Release of a light hydrocarbon gas mixture in a chemical product plant

During the start-up operation (phase) of a cryogenic unit, a separator exploded, releasing ethane, ethylene, methane and hydrogen. The bursting of a tank was caused by a 1600 mm long incipient crack which had been formed at the inside of the tank along a weld line (a ferritic basis metal had been welded with an austenitic weld metal).
The hydrogen induced incipient crack (cracking) was caused by the alternating load during start-up and shut-down phases in the course of the operating time.

The INITIATING CAUSE was hydrogen induced cracking originated by the geometric form variation and the resulting plastic deformation in the area of a longitudinal seam weld. This incipient cracking was favoured by the high tensile (strength) austenitic CrNiMnW weld metal with its particularly accentuated martensitic composite (mixed) zone (area) along the transition between weld metal and basis metal and was also favoured by its physical characteristics which differ strongly from the basis metal.

Possible ROOT CAUSES could two contributions: (i) a wrong execution of the welding (wrong choice of materials, inadequate themral management), aand a a lack of inspection / maintenance.

cryogenic unit, separator, weld

At the moment of the release, the plant was in in the start-up phase of a cryogenic unit.

DESCRIPTION OF THE PROCESS
A cryogenic separator in a refinery has the role to separeate heavier components (like propane, butane, and natural gas liquids - NGLs) to liquefy while lighter ones (like methane or hydrogen) remain gaseous,based on the difference between their boiling points. The process involves compressing, and cooling the gas, then flashing (a sudden pressure drop) to create liquid/vapor phases, followed by fractionation in columns where each component is boiled off sequentially.

Large material damages (losses) due to the destruction of plant parts, estimated as approx. 1.5 million ECU. No persons were injured.
The neighbourhood was strongly disturbed by the noise caused by the bursting of the tank and the consequent flaring activities.

The actions are not reported. Time of the fire also unknown.
Since this was a case of immediate release and ignition, It is very probable that the fire was left extinguish, possibly by cooling the tubes, before starting to intervene to recover the vehicles.