At 9:30 a.m., the operator of a plant that manufactures metal oxides and salts began transferring a nickel nitrate solution to a 23 m³ storage tank. At 10:15 a.m., he realized that the tank’s sampling valve was open and halted the transfer. In the meantime, 4.63 tonnes of 42% nickel solution (i.e. 660 kg of metal) had leaked into the retention basin. But the basin was rendered ineffective due to a ground-level bypass connecting the basin directly to the site’s wastewater treatment plant (WWTP) which had been left open. The plant was switched to a closed circuit but 23 kg of nickel had been discharged into the River Rhône. Sodium hydroxide and clarifying agents were added by hand to accelerate the metal’s precipitation, but production was not shut down. The following day, the level in the WWTP’s primary pond was at maximum level and threatened to overflow. As the water in the settling tank had an acceptable colour and pH level, the operator decided to resume discharging it into the river. The presence of nickel in the discharge (144 kg at 275 mg/l) was not found until the samples collected each day were analysed after two weeks. In all, 167 kg of nickel was discharged to the Rhône, with 30 kg more discharged the following week. No impacts on the natural environment were reported. The inspection authorities for classified facilities were informed.

The operator failed to make sure that the tank’s sampling valve was closed before beginning the transfer. This check is not documented in a written procedure but is part of each operator’s basic training. Not many processes at the site are automated, so most operations are carried out by hand. In addition, unlike the other quarter-turn valves on the site, this valve does not have a cover to prevent actuation errors or involuntary opening. Although it was known that the bypass rendered the basin ineffective, no cap had been fitted. The bypass made it possible to quickly drain the water used to wash the tank to the WWTP (the tank is frequently washed). In addition, the site’s strategy is to quickly transfer leaks of hazardous substances to the site’s WWTP. The basin was therefore merely used as a zero-volume gravity-fed weir.

The partial neutralisation had concealed the settling tank’s acidification, which could have occurred much faster and tripped the automatic closure when the pH dropped below 5.5. Because the effluents were recirculated, it was impossible to treat all the nickel. The WWTP’s capacity (100 kg/h of nickel) was not sufficient to handle a high load under accident conditions (600 kg/h of nickel in this case). The capacity of the flocculation and precipitation tanks was not enough to treat massive pollution of the settling tank in 24 hours. Lastly, the decision to re-open the settling tank was taken only on the basis of the colour and pH level. No analyses were performed to determine the metal ion concentration.

The operator put up signs showing the positions of the valves, made an inventory of the bypasses on the site, raised the pH threshold used to close off the settling tank, and set a minimum volume that must be maintained in the WWTP to ensure longer operation in a closed circuit. A procedure indicating what to do if an incident involving the WWTP occurs was written and the employees received training in implementing the procedure in the event of pollution risks.