Release from a storage at a hydrogen production site

The released occurred at a T-coupling small of a small pipe (possibly around 6 mm in diameter) in a gas storage system consisting of 40 vertical steel cylinders of 50 l each in a container.
At the time of the accident, the hydrogen pressure in the system was 120 bar, the highest value since the installation, well below the design pressure of 200 bar. The pipe was not well mounted and detached from T-coupling at120 bar.
Approximately 32 kg of hydrogen were released in the container during the 1.5 hours from initiation to almost completed blowdown.
The rescue services evacuated the proximity of the site but did not take any other immediate action. They monitored the hydrogen detectors at the fire service panel until it indicated that the concentration was below flammable levels. However, since the concentration was very high during the event, the detectors had been over saturated giving a false negative and thus exposing fire fighters to danger when they opened the container.
There were no ignition, no harm to people or property.

ANALSYIS
The cause is believed to be an incorrectly mounted coupling that detached when the pressure became high enough.
Moreover, pressure testing had not been performed due to a misinterpretation of the legislation: it was understood that a pipe below 25 mm does not require third-party pressure testing. This was wrongly interpreted as no pressure testing was needed. The facility was CE-marked, with a certificate of conformity from the manufacturer and third-party bodies certifying a passed first inspection.

The INITIATING CAUSE was the detachment of a connection under increasing internal pressure.
The ROOT CAUSE was a combination of
(1) Wrong mounting of the coupling. Since the coupling consisted of 3rd party commercial elements, this suggests a lack of knowledge of the site workers.
(2) Failing to execute a tightness test. This was the result of a wrong interpretation of the safety regulations.
A FACTOR which could have been CONTRIBUTED to escalating the incident was the wrong reading of the safety sensors, which saturated due to the high hydrogen concentration and was giving probably a zero signal, upon which the first responders decided to intervene believing that atmosphere inside he container was below the LEL value.

threaded connection

DESCRIPTION OF THE FACILITY
The stationary storage system was part of a renewable hydrogen production plant made up of a PV farm and a water electrolyser.
The storage consisted in a 20 ft ISO-container, and within 40 compressed hydrogen bottles placed vertically. The bottles were standard steel compressed gas bottles of 50 l and a nominal pressure of 200 bar.
The T-coupling which leaked was a compression-grip tube fitting.

32 kg of hydrogen were lost.

The rescue services evacuated the proximity of the site but did not take any other immediate action. They monitored the hydrogen detectors at the fire service panel until it indicated that the concentration was below flammable levels.

(1) For the installation/assembly of this type of facilities, a specific training on the specific coupling techniques is needed.
(2) It is necessary to increase the awareness about pressure testing requirements, also for small pipe dimensions with a nominal diameter smaller than 25mm.
(3) Serial connection of large number of pressure vessels (40 in this case) causes a large number of potential leakage points. The design of such a storage facility should consider minimisation of the number of threaded connections.
(4) It is not correct to use hydrogen safety detector to measure hydrogen concentrations far above the lower ignition limit (LIL). Safety detectors are designed to give a series of alarms when hydrogen concentration in air increase above determined threshold, for example at 10% and 50% of the LIL. Unless they are designed for the whole range of concentration up to 100% of hydrogen, they will saturate at a certain value and may give false negative signals.

The fire fighters were thus exposed to a risk of explosion when they opened the container.

All T-connections on the 40 bottles in the array were checked and re-mounted according to specifications. After which, the system was pressure tested but found not leak tight. The pressure test was therefore terminated and the facility taken out of service awaiting additional actions.
An inspection of the authorities was still ongoing at the moment of the reporting (