
Surpass is a member of the National Association of Chemical Distributors (NACD).
NACD members have developed a program, called the Responsible Distribution Process,
which outlines guiding principles and elements to improve the safe handling of chemicals.
Commitment to the NACD Responsible Distribution Process is a condition of continued
membership. Although the JCAIT understood that Surpass had not yet completed its
program, the JCAIT found several deficiencies in Surpass's management system, such as
undocumented standard operating procedures and lack of process hazard analysis. A
timely and thorough implementation of the Responsible Distribution Process program by
Surpass may have uncovered these deficiencies and led to improvements in Surpass's
system to manage health, safety, and environmental concerns.
The Clean Air Act requires a periodic (every 5 year) review of the list of substances
covered under the Risk Management Program (RMP) Rule. Under a recent modification
to the list of regulated substances for the RMP Rule, only anhydrous hydrogen chloride
and HC1 solutions of 37% or greater will be covered (62 FR 45130, August 25, 1997). As
this incident demonstrates, solutions with HC1 concentrations below 37% may pose
potential hazards to human health or the environment. The circumstances of this incident
should be considered in any future evaluation of how to list HC1 solutions for the RMP
Rule.
29
-------
References
The Chlorine Institute, Inc. Pamphlet 150: Hydrochloric Acid Tank Motor Vehicle
Loading/Unloading. Edition 1. 1996.
CRC Handbook of Chemistry and Physics 1988-89, 69th edition. Boca-Raton: CRC Press. 1988.
Kirk-Othmer Encyclopedia of Chemical Technology, 4th Ed., vol 5, "Chlorine Oxygen Acids and
Salts; Dichlorine Monoxide, Hypochlorous Acid, and Hypochlorites," p. 932ff. New York: John
Wiley & Sons, 1993.
Kletz, T. What Went Wrong? Case Histories of Process Plant Disasters. 3rd Ed. Houston: Gulf
Publishing. 1994.
Lees, F. Loss Prevention in the Process Industries; Volume 2. 2nd ed. Oxford: Butterworth-
Heinemann. 1996.
30
-------
Appendix A
Joint Chemical Accident Investigation Team (JCAIT) members
OSHA personnel who participated in the accident investigation and development of the accident
report include:
Mike Marshall OSHA National Office
Kay Coffey OSHA Albany, NY Area Office
Margaret Rawson OSHA Albany, NY Area Office
EPA personnel who participated in the accident investigation and development of the accident report
include:
Breeda Reilly U. S. EPA Headquarters
Ellen Banner U. S. EPA Region II
Dilshad Perera U.S. EPA Region II
Mohan Hede U.S. EPA Region II
31
-------
Appendix B
Industry Codes
These codes are listed for informational purposes.
D 4097-88 Standard Specification for Contact-Molded Glass-Fiber-Reinforced Thermoset Resin
Chemical-Resistant Tanks, 1995 Annual Book of ASTM Standards, Section 8 Plastics, ASTM 1916
Race Street, Philadelphia, PA. This standard includes requirements for materials, properties,
design, construction, dimensions, tolerances, workmanship and appearance for atmospheric pressure
above-ground cylindrical tanks fabricated by contact molding.
D 3299-88 Standard Specification for Filament-Wound Glass-Fiber-Reinforced Thermoset Resin
Chemical-Resistant Tanks, 1995 Annual Book of ASTM Standards, Section 8 Plastics, ASTM 1916
Race Street, Philadelphia, PA. This standard includes requirements for materials, properties,
design, construction, dimensions, tolerances, workmanship and appearance for atmospheric pressure
above-ground cylindrical tanks fabricated by filament winding. This standard covers both tanks
vented directly to the atmosphere and to tanks vented into a fume conservation system.
Pamphlet 150: Hydrochloric acid tank motor vehicle loading/unloading; Edition 1; June 1996, The
Chlorine Institute. This code presents guidance for the safe transportation, handling, and receipt
of HC1 in tank motor vehicles.
32
-------
Appendix C
Other Accidents Involving Atmospheric Pressure FRP Tanks
Collapse of FRP Tank at Wastewater Treatment Facility
In May, 1995, at a government-owned, contractor-operated facility outside of
Cincinnati, Ohio, a 16,900-gallon fiberglass reinforced plastic tank failed and collapsed.
Personnel were preparing the tank for testing and water was being added to fill the tank to
94% capacity. There were no personnel injured nor environmental impacts. A large portion
of the waste water treatment system was damaged and repairs were estimated at $393,000
and required over a month to complete.
The tank ruptured at its base and collapsed. Investigators found that the tank was
overfilled and estimated that the combined air and water pressure in the tank at the time of
the rupture was greater than 70 psig— approximately ten times the design pressure. The
direct cause of the accident was found to be a design error in the tank overflow line. The root
cause was an inadequate design review. Other contributing factors were also uncovered.
Reference: DOE (1995). Type B Investigation Report Collapse of Tank 343 Advanced
Wastewater Treatment Facility, DOE-FN-0001-95, May 20, 1995.
33
-------
-------
Appendix D
Modeling of Venting System
34
-------
MATRIX Engineering, Inc.
170 Highway 35, Red Bank, NJ 07701 Phone: (732) 747-9111 Fax: (732) 741-5553
Februarys, 1998
Ms. Breeda Reilly
US Environmental Protection Agency
Mail Code 5104
401 M Street, SW
Washington, DC 20460
Re: Surpass Chemical Tank Failure
Dear Breeda:
Presented herein is my report of findings concerning the referenced
matter. This report is based on the documents reviewed to date and the
computer modeling of the tank failure scenario. This report may be
supplemented or amended subject to review of additional documents or other
materials relevant to the case.
Surpass Chemical Company is a repackager and marketer of muriatic
acid. As part of their operation, Surpass historically received tankwagon loads
of 20 Be muriatic acid for repackaging. The deliveries were received into a
fiberglass reinforced plastic (FRP) tank by pressuring the tankwagon to 22 to 25
psig (36.7 to 39.7 psia). On March 8, 1997, during the receipt of a load of
muriatic acid, the FRP tank failed leading to the release of approximately 5000
gallons of acid.
The tank was manufactured by Owens-Corning and is shown on the
design drawings to be 7'-7" in diameter and approximately 18' straight side with
a dome roof. The design specification indicates that the tank was rated for a
maximum 10" WC operating pressure and was supplied with a 24" diameter
hinged vent at the top for vapor relief.
The filling line leading to the tank to which the tankwagon connected was
a 2" diameter PVC line discharging to the top of the tank. The 24" diameter vent
had been modified by Surpass to control the release of acid fumes during
delivery. The 24" hinged opening was bolted closed and a 3" diameter vent line
was mounted on the top of the tank and routed to near grade where it
discharged into a caustic solution in a 50 gallon drum. The purpose of the
caustic solution was ostensibly to scrub the acid fumes to eliminate an
environmental or personnel exposure concern. The drum was reportedly filled
with approximately 20 gallons of solution which was manually blended to 18 wt%
using 50 wt% NaOH solution.
Environmental Compliance Design Project Management Safety
35
-------
The 3" diameter vent line was terminated within the scrubber drum by
fitting a custom fabricated diffuser. The diffuser was fabricated from PVC pipe
by drilling approximately 36 holes, each 5/8" diameter. After the accident, the
diffuser holes were found to be fouled with a crystalline solid.
The results of the initial accident investigation documented a timeline of
events which indicated that the delivery of muriatic acid totaled 47840 Ibs.
(4950± gals) and that approximately 80 minutes after the start of the off-loading,
the FRP tank failed and released its contents. Other subsequent consequences
of the accident led to the release of chlorine gas from a mixture of the muriatic
acid with sodium hypochlorite.
As a result of the failure of the tank, the roof and a 5 foot section of the
shell of the tank separated and flew 15 to 20 feet to the top of the adjacent
building indicating that the tank had been overpressured. A computer model
was created to analyze the pressure profile of the tank during the delivery and to
evaluate the effect of the tank's design features on the pressure within the tank.
The analysis of the pressure in the tank is based on an unsteady state mass
balance calculation routine. The parameters used in the calculations are
discussed below.
The tank pressure is at atmospheric pressure before the beginning of the
muriatic acid off-loading. The submersed diffuser within the scrubber solution
resulted in a backpressure during receiving operations equal to the hydrostatic
pressure of the solution within the scrubber drum. Therefore, no vapor flow
occurred out of the tank until the pressure within the tank exceeded this
superimposed backpressure. Based on the report of 20 gallons of solution in the
50 gallon drum, the maximum backpressure created by this solution is calculated
to be 16" WC (0.6 psig = 15.3 psia).
The delivery of the liquid into the tank caused an increase in the tank
pressure due to the reduction in the void volume of the tank. As the tank
pressure increased above the diffuser backpressure, vapor flow out of the tank
began. The tank pressure at any time was dependent on the rate of liquid flow
into the tank and the rate of vapor flow out of the tank.
The rate of vapor flow out of the tank is a function of the tank pressure
and the diffuser backpressure and was calculated using an orifice flow
calculation method for compressible flow. The pressure drop associated with
the flow of the vapor through the 3" vent line would further restrict flow, however,
this factor was considered negligible and was ignored in the modeling effort.
The rate of liquid flow into the tank is a function of the pressure in the
tankwagon, the differential head caused by the liquid being pumped to the top of
the tank, the backpressure created by the pressure within the tank and the
36
-------
dimensions of the delivery line and hose. The record indicated that the average
rate of liquid flow into the tank was approximately 60+ gpm based on 4947
gallons delivered in 80 minutes.
The computer modeling on this basis indicates that the tank pressure first
increased to overcome the diffuser backpressure and thereafter to establish a
sufficient vapor flow to equal the inlet liquid flow. During the liquid delivery the
tank pressure was calculated to reach approximately 15.3 psia (16"WC gauge).
This is more than the rated pressure for the tank but, nonetheless, failure of the
tank did not occur at this time.
Throughout the delivery of the liquid to the tank, the pressure of the liquid
discharged to the top of the tank is reduced below the pressure of the
tankwagon by both the line pressure drop and the change in liquid head. The
latter is the more significant factor. At the end of the delivery, as all liquid in the
tankwagon and line is vacated, the effect of the change in the liquid head is
quickly eliminated as the liquid in the line is evacuated and displaced with vapor.
At this point, the tank pressure increased sharply as the pressure within
tankwagon is relieved into the tank. The net pressure in the tank is the result of
the flow of pressured air into the tank which is only slightly offset by the
continued flow of vapors out of the tank through the diffuser.
At this point in the delivery process, the tank pressure quickly rises from
its pressure during liquid transfer. The peak pressure in the tank is dependent
on the available flow area of the diffuser. Two scenarios were evaluated to
demonstrate the effect of fouling of the diffuser holes. The first scenario is
based on no fouling. The second scenario assumes that the diameter of the 36
holes in the diffuser had been uniformly reduced to 1/4" from 5/8". Graphs of the
predicted pressure profiles of the tank under these conditions are presented in
the attachments.
Under conditions of no fouling the tank pressure at the end of the delivery
process was predicted to peak at 3.4 psig. Although this is significantly above
the recommended pressure limit for the tank, the prior use of the tank in this
service indicates that this did not exceed the yield point for the tank. In the
second scenario where the diameter was assumed to be reduced to 1/4", the
tank pressure peaks at approximately 12 psig.
The exact pressure which caused the tank failure was not predicted.
However, the configuration of the vent/scrubber system led to the operation of
the tank under routine operations outside the specifications of the manufacturer.
The fouling of the diffuser over time led to the further increase in the tank
pressure which eventually led to failure. The fundamental cause of this accident
was the improper design of the vent/scrubber system to control the maximum
pressure within the tank under foreseeable operating conditions.
37
-------
If you have any questions on this analysis, please contact me at your
convenience.
Very truly yours,
'J. David Calvert, PE, CSP
Attachments
cc: Tom Uden - ICF Kaiser
38
-------
DO
CD
oo
•] --
c
^
ct
• ]
.. c
• 1
39
-------
oo
o
CO
o
IO
o
CO
- O
- O
O
CO
OO
eg
(O
04
Tf-
OJ
04
eg
o
eg
oo
(0
40
-------
•l I
- C
c
u
- C
h- C
c
c
- C.
CO
co
CM
J«W I
41
CO
CD
------- |