Liquid hydrogen release from a rail tanker

A liquid hydrogen rail tank started venting hydrogen, and almost the whole content of the tank, was bout 2600 kg of liquid hydrogen, was lost over a period of seven hours via the venting line of the tank. The wagon with the tank was uncoupled from the goods train and shunted to a siding to allow the release of its content.

A micro-crack in the outer wall of the tank caused the loss of the vacuum and was the initiating event of the release. The loss of the interspace vacuum degraded the related thermal insulation function, with a quick increase of the conductive and convective heat transfer between the tank interior and the environment. The liquid and vapour hydrogen raised the internal pressure to the design value of the safe valves.

The valves were designed to maintain a constant 12 bar pressure differential between the upstream and downstream sides of their shutter (thus the max pressure attained inside the tank during the accident was approximately 13 bar).

The calculations of the referred source (Cancelli et al., Cryogenics 2004) showed that:
1. The released mass was 2490 kg, corresponding the 98% of the one originally stored in the tank.
2. The tank internal hydrogen temperature raised slowly for the first 2.5 hours, until the moment it reached the critical temperature (Tc). Until that time, the incoming heat was used as latent vaporisation heat. After the attainment of Tc, the incoming heat became sensible heat and the temperature raised steeply.
3. The hydrogen flow rate through the safety valves peaked approximately at the same moment (0.4 kg/s, approximately after 2.5 h), corresponding to the moment in which the densities of the gas and liquid phases in equilibrium become similar.

INITIATING CAUSE: the loss of vacuum and of the related thermal insulating capacity of the cryogenic tank was due to the cracking of the external layer.
It is unknown if the crack was caused by the collision with an object, or a design-related event (for example too high local stresses or material ageing)

vacuum thermal insulation

DESCRIPTION OF THE TANK COMPONENTS
The cryogenic tank was of the Dewar type with 70 alternate isolation layers of dextar and aluminium under a vacuum of the order of 10-2 Pa. Its intenal voulme space was 42,000 l.
The internal and the external tanks were kept in place by stainless steel tie rods at the front and rear. They were 61 cm long and 2.5 cm in diameter. The neck of the tank was held by stainless steel tie rods 43 cm long and 2.2 cm in diameter (15 at the front and 15 at the rear).
Four spring-loaded safety valves (two serving as reserve valves) were installed to come into operation at 13 bara to protect the inner tank from excess pressures.
The venting line ran for about 6.5 m before reaching the safety valves, partly in the inner tank and interspace, and the remainder in contact with the air. Downstream from the valves, it was a vertical pipe with a short, horizontal fluted end to prevent rain access.

The loss of 300 kg of hydrogen.

The onsite fire brigade arrived on the scene immediately after the explosion. They first searched for injured persons and did a headcount to check that everyone from in and around the plant was accounted for. The emergency response control centre also requested the intervention of an emergency doctor and ambulances. Smaller cable fires in the plant were extinguished from turntable ladders.
Measuring vehicles were unable to determine the concentration of toxic substances.
The police provided assistance and checked the application of pollution control measures.
The extinguishing water was channelled to the internal clarification plant for treatment.

The safety design of the tank demonstrated its soundness. Despite a massive increase of the temperature inside the tank, the structural integrity was never compromised and the safety measures worked as designed in depressurisation and venting. The slow release of hydrogen, despite the high quantity, never constituted an explosion risk.

Nevertheless, the thermo-physical and fluid-dynamics phenomena involved in such a case are extremely complex. To ensure a similar level of safety also in cases of higher heat flows, these phenomena need to be studied in detail and mastered by design, to further reduce probability of occurrence and possible escalation.