Nitrogen/helium blends are frequently used to leak test hydrogen systems.

Speed of detection, detection limit, location, and cross-sensitivity are some of many criteria that might be used for selecting a detector. A common setpoint for gaseous hydrogen detection is 25% of LFL, or 1% concentration in air. However, the detection limit also depends on the system and exposure. When specific hazards are likely, detectors may have detection limits in the low-ppm range. Lower detection limits can offer earlier detection, but also at the risk of spurious trips.

Leak detection system requirements depend on the system design and applicable codes. The
appropriateness of detection equipment depend on many factors, including the type of system,
application, location, and probability of leaks. For example, hydrogen refueling stations are required by
code to be equipped with leak detection systems.

While hydrogen gas detectors are less effective outdoors, they can be an important safeguard as part of an overall hydrogen system design. They have been used in many cases to automatically shut down equipment and isolate hydrogen supply. Location and type of detectors depend on the system design and siting, but when installed, should be in areas that are most likely to be exposed to hydrogen releases. Dispersion analysis can help with that assessment.

Training personnel and equipping them with portable gas detectors to properly identify the gas that is
leaking can play an important role in both troubleshooting and emergency response.

The manufacturer’s calibration requirements should be followed to ensure proper operation of the
detection system. The requirements will vary depending on the type of detector and the environment in
which they are installed. Calibration can usually be performed by the user/owner if properly trained and
supplied with calibration gas, etc.

The Panel is not aware of any standard for hydrogen detectors for onboard vehicle applications. Some information is available in the SAE Technical Information Report: TIR J3089 Characterization of On-Board Vehicular Hydrogen Sensors, which was published in 2018.

One pertinent reference is a Sandia National Laboratories paper by Schefer et al: Spatial and radiative properties of an open-flame hydrogen plume, Intl J. Hydrogen Energy, 31 (2006): 1332-1340. Information on this and other similar papers are available at https://h2tools.org/bibliography. Further information can probably be obtained from the author of this paper and other papers reporting hydrogen flame spectral distributions. Of course, hydrogen flame sensor manufacturers also have such data and should be consulted.  Specifications on these sensors should provide useful data.

There are many manufacturers of multiple  types of flame detectors and it’s best to seek their input for the advantages of different types for specific applications. Regarding flame detector technology, UV detectors have been prone to false alarms from outside sources such as sunlight, lightning, and welding/cutting torches. The newer triple-IR detectors that are specifically designed and tested for hydrogen flames have been shown to be effective. These detectors are currently in use at many hydrogen systems for detecting both hydrocarbon and hydrogen flames. Also, flame detection will only detect a leak once the hydrogen is ignited.  It is difficult to assess the minimum leak size without specific testing at known hydrogen pressure and the distance of the detector from the leak

The answer will depend on if it is only H2 measured or if VOCs are also included. The type of gas sensor may change since many detectors are limited to flammable gases. To provide feedback, the Panel would require more details on the sensors being used on the project and the failure modes. There are concerns about allowing hydrogen concentrations as large as 3.5% for the high level alarm. Since VOCs are mixed with the hydrogen, the mixture lower flammability limit is probably slightly less than 4%. A typical practice would use 25% of the LFL, or 1% hydrogen), for alarm, not 1.5% to 3.5%. The vacuum may be an issue, as normally these are operated at atmospheric pressure. Engineering discussions with vendors and research would be needed to assure VOCs are not an issue and the vacuum is suitable. Staff at the National Renewable Energy Laboratory have experience with a range of sensors and may be able to help, particularly if the failure mode with the current sensors is known.  As always, experienced sensor manufacturers should also be consulted for assistance.