Explosion due to a release from a hydrogen storage tank

on a Sunday morning, in a ceramic factory, a leak on a 100 m³ tank containing 370 kg of hydrogen caused an explosion. The generated pressure wave causes significant damage to exterior buildings (broken windows, fire starts); a fragment of the tank was found several hundred meters from the place of the explosion. The human toll shows 23 people from the public who were slightly injured. A fire broke out soon after on the site and threatened storage of acetylene and hydrogen fluoride. A safety perimeter of 500 m is set up, road and rail traffic are stopped, the population evacuated.

The tank at the origin of the accident had been put into service in December 1982, after having been modified to notably increase its storage capacity. Its first regulatory check after 5 years of use had revealed nothing abnormal, the second was to take place a few months after the accident. Operating at a maximum operating pressure of 44.1 bar, the storage was replenished as soon as its pressure fell below 15 bar (several times a week). The last loading by an external company had been carried out less than 2 hours before the explosion.

Fatigue corrosion was the source of the leak that caused the explosion. The modification work carried out on the tank and in particular the removal of the roof along the weld had caused a deformation of the tank (oval instead of circular section) and induced tension in the material. Frequent filling of the repository only accelerated the weakening process of the tank.

INITIATING cause was the hydrogen release caused by fatigue corrosion of the tank material (valve, welded area).
On the root cause can only be speculated: the modification work had induced a change of shape of the tank, this could be related to a operation cause or even a design cause.

hydrogen storage tank

The tank at the origin of the accident had been put into service in December 1982, after having been modified to notably increase its storage capacity. Its first regulatory check after 5 years of use had revealed nothing abnormal, the second was to take place a few months after the accident.
Operating at a maximum operating pressure of 44.1 bar, the storage was replenished as soon as its pressure fell below 15 bar (several times a week). The last loading by an external company had been carried out less than 2 hours before the explosion.

MATERIAL LOSS: Outside the establishment the explosion caused about 850 cases of material damages (fires, broken windows, etc.).
COMMUNITY DISRUPTION: The police delimited an area (500 metres large) around the plant. The rail traffic was stopped and the road traffic deviated.

One of the factors contributing to the failure of the tank was the elimination of the roofing of a original weld during modification works. Despite inspections had been regularly executed, these were not capable to detect crack growth there during inspections.
The incident represented a milestone in the way how welds were handled and the occasion to step up efforts to understand welds degradation under hydrogen and to improve standards and regulations related to weld behaviour under dynamic loads and quality assessment of welds.

Since then, weld performance and deterioration assessment received more prominent attention, because of the difficult to establish a method able to relate the load on the materials to the state of the weld seam, and, from there, to be able to predict service life of the vessel.
Moreover, the periodic testing program had to be designed so that fatigue cracks could be detected. Instead of the traditional hydrostatic pressure test, instruments such as dye penetrant testing, ultrasonic testing, magnetic particle testing, or acoustic emission testing were more promising but needed further development and standardisation. The methods could be made more conclusive if used together, for an example ultrasound method carried out in conjunction with an acoustic emission test.