Fire in a hydrogen laboratory

A laboratory worker proceeded to loosen a fitting in one of the hydrogen gas lines. The line was still under pressure at approximately 110 psig (7.6 bar) and connected to the hydrogen supply. The worker had closed a valve on the wrong line before and assumed to work on an isolated section of the hydrogen pipe.

Hydrogen escaped from the loosened fitting and the ejected the section of the pipe. The leak ignited, remaining invisible but manifested itself as a white stream around the hydrogen jet, and a change of noise of the jet. All happened in a matter seconds and the worker could not react, but leaving the lab and pushing the emergency stop button. Someone else pulled also the fire alarm.

These actions were designed to close the main hydrogen solenoid (shutoff) valve. Realising that the hydrogen supply did not close automatically, the local emergency response team closed manually the main hydrogen valve at the hydrogen storage system.

No injury and negligible property damage.

The INITIATING CAUSE was the failing to isolate and depressurise the section of pipe before dismantling it.
THe IGNITION SOURCE was could have been either static discharge or a power bar.

Despite the several safety measures installed in the laboratory, the safety system failed to bring the laboratory in a safe state. The reason was the wrong installation of the shutoff valve and its malfunctioning.
The human error revealed a weakness of the safety system, which was relying on the activation of only one safeguard, the good functioning of which had never been properly tested and verified. The ROOT CAUSE is a combination of human error, inadequate procedure able to avoid confusion of valve and pipes, and a shortcoming of the safety design.

fitting,
pipe,
solenoid valve

The gas lab safety system had a main control panel. Hydrogen sensors, flame detectors, emergency push buttons, fire alarm and sprinkler system were all inputs into the control panel. In the event of an emergency, if any of these systems were engaged, the control panel should de-power the main solenoid valve and stop the gas flow.
At first site, theses preventive and mitigation measures should have been able to minimise the consequences of any mishaps in the laboratory. However, everything depended on the good functioning of one component, the solenoid valve.

In the case of this event, (i) the solenoid valve was not installed according to the manufacturer specification, and (ii) its activation was checked just on the basis of a click of its motor, without verifying that the valve was really able to completely stop the hydrogen flow.
For components with a critical safety role, redundancy should be a standard. Moreover, the correct functioning of the component under all considered accidental scenarios should be the guaranteed by adequate inspection and verification procedures.

1. The hydrogen main shut off valve was installed with the correct orientation.
2. The lab safety system testing procedures were modified. When testing the main valve, its proper functioning was assessed by measuring the downstream pressure after activation.
3. The lab working procedures were modified and training was organised, aiming at avoiding the human error which took place.