The Flood Before the Fire

▶️ Prefer to watch? Every story’s a short video → https://www.youtube.com/@GetSafetyIntel · 📣 Got a near-miss? Share it anonymously (QR + link at the bottom).

A note on the date: today’s exact-date incident is covered in our other brief today; this second story is from this same week — September 20 — a different trade with a lesson worth its own five minutes.

Four years ago yesterday, two brothers went to work at a refinery outside Toledo and didn’t come home, at the end of a shift where more than 3,700 alarms had gone off in twelve hours and two of them were told to drain a vessel “as fast as you guys can.” They may not have even known what they were letting onto the ground. This is a story about what happens to good operators when the system floods them faster than any human can swim.

On September 20, 2022, the BP-Husky refinery in Oregon, Ohio, just east of Toledo, had a bad day that got worse in slow motion. It started with a process upset in the naphtha hydrotreater unit and a loss of containment. Units began shutting down. And through a cascade of operational decisions in an increasingly abnormal plant, liquid naphtha — a light, highly flammable hydrocarbon — ended up flowing into a pressurized vessel called the Fuel Gas Mix Drum that was only ever supposed to hold vapor. The drum filled with liquid and overflowed into the vapor piping that feeds the refinery’s furnaces and boilers, which meant liquid naphtha was now headed toward open flames. That’s a catastrophe waiting on a match.

To relieve the liquid-full drum, employees were directed to drain it “as fast as you guys can.” Two of them — brothers, Max and Ben Morrissey, both BP employees — opened the vessel and released the liquid to the ground. It flashed into a vapor cloud, the cloud found an ignition source, and the flash fire that followed killed them both. Over 23,000 pounds of naphtha were released. The property damage came to roughly $597 million. It was the largest fatal incident at a BP-operated U.S. refinery since the Texas City disaster of 2005, which killed 15.

And here is the detail from the CSB’s final report that should stop you cold: the two men who opened that vessel may have believed the liquid inside was an amine-water solution — not naphtha. They weren’t reckless. They may not have known the thing they were draining onto the ground was one spark away from killing them.

This is HOP with the volume all the way up. Error is normal — but the CSB found more than 3,712 alarms in the twelve hours before the fire, an “alarm flood” that overwhelmed and distracted the board operators and delayed and garbled the response to the alarms that actually mattered. You cannot hand a human being 3,700 alarms and then be surprised when the critical one gets lost in the noise. Blame fixes nothing here: the men who opened the vessel were following a direction, in a plant that had been abnormal for hours, quite possibly with the wrong idea of what was in their hands. Context drove every bit of it. And learning was intentional — except it wasn’t: the CSB found a 2019 incident at this same refinery where naphtha had started filling this same vessel, warning signs were flagged, and no action items were ever developed to stop it from happening again. The plant had already been shown the movie once. Nobody wrote the ending down.

The person closest to the work saw this coming, too. Somewhere in that alarm flood, an operator knew the plant was past the point where the control system could cope. What they didn’t have — what the CSB specifically called out — was clear, trained, exercisable stop-work authority to call it and shut the thing down before it killed anyone. How your organization responds to abnormal situations, and whether the person at the console feels able to hit the brakes, is the whole game.

The floor-level rule is 29 CFR 1910.119 — Process Safety Management of Highly Hazardous Chemicals (verify current text at https://www.osha.gov). Read the parts Toledo failed:

1910.119(f) — operating procedures, including clear operating limits and what to do when you exceed them. A vessel that’s supposed to hold vapor filling with liquid is an operating-limit deviation with a defined consequence — and the response to it can’t be improvised at the moment naphtha is heading for the furnaces.

1910.119(e) — process hazard analysis. The refinery’s own PHAs had actually identified overflow scenarios. The safeguards on paper — instrumented systems, relief valves — didn’t stop the real one. A hazard you’ve identified and then under-protected is arguably more dangerous than one you never saw, because you’ve told yourself you handled it.

1910.119(l) — management of change, and (m) — incident investigation. The 2019 near-repeat is the (m) failure in the flesh: you only get the free lesson if you actually turn the investigation into action items and close them. And the human piece the CSB hammered — alarm management and stop-work authority — lives in the recognized good-practice standards (ISA-18.2 on alarm systems) that a real PSM program is built on.

Translate it to the floor: you probably don’t run a refinery, but you have a control room, a panel, a phone that lights up, or a shift where everything goes sideways at once. Two questions decide whether your people swim or drown. First — when your system floods, do your people have a clear, drilled way to stop the job, and do they actually believe they’re allowed to use it without getting chewed out? Second — the last time something almost went wrong, did it turn into a closed action item, or did it turn into a story people tell? Toledo had the story. It didn’t have the action item.

Most of what we cover is E, H, and S at once, and Toledo hits all three.

Safety (S): the spine — an alarm flood that outran the operators, a liquid overflow into a vapor system the safeguards didn’t stop, a possible mistaken belief about what was even in the vessel, and no exercised stop-work authority to end it. The fix was never a calmer operator; it was a system that doesn’t flood the human and a culture where hitting the brakes is celebrated, not second-guessed.

Health (H): the acute toll was two men killed in a flash fire — thermal and blast injury, the way a vapor-cloud ignition kills, in seconds. Behind the two deaths sits the health load nobody tallies: the coworkers who were there, the responders, the families of two brothers lost in one event.

Environmental (E): over 23,000 pounds of naphtha — a volatile hydrocarbon, a stew of VOCs — was released, much of it burned in the fire, in a refinery that shares its air and its fence line with a residential community near Toledo. Loss of containment on that scale isn’t only a fire risk; it’s an air-quality and emissions event for the people downwind, and it ties straight into the Clean Air Act and RMP framework that governs facilities handling this much flammable material. The same failure that killed two workers put a hydrocarbon release and a smoke plume over a community. It’s never just the S.

The CSB just closed the book on another fatal cracker explosion — the reports are still landing. On September 16, 2026, the U.S. Chemical Safety Board released its final report into the June 4, 2025 explosion in an ethane cracking furnace at the Shell Polymers facility in Monaca, Pennsylvania. So what for safety leaders: the CSB is still doing the work of turning catastrophes into named, fixable causes — furnace and fired-equipment integrity, abnormal-operations management, the same family of issues that shows up in Toledo. When a fresh CSB report drops in your sector, read it like it’s about your plant, because the next one might be. (Source: U.S. CSB, final report Sept. 16, 2026 — https://www.csb.gov/investigations/completed-investigations/)

And another final report a month earlier, on a fatal gas explosion at a coke works. On August 10, 2026, the CSB issued its final report on the August 2025 fatal coke oven gas explosion at U.S. Steel’s Clairton Coke Works in Pennsylvania. So what for safety leaders: two final reports on fatal process-safety events inside a single autumn is not a coincidence — it’s the pattern. Fired equipment, gas systems, and abnormal situations keep killing people in heavy industry. If your process safety program hasn’t been stress-tested against “what happens when three things go wrong at once,” Toledo and Clairton are your reminder to do it before someone else writes your case study. (Source: U.S. CSB, final report Aug. 10, 2026 — https://www.csb.gov/us-chemical-safety-board-issues-final-report-on-august-2025-fatal-coke-oven-gas-explosion-at-us-steel-clairton-coke-works/)

Shared blamelessly, because that’s the only way anyone learns from it.

A water-treatment operator at a mid-size plant is working a rough night — a pump trips, a tank alarms high, and the SCADA screen lights up with a wall of nuisance alarms it always throws when things get busy, most of which the crew learned long ago to swipe away. Buried in that wall is a real one: a valve that’s supposed to be closed is reading open, and a tank is filling that shouldn’t be. The operator, heads-down clearing the noise, nearly misses it — catches it only because a second operator wandering past glances at the mimic and says “why’s that one filling?” They stop, close the valve manually, and avert an overflow. Nobody’s hurt. Afterward they do the unglamorous thing: they sit down and rationalize the alarm list, kill the chronic nuisance alarms that had trained everyone to ignore the screen, and agree out loud that “I’m going to stop the process and figure this out” is always an acceptable sentence on that shift.

The HOP read: the operator wasn’t careless — he’d been conditioned, over months, by a system that cried wolf until the crew stopped hearing it, which is exactly what an alarm flood does to a human being. The catch wasn’t a more vigilant employee; it was a second set of eyes and a screen that hadn’t yet been fixed. Toledo’s board operators had 3,712 alarms and no fresh set of eyes and no brake they felt free to pull. This crew had a coworker walk by and a culture where stopping was allowed. That’s the entire difference between a near-miss you write up and a report the CSB writes about you.

Got a fail or a near-miss? Hit reply, or scan the QR below — we’ll feature it anonymously — no names, no company, no blame. Just the lesson, so the next crew doesn’t learn it the hard way.

Share your near-miss here → https://scantoconfess.com — 100% anonymous, no names, no company, no blame.

Before your next shift, go look at your alarms — the SCADA screen, the panel, the phone tree, whatever floods when your operation goes abnormal. Do two things. One: find the chronic nuisance alarm your crew has trained itself to swipe away without reading, and start the list of alarms to rationalize or fix, because every one of those is teaching your people to ignore the screen that will someday be trying to save them. Two: ask the person at the controls, plainly, “if this all went sideways right now, do you feel like you could stop the job — and would anybody give you grief for it?” If the answer is anything but a clear yes, that’s today’s work. The Morrissey brothers were told to drain it fast. Give your people permission to say stop.

🎬 Rather watch than read? Every Safety Intel story is a short video on YouTube → https://www.youtube.com/@GetSafetyIntel — subscribe so you never miss one. 📸 Share YOUR near-miss (100% anonymous): https://scantoconfess.com — or just scan the QR above. 📤 Forward this to a crew member who needs it today.

Please stay Safe & Hydrated!!!

BP-Husky Toledo Refinery naphtha release and fire, Sept. 20, 2022 — CSB final report (June 24, 2024): 3,712 alarms, 23,000+ lbs naphtha, two brothers killed, amine-water belief, 2019 precursor: https://www.csb.gov/us-chemical-safety-board-issues-final-report-into-fatal-2022-fire-at-bp-husky-refinery-near-toledo-ohio/

CSB full final report PDF: https://www.csb.gov/assets/1/6/final_report_-_20241.pdf

Process Safety Management — 29 CFR 1910.119: https://www.osha.gov/laws-regs/regulations/standardnumber/1910/1910.119

Shell Polymers Monaca final report, Sept. 16, 2026 (U.S. CSB completed investigations): https://www.csb.gov/investigations/completed-investigations/

U.S. Steel Clairton Coke Works final report, Aug. 10, 2026 (U.S. CSB): https://www.csb.gov/us-chemical-safety-board-issues-final-report-on-august-2025-fatal-coke-oven-gas-explosion-at-us-steel-clairton-coke-works/