Lessons Learned from an Incident at a Cryogenic

Gas Processing Facility

Adrian Pierorazio CFEI, PEng
Blast Effects, BakerRisk, Burlington, ON, Canada; [email protected] (for correspondence)
Published online 8 August 2015 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/prs.11763

In September 2009, a fatal incident occurred in a cryo- PROCESS DESCRIPTION

genic gas processing plant. The investigation of the incident The system was designed, built, and commissioned in the
indicated a number of potential issues that may have con- 1960s to convert coke oven gas to fuel gas for other parts of
tributed to or caused the event. These issues include hot work the complex. The system uses a cryogenic process to convert
procedures, electrostatic discharge, electrical conduit sealing, coke oven gas into three streams:
convective “breathing” due to multiple vents, equipment age,
 Downriver gas
maintenance, and worker training. The investigation con-
 Underfire gas
cluded that the fuel for this event was provided by small leaks
 Sub gas
from the product pipes inside of the cold box; the oxygen was
provided by convective “breathing” that occurred due to the The downriver and underfire gases are used as fuels in
presence of two vents from the cold box and the erosion of other areas of the complex. One part of the facility is down-
the flapper valves that were intended to seal these vents; and river, giving rise to the “downriver” name. The underfire gas
ignition occurred due to stray currents that resulted from is a lower pressure fuel and is used as part of the fuel for
poor hot work procedures, and locating the return lead far the coke ovens. The sub gas contains the components that
from the work location. This article provides an overview of have been taken from the coke oven gas to produce the two
the process and facility, a timeline of events, a summary of product fuel streams. The sub gas is processed further in
the investigative process, and a discussion of the lessons another area to remove more fuel gas and further concen-
learned from this event. V C 2015 American Institute of Chemical
trate the heavier components out of the stream.
Engineers Process Saf Prog 35: 143–148, 2016 The processing of the gas is performed using cryogenic
Keywords: incident investigation; cryogenic; hot work; separation. The coke oven gas is flowed over a bed of cold
welding; coke oven gas stones in a series of vessels called “regenerators.” The stones
are cold enough (as low as 22588F) to cause the gases to
liquefy or freeze. The inlet flow is then turned off and the
INTRODUCTION
pressure is dropped. The control of pressure and the use of
sweep gas cause the components to sublimate, allowing the
Incidents that occur in industry often provide key learn-
capture of the three product streams by switching the outlet
ings for other operators. While many of the specifics may
at the appropriate time.
change from site to site, it is valuable to consider how the
The regenerators are clustered into groups of four, called
lessons learned at one facility may apply to others, even if
“quads.” Each quad has its own insulated “cold box” enclos-
the processes are quite different. The incident being consid-
ing the bottom of the regenerators and the valves and piping
ered here is part of a coke oven system but illustrates a num-
there. These quad cold boxes are constructed with an insu-
ber of issues that have wider application. In this case, the
lated double wall with a void for the pipes and valves. The
process is a sublimation process to extract the fuel gases
outer wall is made of steel plate that is seal-welded to hold it
from the toxic and inert components of coke oven gas. The
in place. The inner skin is made of thin metal sheets that are
cryogenic and sublimation aspects are interesting, but should
clipped in place. Mineral wool insulation is installed between
not detract from the key lessons learned. This article starts
these skins and along the floor and roof to help maintain the
with an overview of the process and then presents the facts
about the activities underway at the time of the incident cryogenic process temperatures. Each quad is numbered,
before moving on to a narrative of the event and an analysis with a total of seven quads being installed in the main regen-
of the technical causes of the incident. Some of the interpre- erator area.
tations of the events may differ between the experts who The piping and valves that connect the quads together
investigated this incident, but the purpose here is to promul- are housed in cold boxes at the end(s) of each quad. There
gate the lessons learned. are three such cold boxes and these are designated using
letters A through C. These lettered cold boxes differ from the
quad cold boxes in that they are packed with insulation
This article was presented at American Institute of Chemical Engi- rather than having an insulated double wall around a hollow
neers, 2015 Spring Meeting, 11th Global Congress on Process Safety,
Austin, Texas, April 27–29, 2015. space.
The lettered cold boxes (A through C) are connected
C 2015 American Institute of Chemical Engineers
V through the quads and also through a “cross-tie” box that

Process Safety Progress (Vol.35, No.2) June 2016 143

Figure 1. Layout of processing area. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.
com.]

permits gas to bypass the quads and flow between the A Work was being performed in the main regenerator area
and B Cold Boxes. In this report, the cross-tie box is referred on the day of the incident. Rock wool insulation was being
to using the letters of the cold boxes at the ends of the vacuumed from the west side of Cold Box B. Grinding and
cross-tie (i.e., “AB”). welding repairs were being made at the south end of the AB
The cold boxes are designed to keep the space under the Cross-tie Box adjacent to the north face of Cold Box B, on
Quad 3 regenerators cold. This is done by attempting to seal the south face of Cold Box B (Figure 1), and on top of the
the cold box to prevent air from being introduced into the Quad 5 Cold Box. Hot work was being performed in the
cold box and to prevent gas from inside the cold box from control room 2 building around the time of the incident.
leaking out. This included the insulating work in Cold Box B as well as
The numbered cold boxes each have a nitrogen supply welding and grinding work that was being performed to
and backpressure regulator system. The quad boxes each repair the wall plates of the cold boxes (Cross-Tie AB, Cold
have a single nitrogen injection point consisting of a 2-inch Box B, and Quad 5 being worked on the day of the inci-
nominal diameter pipe penetrating the double-wall of the dent). There were numerous portable gas meters in use in
box. There are two 10-inch nominal diameter vents on the the west half of the B Cold Box, under the AB Cross-tie Box,
quad cold boxes, each equipped with a flapper-style relief and on top of Quad 5 at the time of the incident.
valve to relieve overpressure and to allow air to be drawn in Two work crews were in the area at the south end of the
to relieve vacuum to prevent the cold box from collapsing in AB Cross-tie Box (i.e., on the north wall of the B Cold Box
the event of a substantial vacuum forming in the cold box. near the junction of B Cold Box and Quad 3). A welding
One of the vent lines on each quad box is also fitted with a crew was welding a plate at the intersection of Cold Box B
backpressure regulator with a separate open-ended vent line and the AB Cross-tie Box, within a few feet of Quad 3’s cold
that is designed to maintain a positive pressure in the cold box (see Figure 2). Another crew was working at ground
box. By contrast, the lettered cold boxes receive nitrogen level, preparing to mix concrete to refurbish a footing for
from a sparger pipe and have a single vent pipe that runs up Cold Box B at the time of the incident. It was reported that
to roof level. the welding crew on the north side of Cold Box B (approxi-
mately 50 feet from their return lead) had just struck an arc
SUMMARY OF FACTS on the AB Cross-tie.
The following sections provide a summary of the key Quads 1, 2, and 3 were in operation, along with Cold
facts obtained during the investigation. Box A and parts of the Cross-tie AB box and Cold Box B. At
the time of the incident, one of the control valves in Quad 3
Activities at the Time of the Incident was switching.
Extensive maintenance activities had been underway
since 2005 and were ongoing at the time of the incident. Equipment Condition Prior to the Incident
Quads 1, 2, and 3 and Cold Box A were in operation and The main regenerator equipment was being repaired to
work was progressing on the other equipment. Cold Box B allow it to return to service and some cold boxes had been
was in the process of being rehabilitated, with the east half out of service for some time.
(i.e., the half connected to the south end of Quad 3) having During the refurbishment project, the flapper valves at the
been completed and returned to service weeks before the top of the vent stacks were being inspected. Some of the
incident. The west half of the Cold Box B was being stripped flapper valves were repaired and/or replaced. Following the
of insulation at the time of the incident, in preparation for incident, some of the flapper valves were found to have
repair, reinsulating, and return to service. The insulation was eroded and the seals were missing.
being removed by two vacuum trucks that were parked out- The welding crew on the north side of Cold Box B was
side of the main regenerator area with hoses running into using one of the group of four individual Witherup welding
the cold box. machines that were stacked and/or bolted together near the

144 June 2016 Published on behalf of the AIChE DOI 10.1002/prs Process Safety Progress (Vol.35, No.2)
Figure 2. Model showing work location. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.
com.]

road on the west side of the Control Room 2 area and shared to the hospital for treatment. The fires that followed the
a piece of angle welded to the opposite side of Cold Box B event burned for a number of hours and were concentrated
for their return lead location. The crew had welded a sup- in the area around and under the Quad 3 regenerators.
port to Cold Box B and had used a bottle jack to lift a mem- Inspections after the event showed that the Quad 3 Cold
ber on the bottom of Cross-tie Box AB in order to restore a Box floor had been deformed outward to the point that it
support clip for it back to the Cold Box B structure. The impacted the concrete below, and the walls of the Quad 3
original connections had failed and new connections were Cold Box had become detached and displaced/deformed
being made as part of the refurbishment project. outward on both sides of the cold box to the point where
During the investigation, one expert opined that parts of they were no longer enclosing the space underneath.
the system were beyond their lifespan. Whether this was true A double wythe brick infill wall to the east of Quad 3
or not was subject to debate regarding the assumptions in the failed and fell into the compressor building, away from the
lifespan model. In the end, the company had implemented an Quad 3 Cold Box.
inspection program to find and address cracks and any other The damage was generally confined to the immediate
flaws in the equipment and repair these when found. Many vicinity of Quad 3. The only significant structural damage
chemical processing facilities are of a similar vintage, or older, observed in Cold Boxes A and B were to the walls that sepa-
and rely on such inspection/repair programs to extend the life rate them from Quad 3. This damage resulted in wall move-
of their assets, so this is a common condition. ment that moved the pipes running between these lettered
boxes and Quad 3, resulting in some mechanical damage to
Safety Procedures the valves and piping in the lettered boxes. The walls were
Standard operating procedures in the regenerator area deflected away from the Quad 3 Cold Box.
required workers to use personal gas monitors to monitor The damage patterns are consistent with an “explosion”
for flammable gas, oxygen, and carbon monoxide as well as having occurred inside the Quad 3 Cold Box.
area monitors and periodic checks by operations with a
hand-held monitor. A review of the logs and the results of INTERPRETATION OF THE EVENT
testing of the personal monitors collected after the incident The investigation concluded that the incident that
confirms that none of these gas detectors sensed alertable occurred on September 3, 2009 was a confined vapor cloud
levels of flammable gas or carbon monoxide or an oxygen explosion (VCE). The limited extent of the damage indicates
deficiency in the area prior to the incident. that the energy involved was similarly limited. The damage is
There was a specific requirement for patching the live cold consistent with a relatively small vapor cloud, and the inten-
boxes to be performed as “cold work” with a gas sampling sity of the overpressure also indicates that the VCE was a
port and a nitrogen purge port to allow the atmosphere behind deflagration rather than a detonation.
the plate to be sampled and then purged before hot work (i.e., A VCE is a combustion event and, as such, requires three
welding) could commence under a separate hot work permit. elements to be present in sufficient quantity:
The welding crew was performing hot work at the junction of
 Fuel
the High Box and the Cross-tie Box, specifically on the Cross-
 Oxidizer
tie Box, using hot work, with no separate permit, no purge,
 Ignition
and no continuous monitoring of the sample port.
Each of these elements must exist in balance with the
Damage Sustained others. The fuel must exist in a concentration between the
On September 3, 2009, an “explosion” was reported in lower and upper flammable limits, the oxidizer (oxygen
Quad 3, resulting in one fatality and one worker being sent from the air in this case), must exist at above the minimum

Process Safety Progress (Vol.35, No.2) Published on behalf of the AIChE DOI 10.1002/prs June 2016 145
oxygen concentration required to support combustion, and The following sections provide a summary of how each
the ignition source must be in the correct location and of potential ignition source was evaluated.
sufficient strength to produce a self-sustaining combustion.
Each of these elements is discussed in the sections below. Electrical Equipment or Wiring Inside the Quad 3 Cold Box
One of the first hypotheses considered was that there
Fuel Source could be a short or malfunction of electrical equipment or
The underfire, downriver, and sub gas lines run through wiring in the quad 3 cold box. The only wiring that was
the Quad 3 Cold Box. These lines run at cryogenic tempera- found to run inside the cold box itself was thermocouple
tures. When these lines are serviced and the quad is brought wiring to measure the process temperatures. The voltage
online, the pipes are cooled to below 22508F. The stresses expected from a thermocouple under normal operating con-
due to cryogenic processes can cause flanges and other con- ditions in the cold box would be less than 10 mV and would
nections to leak. When these leaks occur, they are generally be a static voltage; this is well below what would be
small. Even small leaks in the relatively sealed space of the required to ignite a flammable gas cloud.
cold box can produce accumulations of gas and the concen-
Electrical Equipment or Wiring Outside the Quad 3 Cold Box
tration can increase over time. After the incident, the piping
As a corollary to the scenario above, there is an amount
was purged with nitrogen and maintained at a slight positive
of electrical equipment outside of the quad 3 cold box that
pressure of nitrogen. As part of their investigation, handheld
does operate at voltages of 120 volts or more. Since the wir-
quad gas meters were used to detect locations with low oxy-
ing and equipment are in an electrically classified area, the
gen readings. These low oxygen areas were identified as wiring is routed inside of conduit and junction boxes
areas where nitrogen was being dispersed from the piping. intended to prevent potentially flammable gas mixtures from
Several such areas were identified and it was concluded that, being present where significant voltages and/or currents
as expected, there were small leaks in the piping that could were present. To investigate this, the junction boxes near the
allow gas to escape. While flow rates were not measured, it cold box were opened and inspected. It was observed that
was concluded that these were a credible source of fuel for the seals between the quad 3 cold box were of an elasto-
the event. meric caulking-like material, rather than the cast solid mate-
rial that would normally be used in classified conduit. It was
Oxygen further revealed that some of the locations where cast or
As designed (and as in operation on the day of the inci- poured seals would be expected (i.e., where conduit is con-
dent), the nitrogen is introduced at a single point from the nected to a junction box) were missing seals or had gaps in
cross-tie alley (i.e., on the west of the odd numbered quads the seals.
and on the east of the even numbered quads) at the south The conduit was analyzed to determine the minimum gas
end of each quad cold box. With the erosion of the vent concentration that would produce a flame that could propa-
flapper valves, the nitrogen supply was unable to provide a gate through the conduit that connected to the thermocouple
significant positive pressure in the cold box. The leak in the junction box. The pressure produced by the deflagration of
flapper valves could result in the venting of the incoming this minimum gas concentration was then predicted and
nitrogen through the nearest vent (located just south of the compared to the pressure capacity of the conduit and junc-
C regenerator) and the nitrogen would not sweep through tion boxes. The analysis shows that any flammable gas con-
the cold box to purge the air. centration that can propagate through the conduit would
With hydrogen-rich gas (i.e., underfire or downriver gas) cause failure of the junction boxes and/or conduit. Since
leaking from the process piping, the gas mixture would have these components did not fail due to overpressure and since
a lower density (i.e., be “lighter”) than normal air. The result there were no other signs of combustion inside the conduit
of this was that the gas would rise out of the vents creating a and junction boxes, and since there were signs that there
convection loop in which the light gas rises out of the cold were functional seals between the cold box and the electrical
box and denser oxygen-laden air entered the cold box to system outside the cold box, it was concluded that this was
replace it and providing the oxygen for combustion to take not a credible ignition source.
place. Calculations show that this mechanism is a credible
source of oxygen to support combustion in the Quad 3 cold Static Electric Sparks from the Vacuum Truck Operations
box, if sufficient fuel is present along with a competent igni- Work crews were working on west side (quad 4 side) of
tion source. Cold Box B at the time of the incident. As part of this work,
they were using a vacuum truck to suction the old insulation
away to clear the area for rehabilitation. The vacuum hose
Ignition Source
ran east under quads 5 and 6, along the east side of quad 5,
With fuel and oxygen present in the cold box, the only
then under cold box B to the work location. The hose was
remaining required element is ignition. A number of poten-
not conductive hose and so there is a potential for static
tial ignition sources were considered including the following:
buildup and discharge from the hose. Static discharge is a
 Electrical equipment or wiring inside the quad 3 cold box high voltage/low current phenomenon which occurs locally
 Electrical equipment or wiring outside the Quad 3 Cold where a static source with high voltage potential creates an
Box arc to ground. In this case, the potential ignition locations
 Static electric sparks from the vacuum truck operations would be inside the pipe (charged particle to hose) or along
 Static electric sparks from within the quad 3 cold box the path of the hose. Since there was no report of fire or
 Ignition due to hot surfaces in the quad 3 cold box ignition at the work site or vacuum truck, it can be surmised
 Ignition due to deformation/failure of metal components that there was no deflagration propagating in the hose. Since
in the system the hose was generally lying on the ground or near quad 4/
 Ignition due to stray currents induced by process equip- cold box b, any local discharge from hose to structure would
ment in the control room 2 area be unable to ignite a flammable gas mixture in quad 3. As a
 Ignition due to stray currents induced by the welding in result, ignition from static discharge due to vacuum opera-
operation at the time of the incident tions was ruled out.

146 June 2016 Published on behalf of the AIChE DOI 10.1002/prs Process Safety Progress (Vol.35, No.2)
Static Electric Sparks from within the Quad 3 Cold Box of the incident was also considered as a potential source of
Potential sources of static electric discharge inside the appreciable stray currents. In welding operations, currents as
quad 3 cold box were sought. These included moving parts high as 100 amperes, or more, flow through the components
that were well insulated from any metallic structure, and any being welded together. The current then flows through the
particles or gases that can flow in such as was as to develop metal to the return lead to complete the circuit. At the time
an electrical charge. None of these potential sources were of the incident, a crew was welding on the north side of the
found and so this source of ignition was ruled out. B cold box at the west side of the intersection with the cross-
tie box (i.e., approximately two stories up, very close to
Ignition Due to Deformation/Failure of Metal Components in the quad 3). The return lead (colloquially, if inaccurately,
System referred to as the “ground clip”), was located near the mid-
It has been postulated that the ignition may have dle of the south wall of the B cold box near ground level.
occurred as a result of deformation or failure of a compo- This means that the full welding current was being passed
nent. The only mechanism for such an event to occur inside through a large part of the cold box structure as a result of
the quad 3 cold box is a gross failure of the process equip- the way the return clip was located. The combination of
ment. When the process equipment was inspected, the dam- high current source, low conductivity path through the quad
age was consistent with mechanical failure due to the
3 cold box, and the long path to the return lead, indicates
rupture of the cold box under positive pressure. The only
that there is a high potential for substantial currents to be
evidence of any component that may have failed prior to
flowing throughout the quad 3, crosstie box, and cold box b
combustion is the bellows at the bottom of regenerator 3a.
However, this failure appears to be inward and not outward, areas.
indicating that it is more likely due to impact during the fail- Witnesses reported that the ignition occurred at virtually
ure of the cold box, than due to rupture prior to the inci- the same time that a welding arc was struck on the AB
dent. In any event, the deformations in this area are Cross-tie Box. Electrical testing postincident revealed that
relatively small and not credible as being sufficient to heat welding being performed in the plant at that time was pro-
the metal close to the autoignition temperature of the fuel. ducing stray electrical currents that were strong enough to
interfere with electrical testing being performed. This indi-
Ignition Due to Stray Currents Induced by Process Equipment in the cates that the welding practices in place at that time were
Control Room 2 Area transmitting energy to areas away from the work location.
With ignition due to electrical, chemical, and mechanical Locating the ground return lead on the far side of the B
systems inside the quad 3 cold box ruled out, attention Cold Box approximately 50 feet from the welding location,
turned to how operations outside the cold box could pro- increased the energy sent to other areas of the system. Since
duce ignition inside the cold box. The most logical of these it was shown that stray current paths existed in the cold
would be for ignition to occur as a result of electrical current boxes, that the welding was being performed in a way that
flowing through the cold box from external sources. would encourage the creation of significant stray currents
Elementary electrical theory holds that the current along parallel paths in the cold boxes, that there was a corre-
between parallel paths will divide as the reciprocal of the lation in time between the start of the welding near Quad 3
resistances of the various paths. Stated more simply: current and the incident, and since all other ignition mechanisms
will flow through all available paths from a source to a sink considered were ruled out, it is concluded that the welding
(often ground) with low resistance paths getting a higher is the most probable cause of ignition.
portion of the flow than high resistance paths. The current
that flows through alternative paths such as along structural
elements and process piping is referred to as “stray currents.” SUMMARY AND LESSONS LEARNED
The cold boxes run at cryogenic temperatures, making The investigation concluded that the fuel for this event
the resistance of paths though the steel of the structure and was provided by small leaks from the product pipes inside
piping very low. This means that the lowest resistance path of the cold box; the oxygen was provided by to convective
for currents flowing through the structure is through the cold “breathing” that occurred due to the presence of two vents
boxes. A search was undertaken to find electrical equipment from the cold box and the erosion of the flapper valves that
that either did not have a viable local ground connection (to were intended to seal these vents; and ignition occurred due
cause currents to need to go through the structure), or was to stray currents that resulted from poor hot work proce-
capable of providing high enough currents to the system that dures, and locating the return lead far from the work loca-
even a small fraction of the current would be significant tion. While the processes described in this article are not
enough to cause ignition. typical of wider chemical industry, there are some lessons
In the process area, there are a number of compressors learned that can be applicable to other processes. As with
which use high voltages and currents. These are all a fair dis- any incident, the root cause is often a failure of the Process
tance from the cold boxes. The compressors were tested and
Safety Management (PSM) system and this incident was no
found not to be shorted and had reasonable connections to
exception.
ground to account for stray currents produced locally. These
were then ruled out.
The lighting and electrical switches in the regenerator
area are much closer to the cold boxes (implying lower Training and Contractors
resistance to the cold boxes), but none were found to have Despite documentation and training to the contrary, the
shorted to the structure or otherwise lack a proper electrical welders (contractors) did not seem to be aware of the hazard
return path. As a result, these were ruled out as likely igni- of welding on an enclosure that could contain flammable
tion sources. gas. Their personal gas monitors would not be able to detect
a flammable gas mixture inside the cold box from their loca-
Ignition Due to Stray Currents Induced by the Welding in Operation tion outside of the box and the welding current was able to
at the Time of the Incident penetrate the skin of the cold box easily. The safety person-
In evaluating electrical sources outside of the cold boxes nel were responsible for monitoring the gases in the area
themselves, the welding activity being conducted at the time and did not note this discrepancy in the sample location.

Process Safety Progress (Vol.35, No.2) Published on behalf of the AIChE DOI 10.1002/prs June 2016 147
Hot Work Hazard Identification
This incident appears to be a failure of the hot work per- The design, as received from the technology licensor
mitting system in that the sampling and welding practices included two large vent stacks. The presence of two stacks,
did not prevent ignition. Safe procedures were in place, but rather than one, allowed a convection loop to form. The
not followed. designer and operator failed to identify this hazard and mod-
ify the design appropriately.
Mechanical Integrity The nitrogen purge system, coming in at a point rather than
As with many aging facilities, erosion, wear, and fatigue being distributed by a sparger, allowed pockets of air to
were present in this facility. While active testing and repair remain. This hazard was not identified prior to the incident.
programs corrected many issues, the erosion of the flapper The welders were not aware of the hazards of welding on
valves was critical to the safety of the operation. live cryogenic equipment which may contain flammable gases.
Postincident inspections and testing demonstrated that the There was a belief that their flammable gas monitors would
grounding was disconnected in some places. While this was have alarmed if flammable gas was present in the area.
The welders were unaware of the path taken by their
not likely a cause of the incident, it is valuable to note that
welding current. While there was general understanding that
this electrical issue may occur in other process areas and
the current would follow the path of least resistance, this
may go undetected unless specifically sought out.
was interpreted as the shortest distance through the steel
While it is not specifically mechanical, the observation that
without considering parallel paths or the decreased resist-
the seals in the classified electrical conduit and junction boxes ance of steel at low temperatures.
were incomplete, missing, or sealed in a manner that is not
consistent with good practices (i.e., using an elastomer rather
Incident Investigation
than a castable seal) is a valuable lesson as well. The conduit
The cold box had a much smaller event 4 years prior to
and junction boxes had been installed many years prior and, it this incident. Investigators at that time concluded that it was
seems, had not been inspected to determine if all seals were in caused by lightning strike, and so recommended lightning
place. This may have been found if the seals were inspected. rods be installed to prevent a second strike. While there is
debate about the veracity of this cause of ignition, the earlier
Human Factors investigation could have identified the fuel and air sources
The flapper valves were located at the top of exhaust and recommended ways to control them. Had the original
stacks, well above the roof of the process building. To access investigation found the eroded flapper valves and identified
these valves required a crane and basket and so the equip- convection as a means of introducing air into the cold box,
ment was not regularly inspected or maintained. As such, the valves could have been replaced or the vent piping
personnel were unaware of the erosion of the valves, which reconfigured to prevent this from occurring and so could
led to the presence of air in the cold box. have prevented the event.

148 June 2016 Published on behalf of the AIChE DOI 10.1002/prs Process Safety Progress (Vol.35, No.2)