The hazard was not the steel: an oxygen manifold burst on a Gulf site-clearance job

A pressure test is not a qualification. That is the whole of this report in one line.

On 30 April 2026, at about 0700 hours, a contract diver was injured aboard a lift boat performing diving support for decommissioning work at South Pass 28, platform 99, on OCS Lease 00694. He was installing and testing a newly designed oxygen manifold for the diving chamber’s breathing system. He opened the valve on a bottle carrying 2,600 psi. The manifold burst, and metal fragments struck his right hand, arms, chest and face.

He received first aid from a medic on the lift boat, went by crew boat to a nearby facility, and was flown by helicopter to New Orleans for further treatment. There were no other injuries, no damage to the platform and no environmental impact. He later returned to the job and the diving rotation after full medical clearance.

Estimated property damage: $750.

Why a decommissioning audience should read a diving incident

The activity codes on the report matter more than the injury count. The form checks both decommissioning and site clearance. This was not a drilling job or a production upset. It was the tail end of the work our readers sell into — the verification phase that follows plugging and removal, described in the application-to-clearance lifecycle.

Decommissioning safety conversation tends to concentrate on the obvious energy: the lift, the cut, the explosive severance, the tow. This incident involved none of those. The hazard was in the support system that makes diving possible — a gas manifold assembled on deck, on a lift boat, in shallow water, four miles from shore.

The scope of work was site clearance. The energy that injured the worker came from a 2,600 psi oxygen bottle feeding a newly designed manifold.

For a service company, that is the uncomfortable part. A diving, ROV or marine contractor can run a flawless cutting and recovery programme and still carry its serious injury risk in the breathing-gas skid.

What actually failed

The sequence the report sets out is specific, and it is worth following closely because the obvious explanation is the wrong one.

The manifold was new. Its design allowed multiple oxygen bottles to supply the diver’s chamber through a manifold, replacing a single-tank setup. Before the job, it was hydrotested to 7,000 psi in a controlled environment with no failures. The operator’s own investigation found that the diver followed normal installation procedures, and confirmed there were no installation issues.

So the assembly was stronger than the service pressure by a wide margin, and the work was done correctly. It still burst.

The root cause analysis identifies a failure mode specific to oxygen service:

• The ball valves in the manifold were not oxygen-clean. The report notes they possibly contained hydrocarbons or grease, which ignited in the presence of high-pressure oxygen.
• The valves were quick-opening ball valves rather than slow-opening needle valves. Rapid pressurisation created an adiabatic compression heat spike, which ignited the contaminants.

The report states the conclusion plainly: the manifold held 7,000 psi without failure in a controlled environment with inert gases, but the offshore conditions at the time of the incident are what caused the failure. The numbers in the narrative are given as a bottle pressure of 2,600 psi and a failure at around 2,500 psi — in either case, roughly a third of the tested pressure.

A hydrotest with an inert medium proves mechanical strength. It proves nothing about oxygen compatibility, component cleanliness or pressurisation rate. Three variables the test could not see are the three that mattered.

The second cause: a missing procedure

The report records two cause categories, and the second is the one with commercial consequences.

Under management systems, it cites “no or inadequate hazard analysis / written job procedures,” and specifies the gap: the absence of a detailed written Safe Work Procedure for oxygen systems requiring oxygen-clean components, needle valves instead of ball valves, and controlled pressurisation.

That is not a finding about a person. It is a finding about a document that did not exist. The new multi-bottle design arrived on the vessel without a written procedure governing the three things that would have prevented the failure.

The contractor’s immediate response was to stop using multi-bottle manifolds for chamber supply across all company diving operations and revert to a single-tank setup until further notice. A capability was withdrawn because the procedure to run it safely had not been written first.

How the investigation was conducted

Two details in the investigation record are useful to anyone who builds evidence trails from public filings.

First, the timeline. The incident occurred on 30 April. The investigation team reached the lift boat on 4 June — weeks later, which the report attributes to challenging logistics, weather concerns and the remote location. By then the vessel had moved to a different platform in the same field, and the damaged manifold had already gone to a laboratory for analysis. The team travelled from a dock in Venice, Louisiana, aboard a motor vessel, accompanied by an operator supervisor.

Second, a discrepancy. The narrative describes the team arriving on 4 June 2026. The form’s own field for date of onsite investigation records 25 June 2026. The report was approved on 9 September 2026 and printed on 25 September 2026.

Neither point undermines the findings. Both are a reminder of something we keep returning to in reading BSEE signals: a public record is evidence of what was filed, and the narrative and the coded fields of the same document can disagree. Quote the field you actually used.

The team requested a defined document set: the job safety analysis, the initial report of accident or injury to the regulator, the preliminary incident investigation report, the root cause analysis, witness statements, photographs of the damaged equipment, and the equipment inspection report.

That list is effectively a contractor’s audit checklist. If an incident happened on your spread tomorrow, those seven documents are what would be asked for.

No violation, one recommendation

The report records no possible OCS violations related to the accident, and no accident investigation panel was formed.

It does carry a recommendation. The New Orleans District recommends that the Office of Incident Investigations look into developing a safety alert specific to oxygen systems in the Gulf — and specifically, into ensuring that oxygen systems in diving operations use components and equipment designed for oxygen service, rather than the valves and fittings commonly used offshore elsewhere.

That last clause is the generalisable warning. The failure came from applying normal offshore hardware practice to a service that does not tolerate it. Note also that the published version of this report carries Marine Minerals Administration letterhead and explains that MMA personnel are identified in the text as BSEE personnel, a consequence of the reunification of BOEM and BSEE.

What a contractor should take from this

1. Separate strength testing from service qualification. Record the test medium. A hydrotest or inert-gas test does not qualify an assembly for oxygen, and the report proves the gap is not theoretical.
2. Treat oxygen as its own discipline. Oxygen-clean components, cleanliness verification on receipt, and documented cleaning intervals. “New” is not “clean.”
3. Specify valve type, not just pressure rating. Slow-opening needle valves for pressurisation. The report attributes the heat spike to quick-opening ball valves.
4. Write the procedure before the equipment ships. A new design needs its Safe Work Procedure on the vessel with it. The management-systems finding here is about sequencing, not negligence.
5. Include support systems in your decommissioning hazard register. Breathing gas, chamber systems, hydraulic skids and pressure testing equipment sit alongside cutting, rigging and lifting. The campaign’s highest-energy hazard may be on deck.
6. Rehearse the medevac chain. Medic, crew boat to a nearby facility, helicopter to shore. It worked here. It worked because the links existed before they were needed.
7. Keep the seven documents current. Job safety analysis, initial notification, preliminary report, root cause analysis, witness statements, photographs, equipment inspection report.

Why the small numbers matter

A $750 property damage figure and a finding of no violations will keep this report out of the trade press. That is precisely why it is worth reading.

The expensive version of this incident is the one where the fragments travel differently. The procedure gap, the cleanliness gap and the valve choice would have been identical. The only variable was geometry.

Public safety records are among the few places where a service company can study a near-identical job without paying for the lesson. For accounts in a pre-execution window, a demonstrated procedure for oxygen systems is also a differentiator — it answers a question the operator’s own investigation had to ask after the fact.

Sources

• Marine Minerals Administration, Gulf of America Region, Accident Investigation Report (MMS Form 2010), South Pass 28 platform 99, OCS Lease 00694, NuGulf Operating, L.L.C. — incident dated 30 April 2026, approved 9 September 2026, printed 25 September 2026. Published under NTL category "Decommissioning Activities," safety category "Explosion."
• U.S. Department of the Interior, Secretary's Order No. 3451, "Establishment of the Marine Minerals Administration", 10 July 2026 — for the agency naming in the report.
• Electronic Code of Federal Regulations, 30 CFR Part 250, Subpart Q — Decommissioning Activities, including the site clearance provisions at §§250.1740–250.1743.