Process Wednesday: a bad day at the sodium plant

From a French government list of industrial accidents, an anecdote of a fire at a sodium metal packaging facility: 

A fire broke out in a sodium (Na) packaging workshop at 6:45 pm on valves on top of a tank having a potential capacity of 52 t of Na which was fortunately empty at the time. This tank was located next to another tank containing 15 t of Na. These buffer tanks are pressurised at 7 bar in order to transfer Na to the chemical plant and the empty tank, which was to undergo regulatory ten-year testing, was in its final cleaning stage prior to being flushed with nitrogen. 

Three firemen emptied roughly twenty powder extinguishers located in and near the workshop, then threw 40-kg sacks of anhydrous sodium carbonate onto the two torches formed, but the fire was not extinguished. They then attempted to drain the oil system (4 m³) into a buffer tank, but the process could not be completed due to the geometry of the piping. External emergency rescue services arrived in several waves. Attempts were made to place a massive quantity of sodium carbonate on the fire, now several meters high, but were unsuccessful. Fire nozzles were used to project nitrogen on the fire in an attempt to rarefy the oxygen locally. A platform was installed on the tank at around 9:30 pm in an attempt to retain the carbonate. The system was finally purged 30 minutes later by means of a small pipe and the fire's intensity dropped rapidly.  

Although the fire remained contained, firefighting operations lasted 3 hours and 20 minutes.
The considerable amount of heat generated ruptured part of the facility's roof made of eternit and damaged plastic elements at the centre of the workshop. At daybreak, an external rescue team member used a tarp to cover the sodium present in the hall.  

The fire was caused by the ignition of hot oil sprayed from a leaky seal. The leak was caused by excessive pressure in a thermal oil coil heated by exothermic hydrolysis of the residual Na in the tank in contact with steam from the cleaning operation. No notable pollution of the natural environment was observed. The operator undertook the following measures: modification of the tanks' ten-year review procedure (water replaced by a different fluid...), study of the complete drainage of the oil circuits, reorganisation of the internal emergency services (internal contingency plan).

This looked like it happened in France in 1987 (search "Saint Marcel" in the document).

I'm no expert on metal fires, but it sure seemed like they threw everything except the kitchen sink at this fire (fire extinguisher powder, sodium carbonate, nitrogen). I don't quite understand the full extent of how the incident happened; did they really use steam to remove residual sodium from the tanks? I assume there's some other solvent that could be used to avoid the use of steam as a cleaning solvent (I dunno, octanol or something.).

Glad I wasn't there.