VCM is a flammable and carcinogenic gas, used in liquefied and gaseous form to manufacture PVC by polymerisation in reactors. At the end of the reaction, the reactors are degassed to remove the unconverted VCM. The degassing circuit includes a gasometer that serves as a buffer tank.

Vinyl chloride monomer (VCM) leaked from a VCM degassing and condensation circuit at a manufacturing plant from 2:45 a.m. to 9:00 a.m. A valve in the degassing circuit opened at 2:25 a.m. The low gas level alarm went off and the compressors stopped from 2:45 a.m. to 4:45 a.m. At the same time, the photoionisation detectors (PIDs) in a shop housing the reactors being degassed showed VCM concentrations between 5 and 10 ppm. Technicians realised that the valves on two autoclaves were leaking. They tightened the valves and called the maintenance department. At around 8:00 a.m., a peak of 12 ppm was detected on the roof of a laboratory near the condensing unit. The control room was locked down. At 8:50 a.m., the day shift noticed swirling over the degassing section. At around 9:00 a.m., they manually closed the valve that had been left open. The release of VCM stopped.

The leakage of gaseous VCM over 6.5 hours is estimated at 3.4 t. Due to the wind direction (East to West), the MVC plume was directed towards the interior of the site.

The event was caused by the opening of the gasometer safety valve on the degassing circuit due to lack of air. A congestion formed in the distribution system blocking the passage of air. The operator had not identified that the valve could be the cause of an atmospheric release of VCM except in its safety function

The release lasted 6.5 hours as the valve involved was not immediately identified. The operations team assumed that leaks from the valves of 2 autoclaves were the cause of the release. They interpreted the low level in the gasometer as being due to the failure of the compressor bypass valves. Incomplete instructions, a lack of linkage between the various problems in progress and a lack of information available on the screens in the control room were among the reasons for a cognitive tunnel that prevented the operating team from identifying the leak point.

The operator informed all his operating teams about this event and about the action plan resulting from the fault tree analysis subsequently conducted. The operator plans in particular to:

  • review of HAZOPs for systems with equipment that could lead to a release of MVCs to the atmosphere;
  • add alarms to valves that could release VCM;
  • updating of “alarm sheets” to include situations that have not yet arisen or have arisen only rarely on the site;
  • addition of information on the synoptic views and on the screens in the control room.

Download the detailed report in .pdf format (456 Kb)