Around 11:30 am, local residents alarmed by an unusual noise notified a plant that was producing uranium tetrafluoride from uranium ore: destruction of the dike wall on the No. 2 ponding and evaporation basin 180 m long and 15 m high caused all of the liquid held in the basin to be released, while the solid phase was no longer confined either. 15,000 m³ of process water and 10,000 m³ of sludge heavily loaded with nitrates (ammonium, sodium, calcium) and metal hydroxides (Fe, Mg, Mo, V) spilled in a field some 10-ha in size belonging to the operator, yet without reaching the TAURAN canal or leaving the site. The majority of the discharge remained concentrated in a 4-ha area covered by 30 to 40 cm of material, at the base of basins 3, 5 and 6.

Production activity was suspended. The inspection authority for classified facilities recorded the incident first hand the very same day. On 22 March, the operator undertook construction of a bund wall surrounding the affected zone in order to avoid: sludge leaching, flow into the TAURAN canal, and intensification of the environmental impact.

A prefectural order adopted 23 March ordered: an immediate suspension of all effluent discharge towards the basins; confinement of the damaged basin; enhanced monitoring of the natural environment as well as surface water and groundwater sources; an expert assessment of basin stability; and submission of an accident report. The temporary sludge retention installation on Basin 2 included, in addition to the first bund wall, a second water-permeable wall that retained sludge below the level of the basin plus a ditch for collecting filtered water through bund walls by means of reinjection pumping to intact basins.

The discharge caused a considerable drop in the hydraulic pressure exerted by Basin 2 contents on the Basin 3 enclosure. This abrupt variation threatened the durability of Basin 3 as well as Basins 5 and 6, which were contiguous. A public agency specialised in dam monitoring, along with 2 consulting firms, evaluated the stability of the entire facility. The monitoring of Basins 3, 5 and 6 by three weekly topographical recordings (16 blocks), completed by inclinometers placed on the dike crest, visual inspection of crack evolution and piezometric readings inside the basins, revealed no potentially harmful movement.

The physicochemical analyses ordered by the operator detected a rise in the rate of nitrates in the TAURAN Canal, with a maximum of 170 mg/l on 23 March before returning to the nominal value of 50 mg/l on 29 March (these levels were deemed relatively harmless to the environment). The contents of heavy metals (including uranium) were comparable to those observed under normal site operating conditions. The radiological analyses conducted at the parcel boundary indicated an equivalent annual dose generated by the sludge equal to 1 mSv. On 26 March, the operator physically isolated the various zones accessible to the public that were located adjacent to the property boundary.

At the end of March, the operator treated the polluted surface by the use of lime. The polluted sludge and soil were then scraped and placed in definitive storage in Basin 1. On 31 March, a prefectural order authorised the resumption of activity after 12 days of downtime. Since environmental impacts were contained in both space and time, no judicial action was proposed.

Lateral leaks on the body of the dike, composed of tailings from the former sulphur extraction operation, had been detected since 1980, resulting in the installation of a network of drains and piezometers. Impact studies on surface water and groundwater, conducted in 1991 and 1992 at operator request, had concluded the partial permeability of the bottoms of Basins 1 and 2, imposing extensions to the piezometric monitoring network. A follow-up report, dated March 1998, assessed the nitrate-polluted water discharges (77mg/l) of these basins towards the CADARIEGE Canal at approx. 10.5 m³/day: the operator then diverted these discharges into the TAURAN Canal to eliminate the problem of chronic surface water pollution.

The deep cracking of Basin 2 sludge due to the intense heat wave of summer 2003 and heavy autumn rains enabled basin water to infiltrate into the dike body, causing its saturation and a rise in pore pressure, ultimately leading to structural failure. On 15 March, the operator had observed a 15-m long longitudinal crack at the dike crest on Basin 2 and duly informed a consultant, who fearing instability throughout the entire structure had not yet submitted the study’s conclusions at the time of the break.