A container stack is a column, and a column cares about where its weight sits.
Loading rules exist to keep the heavy boxes low, so the mass presses down through the stack into the deck and the lashing gear only has to stop things leaning.
Reverse the order and the gear is asked to do a different job.
Nothing on deck looks wrong when that happens. The boxes are the same size and the stack is the same height.
The difference is in the arithmetic, and the arithmetic is written on a piece of paper ashore.
The failure is already loaded.
Weight at the top is a lever, not a load
Lashing systems are rated for a stack that gets lighter as it goes up. Twistlocks hold the corners together, rods and turnbuckles brace the lower tiers, and the whole arrangement is calculated against a known distribution.
Put the heavy boxes on top and the numbers change in two directions at once. The center of gravity rises, so every degree of roll swings more mass further out.
At the same time the restraint at the bottom is unchanged, which means the forces arriving at the corner castings climb well past what they carry.
The gear does not warn anybody first. Twistlocks and rods either hold or they do not, and the transition between those two states takes no time at all.
A stack loaded this way can cross an ocean without incident and then fail in conditions the ship has handled a hundred times.
The margin is quietly gone.
Waiting for a berth with the sea on the beam
The vessel was a 984 foot container ship rated at 7,471 twenty foot units, holding position rather than steaming.
She was not caught in a storm. Wind was 20 knots from the northwest, the sea was running about 8 feet and the swell about 11 or 12, which is an ordinary winter afternoon on that coast.
What mattered was that a ship with no way on cannot choose her heading, so she lay across the swell and rolled through 18 degrees at the extremes.
That is uncomfortable rather than dangerous for a hull of that size. For a stack that is heavier at the top than the plan assumed, it is the load case that finds the weak point.
Somewhere in the dark the gear in bay 42 let go, and the crew found out when somebody walked out on deck and saw dust in the air.
Nothing else raised an alarm.
One number, entered by hand, thirty nine times
Federal investigators traced it to the loading plan drawn up ashore before departure from Los Angeles on February 6 of 2024, with the loss occurring at 2135 that night about 94 miles south of Oakland.
A fault in the booking system meant the cargo weight for 39 containers had to be keyed in manually, and every one of the 39 was entered as 2,500 kilograms, roughly 5,500 pounds.
Their real weights ran from 54,000 to 63,000 pounds. Each box was therefore carrying around 20 tons that the plan did not know about.
In total about 920 tons went unrecorded, of which roughly 457 tons sat above the hatch covers in bay 42, which is exactly where a stack is least able to absorb it.
Twenty three containers went over the side and another ten were damaged, with cargo and repairs put above 735,000 dollars.
Nobody lost the boxes twice.
Why the ship could not have caught it
The part worth dwelling on is that no instrument aboard was capable of noticing.
Departure mean draft was about 31.6 feet, and in that condition it takes roughly 260 tons to push the ship one inch deeper, in the way that a wrongly described load or an instruction nobody followed passes every visual check.
So 920 hidden tons put her about three and a half inches deeper than the plan predicted, and company policy already allowed a difference of 0.7 feet between calculated and observed drafts to cover ballast, fuel and ordinary variation.
The full sequence is set out in the marine investigation report including the weights and the stack positions.
The discrepancy was smaller than the tolerance.
The rule that was supposed to prevent this
Since the mid twenty tens every export container has needed a verified gross mass before it can be loaded, a rule written precisely because misdeclared weight was sinking stacks.
That rule fixed the weighing. It did not fix the path the number travels afterward, through booking systems, forwarders and planning software, where a single field can be overwritten without anything downstream objecting.
A weighbridge produces a fact. A planning terminal accepts a keystroke, and the trade coverage put the failure squarely there.
The drifting containership behaved exactly as loaded.
The plan described a different ship.
Read the whole thing?
Get the week's signal, not the noise
Our sharpest reporting on energy, climate and nature — free, once a week.
