The Last Five Minutes of a Confined Space Job: Closing the Gap Between Exit and Safe Startup

Confined space work does not end when workers exit the vessel. In fact, the final minutes of a confined space job often determine whether a facility restarts safely or drifts toward catastrophe.

On a humid Louisiana evening in July 2023, a routine chemical process moved toward disaster. The incident occurred at Dow’s Louisiana Operations site in Plaquemine. Inside the Glycol II unit, ethylene oxide began shaping the night’s narrative.

At 9:17 p.m., a fireball illuminated the facility. Moments later, a vessel failed under extreme pressure. More than 31,000 pounds of ethylene oxide released into the atmosphere. Within minutes, a shelter-in-place order spread across nearby neighborhoods. However, the chain of failure started long before dusk. The seeds were planted during the final minutes of a confined space job that everyone believed was complete.

A Story That Begins in Quiet Corners

Two months earlier, a turnaround crew entered the reflux drum under a Permit Required Confined Space permit. Their task involved routine cleaning and inspection. Portable magnetic scene lights illuminated the vessel interior. These low-voltage lights work well in tight spaces and blind corners.

However, the drum soon filled with activity. Scaffolders, electricians, boilermakers, and inspectors moved through the vessel during the turnaround. In that bustle, several lights were never removed.

When the work ended, crews closed the vessel and signed the vessel closure form. The verification relied on a quick glance through the manway. In practice, this process acted as administrative confirmation rather than proof the vessel was empty.

When the unit restarted, the forgotten lights encountered pure ethylene oxide (EO). EO is an efficient solvent but an unforgiving chemical. Over time, it attacked plastics, adhesives, and elastomers inside the lights. Eventually, the materials degraded and fragments began breaking loose.

Metal shards, circuit board pieces, and battery components entered the process stream. These fragments moved through pumps and exchangers and migrated downstream. Eventually, debris reached the product cooler’s pressure relief path, where investigators later documented fragments consistent with the lights.

The first decisive failure was mechanical rather than chemical. A corroded carbon steel fragment struck the rupture disc protecting the product cooler. Normally, the disc opens symmetrically at a set pressure. Instead, the impact created a partial opening and an unintended leak path.

Post-incident metallurgical analysis by the U.S. Chemical Safety Board confirmed the evidence. Investigators found embedded metal fragments consistent with the degraded lights. Furthermore, the rupture disc opening pattern did not match a normal pressure event. These clues revealed how foreign material defeated protective equipment designed for very different failure scenarios.

Where Nitrogen Should Have Been, There Was Only Air

Another vulnerability waited downstream. Between the rupture disc and the pressure relief valve sat a fifty-foot relief segment. The piping measured four inches in diameter and once contained a protective nitrogen charge.

During maintenance in 2020, operators injected about 2.4 PSIG (pounds per square inch gauge) of nitrogen into the segment. On paper, this safeguard seemed sensible. However, the nitrogen charge was temporary and unmonitored. Over time, it slowly escaped through tiny leaks at flanges and fittings.

A pressure transmitter monitored the segment. Gradually, readings drifted from positive pressure to atmospheric conditions. Eventually, the transmitter even recorded slight vacuum levels. These readings clearly indicated air intrusion.

Yet no alarm required investigation. There was no oxygen monitoring system in place. In addition, procedures never required operators to review those readings. As a result, the nitrogen blanket quietly disappeared.

When the rupture disc was punctured in July 2023, ethylene oxide entered a space filled with atmospheric air. This created an extremely hazardous mixture. EO requires very little ignition energy. Certain EO-air mixtures ignite at roughly 0.06 millijoules, which is lower than the spark from walking across a carpet.

Therefore, ignition was almost inevitable. The ethylene oxide ignited inside the relief piping, and the flame front moved rapidly toward the pressure relief valve.

When pressure increased, the valve lifted as designed. However, this time it released flames rather than relieving process pressure. The valve discharged hot gases directly into the reflux drum vapor space.

Inside the drum, ethylene oxide vapor decomposed rapidly under heat. Pressure multiplied until the steel shell failed violently. Soon after, nearby neighborhoods received shelter-in-place warnings.

Investigators later traced the chain of events backward. They found a cluster of ordinary omissions. These omissions occurred during the final minutes of a confined space job, where unreconciled tools and incomplete inspection created hidden hazards.

The Real Cause Was Not Chemistry. It Was Confined Space Closure

The Plaquemine incident did not begin with unpredictable chemistry. Instead, it began with a simple assumption. Workers assumed that when a confined space job ends, the vessel is clean and ready for service.

In this case, that assumption proved dangerously wrong. Operators inspected the reflux drum from the manway while shining flashlights inside. However, they used no mirrors, borescopes, or remote cameras. As a result, the inspection relied heavily on line of sight.

Over time, the inspection process functioned more like a ritual than a safeguard. A signature replaced systematic verification. Meanwhile, the work lights remained hidden against the drum’s curved interior.

Tool control practices created another vulnerability. During the turnaround, workers carried radios, detectors, batteries, and lights into the vessel. However, Dow tracked tools only at the unit level. They did not track tools at the vessel level.

As a result, missing equipment triggered no confined space reentry inspection. Later, investigators discovered five magnetic lights were missing. Yet no procedure required searching the vessel for foreign materials.

The vessel closure form also contributed to the failure. The form required only a witness to look through the manway and confirm the drum was clean. However, it never defined what “clean” meant. Furthermore, it did not require remote inspection or reentry verification.

Consequently, trust replaced verification. Workers trusted that contractors removed their tools. They trusted unseen spaces contained nothing hazardous. In this case, those assumptions proved false.

After startup, the lights degraded in ethylene oxide and released fragments. These fragments migrated through the process system. Each piece represented a failure in confined space closure discipline.

Eventually, the rupture disc confirmed the story. The disc did not burst in its normal symmetrical pattern. Instead, testing revealed embedded metal fragments from the degraded lights. A foreign object opened the disc and created the ignition pathway.

If inspectors had verified the vessel interior properly, the outcome might have differed. Proper confined space inspection would have removed the lights. Without debris, the rupture disc would not have failed prematurely.

Updating ANSI/ASSP Z117.1 for Confined Space Closure

The Chemical Safety Board recommended updating ANSI/ASSP Z117.1 after the incident. For decades, this standard has guided confined space safety practices. It addresses hazard identification, atmospheric testing, and energy isolation.

However, the standard says little about the final stage of confined space work. That stage is vessel closure and startup preparation. The Plaquemine incident exposed this gap clearly.

First, the standard should require formal foreign material exclusion programs. These programs should track tools at the vessel level. Every item entering a confined space should appear in a reconciliation log.

Second, inspection requirements should expand. A quick flashlight scan through a manway should never qualify as verification. Instead, the standard should recommend tiered inspections using mirrors, cameras, or borescopes. In high-hazard environments, confined space reentry may be necessary before closure.

Third, the standard should address connected system hazards. Confined spaces rarely exist in isolation. They connect to piping, relief systems, and vapor spaces that can become hazardous during startup.

Finally, confined space closure should connect to management systems. Closure should require documentation, independent review, and startup authorization. Evidence should include photographs, inspection records, and reconciliation logs.

What We Must Do Next

Preventing the next Plaquemine incident requires more than revising procedures. Organizations must rethink the final phase of confined space work. The last minutes before vessel closure determine whether hidden hazards remain.

First, companies should implement vessel-specific foreign material exclusion programs. Every tool or device entering a confined space must be tracked. Before closure, teams must reconcile all equipment. If an item is missing, the vessel must be reinspected.

Second, inspection practices must shift toward verification. A glance through the manway cannot confirm vessel cleanliness. Instead, workers should use remote visual tools or controlled reentry when necessary.

Third, organizations must address connected system vulnerabilities. Reactive chemicals such as ethylene oxide require continuous vigilance. Inerting systems cannot remain temporary safeguards.

Instead, inerting must function as a continuous safety control. Operators should monitor oxygen levels and pressure readings consistently. A negative pressure reading should trigger immediate investigation.

Finally, leadership must embed closure discipline into organizational culture. Confined space closure should connect directly to Management of Change processes. Startup authorization should rely on evidence rather than assumption.

Closing Thoughts on Confined Space Safety

Catastrophic failures rarely result from mysterious chemistry. Instead, they emerge from familiar routines performed with slightly less rigor than necessary.

The Plaquemine incident illustrates this reality clearly. Ethylene oxide behaved exactly as expected in the presence of oxygen, heat, and confinement. What failed was the human system designed to keep chemistry within safe boundaries.

A forgotten work light, an unreconciled tool list, and a neglected pressure reading created the pathway to disaster. Each issue appeared minor on its own. Together, they formed the classic Swiss Cheese failure chain.

Ultimately, the last five minutes of a confined space job determined the facility’s fate. During those minutes, verification should replace convenience. A vessel is not safe because a form says it is safe.

A confined space is safe only when disciplined inspection proves the system is ready for startup.

Learn How Veriforce Supports Chemical Industry Safety

Managing confined space risks requires more than procedures. It requires systems that help organizations verify compliance, strengthen oversight, and reduce operational risk. Veriforce works with chemical manufacturers to support contractor management, safety training, and compliance programs that help teams operate safely and confidently. Learn more about how Veriforce supports the chemical industry.

About the Author

James A. Junkin, MS, CSP, MSP, SMS, ASP, CSHO is the chief executive officer of Mariner-Gulf Consulting & Services, LLC and the chair of the Veriforce Strategic Advisory Board and the past chair of Professional Safety journal’s editorial review board. James is a member of the Advisory Board for the National Association of Safety Professionals (NASP). He is Columbia Southern University’s 2022 Safety Professional of the Year (Runner Up), a 2023 recipient of the National Association of Environmental Management’s (NAEM) 30 over 30 Award for excellence in the practice of occupational safety and health and sustainability, and the American Society of Safety Professionals (ASSP) 2024 Safety Professional of the Year for Training and Communications, and the recipient of the ASSP 2023-2024 Charles V. Culberson award. He is a much sought after master trainer, keynote speaker, podcaster of The Risk Matrix, and author of numerous articles concerning occupational safety and health.

References

U.S. Chemical Safety and Hazard Investigation Board. (2025). Explosions, fires, and toxic ethylene oxide release at Dow Louisiana Operations, Plaquemine, Louisiana (No. 202303ILA). https://www.csb.gov/us-chemical-safety-board-releases-investigation-report-on-the-2023-explosion-and-toxic-ethylene-oxide-release-at-dow-plant-in-plaquemine-louisiana/