In a PVC manufacturing plant at around 2:30 am, the rupture disk of an autoclave burst resulting in the release of a reaction mixture into the atmosphere (3.75 tonnes of PVC powder and 3.75 tonnes of VCM). On 28th February, a new production run began at 10:20 pm. Polymerisation started at 11:40 pm (expected duration: 2 hrs, 50 min at P = 7.9 bar). At 2:24 am, five minutes before the final breathing (weight of the reactor: 25.5 t), a pressure rise occurred. At 2:25 am, pressure equalled 8 bar; the automatic pressure control system controlled opening of the condenser’s gas extraction valve. At 2:29, the high pressure alarm was activated at 8.3 bar and the water valve on the condensers was opened, but the pressure did not drop. At 2:31, the high pressure alarm was activated at 8.7 bar. At 2:34, reactor pressure was 10.5 bar and manual injection of the inhibitors did not lower the pressure. At 2:36, the two defective rupture disks burst at 15.2 bar. The switch to backup disks enabled isolating the reactor. The reactor weight was 17.7 t and the pressure was 3.8 bar. At 2:52, the other workshop autoclaves were cooled to stop these reactions. Between 2:36 and 3:57 am, workshop chromatographs detected the presence of VCM (0-62 ppm). Atmospheric controls performed at around 5 am by the on-call team both in the workshop and at the site boundary turned out negative. The PVC powder that had fallen onto the site was collected and placed in drums for disposal, but some exceeded the site limitations. During the incident treatment phase, the waste was diverted to the 3,000-m³ pit. The Classified Facilities Inspectorate was notified at 3 pm and recorded the pertinent facts. A press release was issued at the end of the day by the operator.

The next day, removal of domes from the condensers and opening of the filter confirmed the presence of powder in the equipment. The installation was restored and the other workshop autoclaves restarted in the afternoon. The low density of PVC powder manufactured was the reason behind this incident: the abnormal rise in the reactor level led to a partial blockage of the condensers and a deterioration in heat transfer quality. The operator had decided to introduce a set point to better monitor power take-off of the stirring motors during VCM transformation into PVC (liquid mixture, then a mechanical strain due to the formation of PVC): after 2 hours, a low power take-off indicated a far too low density of PVC powder. The operator was also exploring the possibility of adding a low power take-off alarm to warn of the risk of change in density, along with the introduction of a larger amount of densifying agent at the start of polymerisation. The Inspectorate required the operator to provide an accident report and model the explosive areas of the VCM cloud.