This is an unusually good USCG investigation report.
Ship was US flag in gasoline trade between Houston and Tampa.
Ship was repairing a leak between aft ballast tank and 5P cargo,
when she exploded.
Leak was discovered loading Houston,
but they went to Tampa anyway, returned in ballast
to do the repairs anchored inside Galveston Bay.
Despite the fact that she was fitted with inert gas,
5P was not inerted before purging.
The tank was purged by ballasting up
and then draining.
In so doing the tank was overflowed through
the tank lid, but the butterworth openings
were not opened.
This could have left some
pockets of gasoline and gasoline vapors in the tank
in the highest points of the tank.
One calculation indicated that a 1.1 barrel wedge
of gasoline could have existed at the top forward
corner of the tank.
The NTSB calculated that this would have been
enough to put that tankover the LEL
it it totally valporized.
OMI consultants claimed that you would need
2 barrels of gasoline to do this.
The report points out that 5S which had been purged
in a similar manner blew up 2 hours later.
After the tank was drained to a level below
the repair, the tank was closed up.
and not subsequently tested.
Nor was it inerted during welding from the other side.
This was done to save a little off-hire time
by allowing hose testing from the 5P
immediately after completion of the job.
[Also the IG would be dangerous to the repair
team if it leaked into the ballast tank.]
This area of hull structure had a history of leaks.
The ship had also just repaired
two leaks between 5C and the pump room.
In one case, they used a doubler [not Class legal]
because as the Master said the bulkhead was too thin
to gouge out.
The ship has just completed a docking
and an ABS and USCG survey, two weeks earlier.
During this docking, some side shell stiffeners were renewed
in the aft port ballast tank.
The ABS surveyor allowed the repair to be discontinous
at the bulkhead with the new stiffeners terminating
at a landing plate welded to the bulkhead.
[This is contrary to Class rules].
It was a leak at one of the landing plates
that was being repaired at the time of the explosion.
The USCG too some thickness measurements in the bulkhead.
At the level of the landing plate
the 11 mm bulkhead had on average wasted down to less
Other measurements ranged from 22% to 47% wastage.
The lading plate weld was badly undercut,
reducing the bulkhead thickness by more than 70%.