Yes, numerous incidents have occurred where frozen air (which contains oxygen) has built up within a hydrogen process or vent system. These incidents with vent systems incorporate more than just a vent stack, but include a vent system consisting of additional atmospheric equipment (such as a tank) where the equipment stays cold and allows air into the system in contact with a cold hydrogen stream. 

Vent systems are at risk since they are “open” to the atmosphere and certain flow conditions might result in the aspiration of air into a cold hydrogen flow which then leads to the freezing of the air. 

A small quantity of solid air can create an explosive hazard which then leads to cascading failures from the initial incident. Solid air in hydrogen is also shock sensitive which can lead to unexpected ignition.

Flame arrestors can be installed on hydrogen gas vents. The purpose of a flame arrestor is to prevent the migration of flame backwards and upstream into the vent stack or system itself. Generally, flame arrestors are not needed since: 1) the vent stack should be designed to withstand fire or explosion within the stack, and 2) the process generally does not contain a flammable mixture within it, so there is no danger of the flame propagating into the system. In these cases, flame arrestors are not needed and should be avoided since they create additional hazards of blocking or restricting the vent flow. 

There are situations where flame arrestors can be appropriate. An example would be a vent stack from a compressor crankcase. Compressor crankcases are often vented to atmosphere, and where there is a possibility of hydrogen leakage into the crankcase, that vent might use a vent stack to ensure that leakage goes to a safe location. Since the crankcase might then have an air-hydrogen mix, a flame arrestor can prevent the propagation of flame backwards into the crankcase if it were to ignite on the vent stack. The crankcase is usually confined and congested so it could result in an overpressure significant enough for it to rupture. 

The European Industrial Gas Association (EIGA) Doc 211/17, 7.4 does not allow flame arrestors due to the backpressure associated with them.

We are not certain what an inverted vent top is. If this means the hydrogen flow is pointed downward in any way towards grade, then yes it must be avoided. Less dangerous vent gases can be pointed downward, especially those that mix with air rapidly (nitrogen/oxygen/argon). Regardless, reaction forces must be taken into account for any relief valve activation or flow

The distances provided are minimums. While they might be sufficient for most vents from small systems, larger vents will require both a dispersion and radiation analysis to determine the height needed.

Yes, small flowrate vents may be invisible, particularly in daylight. Sometimes it may still be possible to see heat striations in the air from the heat generated by the fire, but it can be difficult to discern at low flowrates. 

It is normal for some air ingress to occur from the vent stack outlet. This is not a hazard if the stack has been properly designed to withstand an internal explosion or fire. Once hydrogen flow from a device is initiated, it will sweep out any air that might be in the stack. Generally, if the vent rate is insufficient to sweep the air out, then it’s also insufficient to freeze or liquefy air in the stack. However, it’s important to prevent air from being pulled into the stack from a venturi effect, so leaks or holes where air can enter the vent piping should be eliminated. 

It is not possible to define ignition potential by just velocity without more data (i.e. pressure, materials involved, direction of impact). Due to the multiple methods of developing an ignition source (friction, impact, electrical charge) and the low ignition energy, it is assumed that hydrogen in the air will ignite (between 4 -74%), as it does 30-40% of the time with no known ignition source (see GH2 chart below). Therefore, to try and manage impingement by velocity as an ignition source is not a practical method to assure no ignition.

Other Information:
The ignition energy of hydrogen is .02 millijoules. By definition, a joule is equal to the kinetic
energy of a kilogram mass moving at the speed of one meter per second.

From “Mechanical Sparks as an Ignition Source of Gas and Dust Explosions” from The Italian Association of Chemical Engineering Online:

Mechanical sparks are small particles, which due to the impact between two objects are torn loose from the surface of one of the two colliding objects. The kinetic energy is turned into heat and deformation work. Mechanical spark generation is dependent on the pressure with which the one object is working against the other, the relative speed between the objects, the friction coefficient and the hardness of the materials involved.

Extrapolation of the experimental results using a model it could be shown that incendive hot surfaces can be generated also at relative speeds of < 1 m/s.

Additionally, tests were performed using a file traveling at 1m/s against a metal surface, and the ignition of hydrogen over many concentrations was observed.

While HSP members have limited experience with MFCs in experimental setups, the Panel does not consider them to be reliable to provide a positive flow shutoff. For safety, a shutoff valve in series is recommended. Projects will also need to consider hazardous electrical rating and location when flowing H2. Regarding Coriolis mass flow measuring devices, Coriolis flow meters measure mass rate changes by oscillating a flow tube and measuring tube distortion response. Measurement resolution is better with heavier, denser materials. Since hydrogen is the least dense gas, the response is much less than with denser materials. They are also expensive and very challenging to operate at low mass flow rates. 

The HSP is aware of a few used with electrolyzers for flow confirmation at the 500 slpm range but lacks feedback on how well these worked.  The project should consider working with the supplier regarding this application. There is a special MFC for vacuum application that has been observed to work well, the only difference is the valve on this is toward vacuum. On the other hand, MFCs made for pressures of atmospheric or above could be used in slight vacuum (up to 10 psia) and low flow rates (1-2 slpm) but are not very accurate due to the expansion of gas at the exhaust of the valve; it depends on the compressibility of that particular gas. 

The recalibration time recommended is 2 years. If there is a valve leak, check the inlet-outlet valve opening pressure ratings on the spec sheet. From a safety perspective, it is not recommended that an MFC be relied on to provide a leak-tight seal against hydrogen. Further, MFCs are not accurate if the gas has other constituents not originally included in the calibration. For instance, humidification of fuel cell supply gases would significantly change the accuracy.   Also, given the inherent risk in mixing H2 and O2 in a vacuum chamber, the HSP recommends a rigorous, multi-party, hazard analysis.