An ester was being reduced by sodium borohydride. During the reaction several errors were made, and several warning signs were ignored, with the result that, eventually, the reactor contents completely ran away, with temperatures far exceeding 300°C.
Immediate on-site investigation was carried out
- The Ester should have been added directly to the vessel, it was in fact added to the doser, which filled to capacity.
- Following the ester, was the solvent charge, this was also sent (in error) to the doser which over filled to the drain, leaving a doser still full of (mainly) ester.
- Following these two errors was a shift change. The new shift on seeing the full doser, ran the contents down into the vessel, even though they knew that the doser should have been empty at this stage.
- The vessel was then cooled to -10°C, this took longer than normal (due to the probe not dipping, but there was no sight glass on vessel)
- Sodium borohydride was then added at -10°C, there was no expected exotherm, so the addition was very facile (ester was in fact frozen, and the solid sat on the top)
- Obviously a reaction was slowly taking place, and eventually that was enough to melt the ester, with the result that the neat ester was now mixed with pure sodium borohydride, and the rest is history...
Ignoring the human error of sending to the doser, there were two obvious places where warnings were ignored, which could have prevented this runaway
- The mistake of sending to the doser, should have been flagged by the new shift - if it's not in the process then they should have queried.
- The abnormal cooling time
- The sudden lack of a known exotherm, when the solid was added, should also have been queried
This shows that operators must be trained into correct following of process, and not try to do changes to correct mistakes of others. Plus to try to observe the unusual and report it immediately to a senior person.