Hydrogen explosion in the chlorine production unit of a chemical plant

The sodium chlorate production unit had restarted after a maintenance shutdown phase. The mercury cells had been under nitrogen flushing since the beginning of the shutdown to avoid the presence of oxygen (prevention of the risk of explosion of the oxygen/hydrogen mixture).
Although the ARIA report (see references) does not mention it, the shutdown probably had been caused by a hydrogen leak signal.

Around mid-day a new hydrogen leak was detected on the nozzle of a cell collector which had just been repaired. This detection stopped the nitrogen flushing, to allow maintenance to intervene safely (avoiding anoxia hazards). Two hours alter a large explosion occurred in the electrolysis room.

The emergency plan was started. Ten minutes later, the operator noticed that the electrolysis cells are still powered up, while the procedure for shutting down the unit foresaw the cutting off their power supply. He then takes the unit to a safe state and drains the installations, then notifies the authorities.

The INITIATING cause seems to be related to the wrong reading of eh voltage of the cells.
According to the ARIA report (see references), the internal investigation showed that the potentiometers of the electrolysis cells displayed zero intensity when the unit was switched off at in the morning at the moment of the first shutdown. This explains why the shift operator did not switch off their power supplies. When the unit was restarted approximately 2 hours later, the arrival of liquid in the cells still under voltage caused the production of hydrogen and oxygen. Later on, the nitrogen flushing of the cells was stopped for maintenance work, and the two production gases accumulated and mixed beyond the lower ignition limit.

electrolyser cell

The plant was restarting after a previous emergency shutdown.

Two operators were evacuated; slightly injured in the ears, they were out-of-work for 5 days.
Material damage was noted in the electrolysis room: cell mats, collector rupture discs, fibrocement sheets on the roof and cladding, some of which were projected a few metres away, damaged pipes.

The ARIA report (see references) does not mention any lesson learnt. While it reports that the asbestos-containing panels of the unit building were replaced, nothing is said on corrective actions. Also the reasons for the first shutdown, and its corrective actions remains unclear.

The failing of de-powering the electrolyser cells during the second shutdown reveal an inadequate operation procedures, and probably a lack of correct risk assessment. usually, this type of accidents are the occasion for a more realistic risk assessment, including a review of the safeguards performance. Nothing is said on this event.