In a packaging varnish manufacturing plant, at 5 am flameless combustion activated the smoke detector located in a polymerisation reactor building. This detector sounded an alarm in the control room and at the establishment’s gatehouse. The alerted technicians observed a release of smoke from a flange of the reactor’s heating system. The team leader decided to transfer the contents of the reactor to a safety buffer tank. A second team leader and a technician, working on the reactor’s aluminium metal sheets, noted an already old heat transmission oil leak, and the rock wool (insulating the reactor’s lagging) was hard and blackened. The leak was located on a weld of the reactor’s heating stage second return, at the separation of the half-shell towards the flange collar of the heating coil. It concerned a mineral oil, whose self-ignition temperature is greater than 250 °C. During the batch started on 21/10 at around 3.06 am, this oil soaking the lagging rose to 230 °C and finished by causing flameless combustion in the soiled rock wool. Once the metal sheet had been dismantled, the wool was sprayed using water fire extinguishers, then at 5.20 am, it was removed and soaked in a can filled with water. The situation was considered under control. However, firefighters were alerted by the external company in charge of the establishment’s remote monitoring, as the guard patrolling when the fire alarm went off could not be contacted on the portable radio (poor battery connection). Indeed, the procedure specified that in the event of failure to mobilise the guard, the company had to call the external emergency services. A pollution response unit was sent and went to the “Shipping” warehouse instead of workshop R7, as the pre-determined fire plan that the firefighters had was incorrect. The operator planned to replace the reactor’s lagging with double-jacket lagging by 31/12/2006 with the installation of a gutter in the low section for easy leak detection. A new guideline was established to check that the portable telephone of the establishment’s guard was operating correctly. A test call (made from outside of the establishment) was made on each 18 h shift change. In addition, the pre-determined fire plan (PFP) was updated. It should be noted that this anomaly had already occurred on the same reactor on 16/02/2004. The faulty weld was repaired on Monday 23/10 in the afternoon and the reactor was placed back in service the next morning. The consequences of the incident were minimal, a few oil residues had been projected on the concrete flagstone on the 1st floor of the workshop, and the fire extinguishing water (a few litres) was recovered and treated as industrial waste.