At 10:00 a.m., an electrical power failure caused the emergency shutdown of several facilities on a petrochemical platform. According to the established safety procedures, products manufactured on the platform were flared off at one of the facilities. As the platform’s steam production plant had also been affected, flaring was not optimal, and large plumes of smoke were visible outside the site. The operator initiated the internal contingency plan at 10:30 a.m., and the personnel of a neighbouring site were confined to the premises. The power supply was then restored at 11:20 a.m., and the internal contingency plan was lifted. The steam boiler was then restarted the following day at around 6:00 a.m., causing the black fumes to stop. It took several days to put the workshops back into operation, resulting in occasional flaring episodes which had been foreseen in the operating procedures.

The classified installations inspectorate was present at the Prefecture, where a crisis unit was activated. The inspectorate stated that the atmospheric emissions associated with the event were unusual, both in their nature and volume. A prefectoral emergency measures order required that the operator submit an incident report within 15 days and conduct a study on the health and environmental impact of the flaring episode.

The association monitoring the region’s air quality reported that several individuals had experienced health symptoms such as eye and nose irritations, headaches, etc. The association set up specific means to monitor the air quality around the site for three days. The analyses made it possible to determine that the local residents had indeed been affected by nuisances (particularly olfactory), sometimes accompanied by symptoms.

A wiring error had caused the loss of the platform’s main power supply during work on the network. A polarity inversion caused the circuit breakers on the platform’s three power networks to open simultaneously. This common-mode (no polarity separation of the three networks) had not been identified before. Restoring power to the networks was hampered by a lack of proper identification of the networks.

Following this accident, the operator implemented the following actions:

• study the total separation of the site’s three main feeders;
• have work involving modifications on the main power supplies validated by a third party;
• improve network identification.