In a 3,000 m² vinegar factory, a 150 m³ tank containing 40 m³ of alcohol vinegar exploded at around 9:40 a.m. while undergoing welding work. The tank, measuring 13 m in height and 4 m in diameter, was projected like a missile. It flew through the roof and landed 80 m from the building, on the Nîmes-Grau du Roi railway line. The emergency rescue services took 2 injured employees to the hospital, one of whom had been badly burned. Rail traffic was interrupted. The spilt alcohol was collected in the holding tank. The building adjoining the tank was severely damaged and was near the point of collapsing. The adjacent tanks, not anchored to the ground, had become deformed as a result of their displacement and being thrown into one another. The firefighters extinguished a fire in offices, probably caused by damage to the electrical installation.
The plant was in the process of coming online following the relocation of the site. The pipework assembly operations had not yet been completed. Due to the lack of storage space, the work-in-progress was stored in the tanks that were waiting to be connected. Tanks containing liquids with less than 141% alcohol could be assembled without having to be emptied first.
The tank in question had been used to recover a mixture of alcohol from transfers made five months earlier. It was estimated to have an alcohol content of 9%. The planned operation consisted in mounting two support legs to maintain the pipes in place. One leg was to be installed at 2.10 m above the ground (1 m below the liquid level) and the other leg at 5.70 m (2.60 m above the liquid level). The explosion occurred when the 2nd weld had just begun. The welders heard a decompression noise a few seconds before the explosion.
The operator’s post-accident assessment concluded that the mixture in the tank had an effective alcoholic strength near 20%. The assessment carried out by the expert showed that the operation performed on the tank below the liquid level most likely resulted in localised heating, bringing the temperature of the liquid from 30 °C to a temperature close to the flash point (36 °C for a 20% mixture). This rise in heat undoubtedly created an explosive atmosphere in a part of the tank’s vapour space. The 2nd welding operation at the gas phase caused a hot spot sufficient to ignite the mixture. The study showed that only 10 to 20% of the tank’s vapour space, i.e. at the LEL concentration, can cause the effects observed. Another possibility is that the current used for the welding operation resulted in an electroerosion phenomenon with the generation of hydrogen since the stainless steel tank contained acetic acid (diluted acids can attack iron, resulting in the release of hydrogen). If the ethanol vapours already present in the vapour space are taken into account, it is possible that just a small amount of hydrogen (4% LEL) was sufficient to make the whole mixture flammable.
From an organisational point of view, the accident resulted from a combination of several causes and circumstances:
- an error in the assessment of the alcohol concentration. The mixture had been produced five months earlier. It was atypical, since the water/alcohol mixes habitually used at the site do not exceed 11% alcohol. It had not been identified as such.
- work had begun without a prevention plan or a fire permit, in violation of the prefectural authorization order,
- commissioning of the plant without conducting the compliance inspection provided for in the prefectural authorization order,
- a vapour phase concentration, before work had begun, close to 2/3 of the LEL due to high external temperatures (35 °C).
Also, the system had all the conditions required to generate the observed effects:
- pressure-resistant tank, robust at the ferrule/dome connection,
- elongated tank, and compression of gases during the blast resulting in the rupture of the ferrule/bottom connection,
- presence of liquid at the base of the tank, ejected at high speed, increasing the propulsion efficiency.
However, taking into account the available energy, the shock wave was dampened due to the amount of energy required to rupture the tank, the roof and the cladding and to propulse the tank, thereby limiting the consequences of the explosion.
The commissioning of the new site, scheduled to take place in late August, was pushed back.