Leak from a forklift during refuelling

A hydrogen leak occurred during refuelling of a fuel cell-powered lift truck. The forklift was depowered. This was the first refuelling after the replacement of the in-tank shutoff solenoid valve of the onboard hydrogen storage tank.

The event occurred during the final pressure testing of the repaired system when an O-ring failed at approximately 4500 psi (ca. 310 bar), releasing the entire contents of the tank in about 10 minutes.
The dispenser hose/nozzle was disconnected, and the leak location was quickly isolated to the tank/valve interface.
The leaking hydrogen did not ignite, no one was injured, and there was no property damage.

The reason for the leak was the failure of the O-ring to guarantee the required sealing above a certain pressure, due to the wrong (lower than needed) torque forces applied to the valve-tank connection during the valve replacement.

The INITIATING CAUSE was the inadequate sealing capacity of a valve O-ring, due to inadequate compression at the tank/valve interface.

The company responsible for the forklift services was subcontracting to another company, which reported that the O-ring was clean, lubricated, and tight when installed.
However, the investigation revealed that there was insufficient insertion depth of the in-tank solenoid valve, and therefore the O-ring was unable to provide a sufficient seal.
At the specified tank manufacturer’s torque requirement, the internal thread of the tank did not allow full engagement. The tank threads had been modified on other tanks of the same model, but not on the specific tank involved in this event. This tank would have required substantially more torque than the one provided foreseen by the procedure used by the servicing company.
The tank supplier has begun modifying its tanks to accommodate the valve, so newer batches were compliant, while the earlier systems require a higher torque to ensure a good seal.

The ROOT CAUSE could be identified as a combination of shortcoming in installation of components and their maintenance and repair. There is also a factor related to inadequate management of the relationship between contractors and qualification or verification of a repair or replacement.

O-ring, solenoid valve

The fuel cell systems for the lift trucks were originally built by Company A and placed into service in 2009. Company B assumed the service and support responsibilities in 2011, and then subcontracted the service to a third Company C. The only instructions given to Company C regarding the tank valve and the O-ring were to follow the tank valve manufacturer’s torque specification for ensuring a seal.

For a review of the state of the art of the hydrogen and fuel cell forklifts technology at the time of the incident, including their refuelling infrastructure, see NREL report quoted in the References (page 9 and page 15).

The only economic loss experienced what the loss of the onboard hydrogen tank content, which in most ofthe forklift is approximately 1 kg.

The dispenser hose/nozzle was disconnected, and the leak location was quickly isolated to the tank/valve interface. A 30-foot boundary around the lift truck was cleared of personnel and equipment. The combustible gas detector mounted on the wall above the hydrogen dispenser did detect an increase concenrtation of hydrogen, due to the action the facility ventilation system which was providing enough air flow to disperse the leaking hydrogen throughout the event.

The incident highlighted the importance of a trustful and stable supply chains for the components for a specific hydrogen technology, including their maintenance and repair. In this event, multiple companies were involved, and communication among them was inadequate. The original manufacturer had closed their fuel cells and hydrogen business and disbanded its team, when passing the know-how and contractual commitments to another company, which subcontracted the service to a third company. Technical knowledge and detailed know-how were not properly transferred. A similar leak had already been experienced at the same facility, but there was no corporate memory of the repair or the underlying failure mode.
In situation like these, the availability of a well-documented design history and operating records would guarantee optimal knowledge transfer of and facilitate prevention of mishaps such as the one occurred in this case.

Another lesson related to the need to qualify high-pressure components and to verify the continuity of the qualification following repair.

A third lesson learnt regarded the preventive and mitigating measures in place at the facility.
The combustible gas detector mounted on the wall above the hydrogen dispenser did not trigger an alarm because a large overhead facility fan was providing enough ventilation to keep hydrogen concentrations in the surrounding of the detector below alarm threshold.
Indeed, as calculated by Buttner et al. in the NREL report referred to (see References), assuming 1 kg of hydrogen released uniformly in a 25,000-ft3 facility would correspond only to a concentration of 1.7% by volume, less than half of the hydrogen LFL. However, a uniform distribution of hydrogen in confined spaces is not a safe assumption in risk assessment, due to possibility of accumulation under present internal structure. A detection system deployed in the most critical points provides an independent assurance of safety especially during high-pressure hydrogen transfers, which are known having a high incidents potential.