At around 1:10 p.m. in a chemical plant, the temperature and power required for the mixing process in a 9 m³ reactor increased during batch production of polyester resins which involved the introduction of epoxy resin with methacrylic acid. This indication of high viscosity in the mixture characterises the start of accidental polymerisation in the mass in a reaction medium. The technician alerted the production managers who activated the reactor’s internal cooling system and decided to spray it down from the outside with 2 emergency hoses. Once alerted, the management went to the workshop at 2:45 p.m. Management decided to push depleted air (100%) into the reactor to obtain “bubbling” in the matrix and empty the methacrylic acid loading hopper to prevent any further introduction of reagent. At 4 p.m., the temperature had exceeded 250 °C, which is the detection limit of the sensors installed in the reactor. At that time, the manager decided to optimise cooling via the exchanger, switched to total nitrogen bubbling to avoid ignition, stopped polyester production and placed the reactors in fallback mode. He shut down the VOC treatment unit and switched to a cold chimney. Then he checked the site’s firefighting water capacities to continue cooling, lock out electrical equipment to isolate the reactor and set up overnight monitoring.

From 2 a.m., the temperature dropped to 110 °C by 10 a.m. The VOC treatment unit and the polyester production were restarted. The reactor returned to room temperature 2 days later, and the alert was lifted. As the product had been polymerised in mass in the reactor, the latter had to be cleaned and refurbished. This process took over 2 months and cost 100K€. Commercial losses were estimated at 60K€, and customer deliveries were delayed 2 to 3 weeks as they had to be handled by a subsidiary located abroad.

On the day before the incident, the reactor was loaded with epoxy resin, but measurements to validate the “epoxy index” indicating the quantity loaded and the functionalities did not provide a compliant result. By applying the standard manufacturing procedures, the technicians were able to carry out additional measurements and contacted the R&D department for verification. The measurements performed by the R&D department showed that the “epoxy index” was correct but there were indications that there had been a deviation in the measurement method used by the production department. During these various validation operations, the resin had remained in the reactor at 100 °C for more than 16 hours instead of 1 hour, which probably changed its characteristics and promoted polymerisation. The following day, after launching a batch at 12:00 p.m., a control sample could not be taken at 12:57 p.m. since the sampling port was blocked by reagents in the process of curing. The stirring system was then shut down for several minutes in order to take a sample through the reactor’s manhole, which may have accelerated the accidental polymerisation reaction.