They Went In to Drain Water
The easy version of DuPont La Porte is that four workers opened valves on a line full of poison without wearing respirators. That is true, and it is the least useful sentence anyone has written about that night. Because the crew that walked in did not think they were draining poison. They thought they were draining water. The building had spent years making that mistake inevitable.
How the unit worked.
The Lannate unit at DuPont's La Porte plant made methomyl, an insecticide DuPont sold under the trade name Lannate. One of its feedstocks was methyl mercaptan: the rotten-cabbage smell that gets added to natural gas so you can detect a leak, except here it ran through the pipes as a liquid under pressure, in bulk, and it is highly toxic. Its immediately-dangerous-to-life-or-health (IDLH) level is 150 ppm. The process ran inside a multi-story manufacturing building that also held chlorine and methyl isocyanate. A waste gas vent header ran through the third floor that carried process vent gas off to an incinerator. That header had low spots where condensate collected, and for years operators believed that condensate was mostly water. It wasn't.
They drained it by hand, through valves, inside the building. The building's only defense against a toxic cloud was two rooftop ventilation fans, and those fans were sized for a different problem: keeping flammable gas below 25% of its lower explosive limit (LEL). Nobody had designed them to handle a toxic release. Hold onto three pieces of equipment: the vent header, its drain valves, and those two fans.
FIGURE 1 - Waste Gas Vent Header Piping / drain valves (yellow + red circles). Source: CSB.
What happened.
FIGURE 2 - Release-pathway diagram Source: CSB
The week before, water got into the methyl mercaptan feed piping, partly because a safety interlock had been bypassed for a spring turnaround and never restored. The unit shut down on November 10. A time where the temperatures in Houston were around 40°F. These low temps caused the water and methyl mercaptan to freeze together into a clathrate hydrate, an ice-like solid that plugged the pipe. For days, operators fought the plug without knowing what it was. On November 14 the technical team concluded it was a hydrate and told operators to melt it with hot water. To relieve the pressure that melting would create, operators opened valves between the methyl mercaptan piping and the waste gas vent header. That put liquid methyl mercaptan into a header built only for vapor.
Early morning on November 15 the plug let go, and roughly 24,000 pounds of liquid methyl mercaptan flowed toward the third floor. At 2:51 a.m., high-pressure alarms went off. Operators read them as the same old water-condensate problem and went to drain the header.
Between 3:01 and 3:13 a.m., the shift supervisor opened two sets of drain valves inside the building. Liquid methyl mercaptan came out instead of water, flashed to gas, and filled the building at about 9,000 pounds an hour. Four people died: shift supervisor Wade Baker, and operators Crystle Rae Wise, Robert Tisnado, and his brother Gilbert "Gibby" Tisnado, who died trying to fit an escape respirator onto Robert's face before he could connect his own air. The CSB found they all died of asphyxia and acute methyl mercaptan exposure.
The release ran for about three hours. The storage-tank pump was not shut off until 6:02 a.m.
The easy story.
Run the hurried investigation.
Did the workers open the valves? Yes.
Were they trained? Yes.
Did written procedures exist? Yes.
Did they wear respiratory protection? No.
Did a supervisor lead them in? Yes.
Every box checks. You can close the file on "workers entered a toxic atmosphere without protection and drained a line they never should have opened." It is technically accurate but when it comes to truly drilling down, it is useless. Useless because it treats the last hands on the valve as the cause instead of asking the only question that matters: why did a trained crew, led by a supervisor, walk into that room certain it was safe? Answer that, and the operators quietly disappear from the story.
What actually set it up.
FIGURE 3 - The 2011 hazard-evaluation worksheet that flagged the hydrate risk (Source: CSB)
They were certain it was condensation because for years the answer had been exactly that. Draining the vent header inside the building was a normalized task.
What changed that night was invisible to them.
A string of piping moves made to melt the hydrate had rerouted liquid methyl mercaptan into a header that was never meant to carry it, and none of those moves went through management of change.
The CSB found no MOC for heating the piping,
no MOC for the new alignment,
3rew used the one detector they had left, their own noses. In the 17 hours before the release, troubleshooting set off 32 methyl mercaptan gas alarms, and everyone treated them as normal. That is normalization of deviance, and Andrew Hopkins would tell you it is an organizational habit, not a personal flaw. DuPont's own auditors had flagged the ineffective ventilation about five years earlier. A 2011 process hazard analysis had flagged the hydrate risk. Both findings sat. The evidence points one way: the people who could have stopped this were upstream of the valve, and years early.
Figure 4 = Barrier analysis of the DuPont La Porte release. Six independent safeguards were meant to stand between 24,000 lb of liquid methyl mercaptan and the crew
How they got past human error.
Here is the method that gets you past "operator error," and you can run it at your own site this week. It is barrier analysis, often drawn as a bowtie. Put the hazard in the center: 24,000 pounds of liquid methyl mercaptan under pressure. Put the people on the other side. Then list every barrier that was supposed to stand between them, and check each one honestly.
Reading straight down the verdict column, the gaps are the story:
MISSING — Inherently safer design, so liquid toxic can never reach an occupied building. The header was never meant to carry liquid, and nothing physically stopped the alignment that let it.
MISSING — Management of change to catch the new piping line-up. No MOC was done, on any of it.
INADEQUATE — A safe-work procedure for draining a toxic line. Draining inside the building was routine, and the "restricted access" the rules called for was never even defined.
MISSING — Toxic gas detection with an alarm the crew could see or hear. No lights, no horns, hydrogen-sulfide proxies on a screen in another room.
DOWN & UNDERSIZED — Ventilation sized for toxicity. Both rooftop fans were down, and the design was too weak for a toxic release even when they ran.
MISSING — Respiratory protection and a controlled emergency response. When the distress call came, more operators ran in with nothing on their faces, because no alarm told them the air was lethal.
Six barriers. Every one was already gone before the shift supervisor touched a valve. Chart it that way and "why didn't the operator wear a respirator" stops being an interesting question.
The honest finding writes itself: you cannot put the cause on the last person in line when management had already removed every layer in front of them. Peter Susca calls those upstream calls the work of hazard gatekeepers, the leaders who decide which hazards get to pass through the gate. At La Porte the gate was standing wide open.
The gap.
Every one of those barriers existed on paper somewhere. DuPont had a process safety management system, a technical standard that named the hydrate risk, auditors who flagged the ventilation, procedures for breaking into lines. The gap was not a missing program. It was that none of the programs reached the third floor of that building on a cold night in November.
Trevor Kletz spent a career on the single idea that would have made all of it moot: what you don't have can't leak. The strongest barrier is the one you never need, because the hazard was designed out. A vent header that physically cannot carry liquid methyl mercaptan does not depend on a fan, an alarm, a procedure, or a respirator being in place at 3 a.m. Every control that relies on a person doing the right thing under pressure will eventually meet a person under pressure.
That is why inherently safer design sits at the top of the hierarchy of controls and administrative rules sit near the bottom.
La Porte is what the bottom of that hierarchy looks like when the top gets skipped.
Monday morning checklist.
Three things you can check at your own site this week. Save this, or run it as a ten-minute toolbox talk.
Find one toxic or flammable material that can end up somewhere it was never designed to go. Trace a drain, a vent header, a relief tie-in. Ask what physically stops the wrong phase or the wrong chemical from reaching an occupied space. If the answer is "a procedure" or "the operator knows," you have found a La Porte in waiting.
Pull your open "urgent" work orders on safety-critical equipment: gas detectors, ventilation, alarms, interlocks. Find the oldest one. Nearly a month is how long DuPont's ventilation fix sat. Ask who is allowed to keep a unit running with that barrier down, and whether anyone has to sign for it.
Walk your worst enclosed space and ask how a worker inside learns the air has turned lethal. If the alarm only lives on a screen in the control room, they don't. Put the warning where the people are.
None of these needs a capital project. They need someone to look before the night the alarms get normalized.
The four people who died at La Porte were not careless. They were the last barrier in a building where every barrier before them had already been removed. Blaming them is the cheapest possible reading of that night, and it is exactly how the next plant keeps its gates wide open.
If you have ever drained a line that everyone swore was just water, tell me about it in the comments. I read every one.
Before Human Error is a weekly teardown of how good operators end up carrying the blame for bad systems. The inbox version lives at Before Human Error on beehiiv, and it lands a day before this hits LinkedIn. Subscribe and I'll send you the one-page Process-Safety Teardown Checklist: every Monday-morning check from these issues, on a single page. The link is in the first comment. You can also subscribe here on LinkedIn to catch each edition.
Sources: U.S. Chemical Safety and Hazard Investigation Board, Investigation Report 2015-01-I-TX, DuPont La Porte (final report, 2019); U.S. Department of Justice and U.S. EPA sentencing releases (2023). Figures reproduced from the CSB final report. (Sponsor slot: this issue fits a toxic-gas-detection or PSM software partner.)