The following is roughly what I intended to say during last nights Disastrous Dinners talk.
I didn't stick too strictly to the script, but you get the idea. There was a good discussion after the talk with extra questions.
If you want to attend the next talk in September, sign ups are now available: By Gum - A Design Error
The night of the event
John Sheptor joined imperial sugar in February 2007. Previously he had worked in various management positions at different chemical companies, and had joined imperial sugar as Chief Operating Officer, with the intent of him receiving a handover before becoming CEO at the next annual meeting at the end of January 2008.
Imperial sugar was headquartered in Texas in a city called ‘Sugar Land’, though they don’t do any sugar processing their anymore. Imperial sugar had two manufacturing sites, one in Louisiana and one at Port Wentworth, Georgia which they bought in the late 90s.
On the 7th of February, at 7:15pm, Nine days after becoming CEO, John Sheptor was getting a tour of the Port Wentworth facility. I never did work out why he was touring the factory at that time of night.
John and three other employees were walking towards the packaging plant, a 4 story building that would take white granulated sugar and convert it into various final products, then package them ready for distribution. As they approached the building they heard a thud. They thought it sounded like someone had dropped an industrial roll of packaging paper from a forklift. Three to five seconds later, a large explosion blew open the doors of the packaging plant and knocked the tour party off their feet.
Security cameras recorded a giant fireball above the factory. The blast buckled many of the brick walls and concrete floors blocking exits. The water pipes that lead to the sprinklers were sheard, rendering the fire suppression system in the buildings ineffective.
Eight people died that night. Another six died from their injuries in the subsequent days. A further 36 were treated for burns in hospital.
The fires in the silos continued to smolder for another week before they were completely extinguished. The silo complex and the packaging plant were destroyed in the explosion and subsequent fires and required completed demolition and rebuild from scratch.
What a start to a new job.
I am sure you remember the experiment with flour in school. Flour sitting in a heap on a spoon will smolder away slowly, but give it a burst of air and create a dust cloud you get an impressive flash. For dust to explode, it must be in the form of a cloud. In needs to have easy access to air to burn instantly.
But here is the thing: Dust clouds are actually quite difficult to form. To get to the stage of making it explosive, the dust cloud needs to be thick enough that visibility is reduced to less than a meter. In an open environment, such as this room, the dust would likely settle out quickly, or the general ventilation would move it on.
Yes you can get explosive clouds but they tend to be small or short lived. But the small ones can become bigger.
That small explosion may not be enough to do much damage on its own, but what it can do is lift any sitting dust into the air. Any dust that is on a surface such as a floor, equipment, light fittings or beams, can get lifted by the pressure wave and then ignite, causing a much larger explosion.
The secondary explosions are the ones that do the damage. That is what happened at Imperial Sugar, the large quantities of sugar that were sitting around got lifted and set of a series of cascading explosions that destroyed multiple buildings.
You need to keep good housekeeping standards to ensure there is no fuel for secondary explosions.
The facility was built in the early 1900s and started initial production in 1917. Over the years, the plant was expanded and modified. Extra equipment was added or upgraded.
In the 1960s, a note was sent from an engineer to management stating that the problem was so serious and dangerous that it was knowingly wrong. The note also said that no outside help is required. In other words - we know we have a problem and we know what we need to do about it.
Unfortunately it didn’t get any better
Over the years, a number of small explosions occurred, though none triggered any secondary explosions. There were procedures and policies detailing the housekeeping standards and inspection regimes, however these were never properly enforced.
Piles of sugar regularly built up and the air was often filled with more dust. It is difficult to keep on top of housekeeping when more dust is continually being generated. The equipment wasn’t designed to seal the sugar in, just to keep contaminants out.
There are pictures of the site before the explosion showing the thick layers of dust on the plant. Tellingly, in one of the images, an electric motor is shown. Someone had installed a metal shield or roof over the motor to stop sugar dust landing on the top of it.
A report by an independent contractor was delivered less than a week before the event detailed the failings of the ventilation systems:
- System was undersized
- System was in a poor state of repair
- Many of the extraction lines were plugged with sugar
The final factor that lead to the explosion was a change made to the plant in the year before the accident. When the factory was first built, food standards were less of an issue. The conveyor belts that transported the sugar below the silos was uncovered and posed a contamination risk. Items could fall onto the belt and get into the product. To stop this, in 2007 a cover was installed over the belt. This kept debris out of the sugar, but was not installed with explosion vents or connected to the dust extraction system.
Previously if a blockage caused a dust cloud to form, it would dissipate quickly because of the general ventilation within the corridor and the relatively large volume of the corridor. Now the dust cloud was trapped within the covers, it was able to build into a dense enough cloud to exide the explosive limits. Eventually the cloud found an ignition source, believed to be an overheated bearing. This lead to the initial explosion within the casing. This disturbed the sitting dust and lead to the next explosion, which continued to cascade through the complex.
This is a case of Normalisation of Deviance.
Despite the company being aware of the dangers and having housekeeping policies in place, it became normal for things to be covered in dust. It became accepted that the equipment leaked and dust would build up. Even if someone did get the place clean, it would be covered in dust by the next day so why bother?
The tasks that would have allowed them to keep the place clean, improving the dust extraction systems, were important but not urgent. For at least 40 years, they were aware of the problem, but never got round to fixing it. Something more urgent kept coming up.
It has waited years already, what difference does one more day make? It was important but not urgent. Until it was too late.
What other problems are important but not urgent? What other problems are not a big problem for now but may suddenly become a very big problem?
The talk was primarily based on the CSB investigation reportGo Top