A 600 kg/h leak of dimethyl sulphide (DMS, flammable with a characteristic odour) occurred in a thiochemicals plant at a chemicals complex. The leak was discovered at around 6:10 a.m. on the flange of a cap on a pipe used to supply a distillation unit with DMS from a storage tank. The plant operator shut down the unit, put up a safety cordon, and doused the leak with water nozzles to knock down the vapours. The pipe was depressurised and the flange was replaced.

An estimated 7 tonnes of DMS either evaporated or leaked out into the plant over 13 hours. A total of 30 m³ of polluted soil was excavated with a mini excavator and disposed of. The 10 m³ of extinguishing water was pumped using a vacuum tanker. The initial piezometric analyses did not reveal any pollution of the underlying water table. The cable conduits located near the leak were damaged by the DMS. As odours were noticed at the site’s boundary, the plant operator issued a press release.

The leak was caused by a worn seal on a cap. The seal had burst at around 5:00 p.m. the day before, a few hours after it had been moved from the ‘full side’ to the ’empty side’ in order to supply the unit as it was being started up. The operator had not noticed anything unusual while making his rounds at around 5:30 p.m. However, at 9:30 p.m., the complex’s central control station informed the plant’s control room of odours in the unit. The operators thought that the odours were caused by the release of a sulphur-containing chemical a few hours earlier on the flare stack of a neighbouring plant and left it at that.

Starting at 10:00 p.m., the stationary detectors at the nearby plants (but not at the distillation unit) began going off at regular intervals. Two new rounds were conducted, but the area where the leak was taking place was not checked and the various control rooms at the complex reported nothing unusual. At 4:30 a.m., a member of the central control station went on a new round. His portable detector confirmed the presence of a leak in the unit. He alerted the plant’s control room and an operator dispatched to the plant located the leak.

The analysis of the accident revealed multiple compounding causes:

•  The portable detector used by the first auxiliary operator was insufficiently sensitive (5 ppm detection threshold).
•  The area where the leak occurred was not checked during the first round (the procedure was too long and did not cover all at-risk areas).
•  The section of pipe where the leak occurred was inadequately lighted.
•  Unlike the electrochemical detectors used at the neighbouring plants, which are also able to detect DMS, the two stationary semiconductor detectors fitted on the pit around the tank and the cap could only detect hydrogen sulphide.
•  As there was no flow rate alarm for the DMS tank, it was impossible for the operators in the control room to detect leaks in the absence of pressure surges in the tank during the 13 hours of the leak.

The operator implemented a number of actions:

•  It reinforced the rounds procedure and refocussed rounds to at-risk areas.
•  It fitted new DMS detectors that use a different technology, checked the detection abilities of the detectors used for other hazardous products on the site, and revised the design rules that led to the wrong type of detector being selected.
•  It replaced the seals with seals of the same type and introduced a procedure for fitting and performing preventive maintenance.
•  It commissioned a new safety study to improve leak detection from the control room (video camera, additional detectors, etc.).