Hydrogen leak on a fuel cells bus

The event occurred in a bus garage, while emptying and purging hydrogen tanks on board of a fuel cells and hydrogen bus. This process was necessary before staring bus preventive maintenance and was executed via an electronic control system. However, the solenoid valve one of the tanks was mechanically stuck. In these cases, the valve manufacturer had foreseen an alternative manual process using an override tool. The process was executed according to manufacturer’s instruction, but the valve still did not open to evacuate hydrogen.
The fuel cell engineer tried to rectify the problem and started removing the override tool, but forgot to unwind the stem. The sealing O-ring of the tool was damaged and caused a hydrogen release in the vicinity of the manual override tool.

The building was evacuated, the extraction fans set manually and the main shutters open to assist with ventilation and dilution of the hydrogen leak.
During the incident the emergency services were alerted. The fire brigade could monitor hydrogen concentration. The highest hydrogen level recorded was 46% of the lower flammable limit on one sensor located directly above the leak.
therefore levels were judged acceptable from an explosive risk perspective.

Approximately 5 hours later the fire brigade left the site and the depot retuned to normal business, but still subject to the control actions until depletion of the remaining hydrogen.

The INITIATING CAUSE, as established by the investigation of the valve manufacturing, was a broken wire of the solenoid coil.

Contributing cause was the failing of the manual override tool process to release the hydrogen from the tank. What it followed was an error by the engineer in an attempt to dismantle the manual tool. This caused the damage of an O-ring.

on-board compressed hydrogen storage

The bus was in the garage (depot) for defueling.
The manual purging process had been previously undertaken unsuccessfully a few times by the fuel cell engineers. This was a rare, but not exceptional process.

There has not been any consequence.

The bus company requested by the valve manufacturer to investigate the solenoid valve and to issue a report.

The building was evacuated, the extraction fans set manually and the main shutters open to assist with ventilation and dilution of the hydrogen leak.

During the incident the emergency services were alerted. The local fire brigade put a 50 metre cordon round the fuel cell building until the hydrogen levels subsided. Hydrogen detectors were installed in the depot. The fire brigade could therefore monitor the levels in the workshop atmosphere and judged it as acceptable from an explosive risk perspective. The highest hydrogen level recorded was 46% of the lower flammable limit on one sensor located directly above the leak.

Approximately 5 hours later the fire brigade left the site and the depot retuned to normal business, but still subject to the control actions until depletion of the remaining hydrogen. Before leaving the scene, the fire brigade agreed the following with the buses company:
1. The fuel cell building would remain cordoned off.
2. The hydrogen detection panel would check every hour and the hydrogen levels recorded until they reached zero.
3. The fuelling compound would not be turned on until as late as possible.
The above three actions were undertaken and it was noted that in the early morning of the day after no hydrogen was detected in the hydrogen fuel cell workshop.

A further consideration was the possibility of air getting into cylinder once the pressure stabilised potentially causing a flammable environment, although there would be no ignition source. This was relayed to the local fire brigade, and after discussion it was decided that the risk was low. Nevertheless it was agreed that the cylinder would be purged with nitrogen.

According to the valve manufacturer,
(1) the solenoid valves of the type mounted on the buses were not fabricated anymore, but were still supported and did not represent any additional hazard than their new version.
(2) The manual tool was safe provided the instructions were thoroughly followed.
To ensure that this will happen in the future, a new detailed procedure was issued in case of failure of the solenoid valve. One new procedure step is the execution of the manual operation outside the depot, to ensure maximal hydrogen dispersion.

This incident is basically a near miss with non negligible hydrogen release) highlights the difficult to build a trustful supply chain for all components and sub-components i an period in which the components and systems have not yet achieved the mass production rates making easy quantification of failure statistics and identification of all possible failure modes.