How Fast Can You See a Hydrogen Leak?

There is no single answer, because there are two different events. A slow leak from a fitting or a permeation path disperses over minutes and is a genuine detection target. A thermally activated pressure relief device empties a 700-bar tank in under 75 seconds — and by design only opens once a fire is already heating the cylinder. One is a gas-detection problem. The other is not.

Two hydrogen events, not one

Conflating them is the most common error in specifying hydrogen sensing for a vehicle space. The slow case is an unintended release: a loosened fitting after road vibration and lashing, a degraded seal, permeation through a joint. Flow is low, momentum is low, and buoyancy plus the deck's own airflow decide where it goes. The fast case is a designed release: the TPRD is there to prevent a pressure-vessel rupture in a fire, so it vents the tank deliberately and completely.

The distinction matters because it sets what detection is being asked to do. Against a slow leak, a gas channel is looking for an accumulation before it reaches a flammable concentration — hydrogen's lower flammable limit is 4% by volume in air, and its minimum ignition energy is roughly 0.017 mJ, low enough that ordinary static discharge is a competent ignition source. Against a TPRD discharge, the fire has already happened; what a gas reading adds at that point is confirmation, not warning.

The slow leak: minutes, and geometry decides

Leak position changes the dispersion field more than leak rate does. Work on hydrogen leakage during the shipping of fuel cell vehicles, published in the International Journal of Hydrogen Energy, simulated releases at different locations and ventilation conditions and found the direction of diffusion strongly dependent on where the release originated: a leak at compartment-ceiling level spread horizontally far faster than one at floor level, reaching a maximum horizontal diffusion distance of 5.04 m within 540 seconds.

The shape of that spread is not the tidy expanding hemisphere the textbook buoyant-plume picture suggests. A companion experimental study, which put a fuel cell vehicle inside a 40-foot container and instrumented it with eight hydrogen sensors, found the released gas ran mainly along the nearest wall to the top of the container and only then diffused outward. Hydrogen in a confined, obstructed space is boundary-following: it uses the vehicle bodies, the bulkheads and the deckhead structure as guides. Two nominally identical leaks a few metres apart can therefore produce very different concentration histories at the same measurement point.

The TPRD event: seconds, and already a fire

This one is outside what any gas channel can usefully warn about. Onboard hydrogen storage is fitted with a thermally activated pressure relief device so that heat from a fire triggers a controlled blowdown rather than a tank burst. Typical TPRD orifices are in the 2–5 mm range depending on manufacturer, and published work on 700-bar systems puts total blowdown at under 75 seconds. Tunnel and car-park dispersion studies of a TPRD release under a vehicle describe a high-speed jet that impinges on the ground, spreads horizontally, recirculates under the chassis and fills the space beneath the car in about one second.

Read that sequence in the right order. The device is thermally activated, so the trigger for the release is a fire already in contact with the cylinder. A hydrogen alarm during a TPRD event is not early warning — the thermal channel should have been in alarm well before the relief device opened. Specifying hydrogen sensing on the promise of catching a TPRD discharge is specifying it for the one hydrogen event it cannot get ahead of.

Why the airflow that protects also hides

Ventilation is simultaneously the primary control measure and the primary reason the signal is weak. SOLAS Chapter II-2 Regulation 20 requires permanent mechanical ventilation on closed vehicle and ro-ro spaces, at a higher rate during vehicle operations when engines are running and a lower continuous rate on passage — commonly cited as 10 and 6 air changes per hour respectively. The CFD literature on hydrogen in deck compartments is consistent that ventilation is the most effective control available: improper ventilation lets hydrogen accumulate in a very short period, and optimised airflow prevents it.

The same airflow dilutes what a detector is trying to measure, exactly as it does for smoke and cell off-gas. The practical consequence is that a hydrogen reading is uninterpretable without knowing the ventilation state that produced it: a given concentration at 6 air changes per hour and the same concentration at 10 mean very different source strengths. Ventilation status belongs in the detection logic as an input, not as a background assumption — the same conclusion the smoke and off-gas channels reach from the other direction.

What a hydrogen channel can honestly promise

A bounded but real capability: minutes of warning on an unintended release, and nothing on a relief-device discharge. That is worth having — hydrogen's 4% lower flammable limit and 0.017 mJ ignition energy mean an undetected accumulation in an enclosed deck is a serious event, and the dispersion work says a slow release stays findable for several minutes before it has travelled far. It is also worth stating the limit plainly rather than selling the channel on the dramatic case.

Sensing-mode choice follows from the same split. Catalytic bead elements respond fast but need oxygen and are susceptible to poisoning over a long voyage; electrochemical cells drift and carry cross-sensitivities that a mixed vehicle deck will exercise; thermal conductivity sensing is unaffected by poisoning and remains linear at high concentration but is weaker near the low end where an early warning has to live. None of them is a single answer, which is the same conclusion the lithium-ion off-gas work reaches — the useful design is multi-modal, and the hydrogen channel earns its place on the slow-leak case, not on the TPRD case.

Sources

• 1. "Effects of leakage location and ventilation condition on hydrogen leakage during shipping of fuel cell vehicles", International Journal of Hydrogen Energy (ScienceDirect S0360319923063954): diffusion direction strongly dependent on leak location; ceiling-level leaks spread horizontally much faster than floor-level; maximum horizontal diffusion distance 5.04 m within 540 s.
• 2. "Experimental and simulation study on hydrogen diffusion and venting inside a container carrying fuel cell vehicles", International Journal of Hydrogen Energy (ScienceDirect S0360319924044082): fuel cell vehicle leak inside a 40-foot container instrumented with eight hydrogen sensors; released hydrogen spread mainly along the nearest wall to the top before diffusing.
• 3. "Research on Hydrogen Leakage Risk Control Methods in Deck Compartments of Hydrogen Fuel Cell-Powered Ships Based on CFD Simulation and Ventilation Optimization", Fire 8(10):400 (MDPI, 2025), doi:10.3390/fire8100400 — ventilation as the dominant control on hydrogen accumulation in deck compartments.
• 4. TPRD behaviour — Ulster University hydrogen safety engineering work and "Numerical Investigation of Hydrogen Jet Dispersion Below and Around a Car in a Tunnel", Energies 16(18):6483, doi:10.3390/en16186483: total blowdown of a 700-bar tank in under 75 s; TPRD orifice diameters typically 2–5 mm; under-chassis fill in approximately one second from an impinging jet.
• 5. SOLAS Chapter II-2 Regulation 20 — permanent mechanical ventilation required in closed vehicle, special category and ro-ro spaces, independent of accommodation and machinery ventilation; commonly cited as 10 air changes per hour during vehicle operations and 6 continuous on passage.
• 6. Hydrogen properties (lower flammable limit 4% vol in air, minimum ignition energy ~0.017 mJ, density roughly one-fourteenth that of air) — standard reference values, PNNL Hydrogen Tools (h2tools.org). FCEVs are ordinarily presented for shipping with only a small residual tank pressure.