Release of hydrogen in a gas cabinet of a laboratory

The event occurred in a research laboratory on renewable energies and decarbonisation. The Lab Manager had swapped a hydrogen cylinder, together with the delivery personnel. They secured the cylinder, located in a fire-proof cabinet, and performed a leak test by means of the soapy water method, after which the the gas cabinet and gas room were closed.
Less than 2 min later, a high-level alarm (10% or over of the Lower Explosive Limit (LEL) activated, mandating the evacuation of the personnel. The alarm was issued by the explosive gas sensor in the hydrogen cabinet, measuring 37% of the LEL of hydrogen.
The Lab Manager could monitor the sensor reading remotely; when he noticed that the hydrogen concentration was not increasing, s/he eliminated any personal item which could provide a source of ignition and re-entered the gas room and sealed the hydrogen cylinder. The gas rapidly dissipated, and 40 minutes after the alarm personnel was allowed to re-enter the building.

The INITIATING CAUSE leak was caused by the difficulty to tighten the connection, due to the use of steel instead of brass in the cylinder connector. The Lab manager and the assisting personnel had to use abnormal force. The leak was probably too small to show bubbles during the soapy water test. A prolonged soap test would have probably been able to detect the small leak.
This would have avoided the need to re-enter the room and the cabinet, which was containing more hydrogen than just after the installation of the cylinder.
The ROOT CAUSE was a shortcoming in the procedure related to the installation of the cylinder and the leak test. A CONTRIBUTING CAUSE could be identified in the failure to realise the increasing difficulty of the tightening, which could bring to some leak, and to notify it timely to management.

nothing

DESCRIPTION OF THE SAFETY SYSTEM OF THE LABORATORY
The laboratory had a safety system based on 111 gas detectors, all of which serviced quarterly, and all of which able to trigger the automatic closing of all gas lines via solenoid valves. A control panel located outside of the laboratory allowed the gas levels of any sensor to be monitored remotely. Alarm conditions were immediately highlighted, making easier the immediate identification of deviation from normal conditions.
Gas cylinders were stored in two gas manifold rooms equipped with 4-hour fire rated doors .The rooms were not accessible to general staff. Cylinders of flammable, explosive and oxidizing gases were kept in explosion-resistant and fire-proof cabinets. The gases were piped through flashback resistors and coaxial cables able to detect automatically damages or leaks, and to trigger the closing of the solenoid valves.

No damage

The high-level alarm on 10% LEL triggered an automatic evacuation of the personnel.
The leak could not be stopped by the automatic shutdown od he hydrogen supply system, because it occurred at the hydrogen cylinder. Therefore, the Lab Manager had to perform a in situ manual operation.

The laboratory affected by this event had a well-designed safety system. The safety system worked as designed, and it was possible to adopt mitigating measures which avoided escalation.
Nevertheless, the system showed some vulnerability to leaks located before the automatic supply shutdown system, in combination with the increasing difficulty in executing the tightening procedure. Moreover, the procedure / instructions did not consider a scenario of a slowly developing leak.

This case highlights how challenging is to perform an effective safety design of a complex set of activities and operations, by considering all possible scenarios. The case also reveals the importance of a timey feedback from operations, provided by workers made aware and trained on hazards and risks. The difficult of the tightening should have been reported well before the occurrence of this event, and this feedback could have brought to an earlier improvement of the connection.

On the persistent shortcoming of the soap test, see for example the old event HIAD_038.

(1) The procedure regarding the exchange of cylinders were updated, by requiring that after changing flammable, oxidizing or explosive gases, personnel shall remain in the gas supply room for two minutes. In the event of a leak, staff would then be able to close the main cylinder valve, before proceeding to evacuate personnel.
(2) The new procedure was communicated to the laboratory staff by updating the training material and by installing boards on the doors of the gas supply rooms.
(3) The laboratory consulted the cylinder supply company the difficulty of tightening the cylinders’ connection. A new connection was designed , which made use of an O-ring on top of the metal-to-metal sealing surface system. The O-ring allowed the gas cylinder to be properly sealed without need for the extreme muscular effort that was previously necessary.
(4) All connectors of the hydrogen lines were replaced with the new model.
(5) Instructions detailing how to inspect O-rings for damage and to report any necessary replacement, were placed in the gas supply room.