TIA 1783 points out a valid concern about how to address the electrical classification zone around a liquid hydrogen system. The existing requirements specify 3' around the outlet of the stack for Division 1 and 25' around the outlet of the stack for Division 2 area. These distances are historical and date back to the 1960's. They are a "one size fits all" simple approach that is easy to implement. Most existing systems use these distances for hazardous area plot plans and equipment selection. However, the distances are conservative for some systems, but also may not be sufficient for larger systems with higher flowrates.

These classified areas apply both to the system itself (for leaks) as well as the vent system outlets. This becomes more complicated since leaks are not expected, but may occur. In contrast, vent stacks are frequently used and hydrogen is expected to be vented since that is the stack’s function. Therefore, vent systems should be designed to properly vent the hydrogen to minimize impacts to personnel, facilities, and the environment.

The currently adopted editions of NFPA 2 are minimum requirements, but best practice for vent stacks would be to follow principles in NFPA 497 to ensure that the specific system and stack classified areas are developed with the actual parameters based on flow, direction, height, physical design, etc. The expected vent flows should be modelled and the classified areas developed accordingly, while using current NFPA requirements as a minimum for both the system and vents. NFPA 2 will eventually be updated with a new table in the future that is similar to, but not the same as, the proposal in TIA 1783.
 

BACKGROUND:
A Division 1 classified area expects hydrogen to be present for some portion of the time during normal operation. A Division 2 area only expects hydrogen to be present under upset conditions. Since systems are designed not to leak, and since leaks are not normal, the area around piping/tanks/etc. is normally considered Division 2. However, in certain areas such as fill connections and vent systems, hydrogen gas releases are expected. Hence, these areas are considered to be Division 1. The 3' extent of Division 1 area is somewhat arbitrary. A fill connection might have a small release of hydrogen during disconnection. Alternately, a vent stack could release much more substantial volume of hydrogen: a relatively small volume from normal operations to a very large release from a pressure relief device. TIA 1783 was correctly noting that since relief devices are of known sizes, then the classified areas should be based on the actual modelling of the relief rates and not just depend upon the traditional 3' or 25' distances.

For example, if it is known that a production system will vent a quantity that will result in a cloud that is 50' radius every time that it shuts down, then the classified area(s) should be much larger than the prescribed area. Similarly, if there is a very large relief device that isn't expected to operate, but might reasonably operate during the life of the system, then a similar analysis should be done.

TIA 1783 tried to express this in a new table but requires additional dialogue and analysis. Documents such as NFPA 497 (as referenced in the TIA), and API Standard 520 can be used as a best practice to develop the appropriate classified areas based on release rate models.

Classified areas are often shown in 2-D on drawings, but they are more accurately portrayed in 3-D (e.g. a "sphere"). In these situations, the height of a vent stack is a key dimension to ensure that the hydrogen cloud, radiation, and overpressure don't significantly create harm, especially considering hydrogen’s buoyancy. A properly designed vent stack should ensure that the momentum from the release further facilitates the upward direction from buoyancy, thereby reducing the extent of the classified area in the downward direction.

There are several concerns with “snuffing” a hydrogen fire from a vent stack. Most importantly, snuffing a hydrogen fire before the hydrogen is isolated can lead to the buildup of a hydrogen vapor cloud, which may then re-ignite, especially with hot surfaces available from the previous fire. The largest hazard is an explosion of the vapor cloud caused by delayed ignition.  It’s always better to isolate the hydrogen at its source to extinguish the fire as fuel runs out. 

Snuffing systems have been used in the past for vent system outlets mainly due to the negative   perception of a visible hydrogen flame at the top of the vent stack, particularly at night.  The success of these systems was marginal since high and sustained rates of inert gas were required to snuff the flame and sufficiently cool the piping outlet to prevent the venting flow from reigniting.  Generally, it’s preferred to design the vent system such that it can withstand a worst-case continuous fire on the outlet without affecting its integrity or surrounding exposures.  If those criteria are met, then it’s inherently safer to allow the vent to burn than to try to snuff it. 

Welded systems are generally preferred, where possible, to reduce the likelihood of leaks.  Generally, even welded systems will need non-welded joints (e.g. unions, flanges, etc.) to allow maintenance replacement of components.  A low-pressure system would not be an exception to this preference.  However, piping at lower pressures and smaller sizes will leak less hydrogen and have less probability of a leak. 

It is common to have mechanical joints for small piping (1/4” to 1/2”) for gauges, transmitters, and other measuring devices.  Additionally, leak rates are lower and are usually low risk, especially in large open areas for National Pipe Taper (NPT) threads below 1.5”. NPT joints are not recommended above 1.5” since they are susceptible to larger and more frequent leaks. Compression-style joints at low pressure have been shown to be largely leak free when installed correctly but also offer a risk of complete separation when assembled improperly. It is important to match the type of fitting to the application (purity, pressure, pipe size), allowable leak rate, and maintenance needs.

It is best to avoid planned blowdown of large amounts of hydrogen inventory at high flowrates if possible.  Low flow releases from vent systems are normal and occur for purging, delivery operations, and maintenance activity.   A challenge with high flow blowdown of a hydrogen system is that venting large quantities of hydrogen can itself be a hazardous activity.   Large blowdowns at high rates from vent systems can lead to jet fires and explosions after release to the atmosphere.

Flaring can be an option.  However, if flare stacks are used, they must ignite before the hydrogen reaches the end of the vent stack, so that a delayed ignition of the hydrogen does not occur, as this could create damaging overpressure.   A flare system is a complicated design for hydrogen. It is not normally a best practice unless the timing of the release is always known, and the flare cannot be extinguished until the hydrogen flow is stopped. Flares are generally only used at large production facilities which have the necessary infrastructure. 

A best practice for any storage system is to site the storage vessels away from any flammable substances and/or protect the vessels with barriers or insulation. It’s inherently safer to avoid   fire exposure onto the vessels, especially since relief devices may not be well suited to protect a vessel in the case of an impinging fire.  Similarly, there may be other methods to limit the H2 released by reducing the size, type or quantity of safety devices on a storage system. 

A best practice, when the storage vessels are not subject to an engulfing fire, is to use reclosing safety devices, such as spring loaded or pilot operated safety valves.  These do not empty the entire contents of the tubes, but open just to maintain the pressure within design criteria. 

Where it may be impossible to completely eliminate engulfing fires, rupture discs or thermally activated pressure relief devices (TPRD) are often preferred since once they activate, they will continue to vent until all pressure is released.  This is important since the fire may weaken the vessel while still at the reclosing devices’ setpoint, causing a vessel failure and a large sudden release of its content. However, non-reclosing relief devices can also be prone to inadvertent or spurious activation.  This can result in unnecessary and unwanted releases which can cause hazardous situations from high reaction forces and large quantity of the release. 

The design of vent systems is critical to the safety of the system. From a process perspective, the pipe design must be sufficient to withstand back pressure, internal pipeline pressure, deflagration pressure, thrust forces from the flow, and must be of a sufficient size to not create a restriction that prevents proper flow or activation of the devices. 

The vent system design starts with relief devices. Devices must first be sized based on the expected demand cases.  Documents such as CGA S1.3 or API 520 can provide guidance to size devices for fire situations.  It’s up to the designer to develop potential relief cases for other process demands, such as runaway heaters, compressors, backflow, etc. Once the relief devices are properly sized, then the discharge piping can be designed such that flow is not restricted.  For example, ASME relief valves will usually require the pressure drop on the downstream side of the device to be no higher than 10% of the set pressure.  Rupture discs don’t have the same requirement, but it is still necessary that the relieving capacity is not reduced below the demand case.  It’s also important to size the relief discharge piping for multiple devices where these devices (from different vessels, piping, etc.) could potentially vent simultaneously.  This is a particular concern where there may be many, sometimes even dozens, of devices on pressure vessels used for fire protection where all vessels can be exposed to fire at once.  As a rule, discharge piping will nearly always be larger than inlet piping, and the piping must be larger than the device orifice.  If there is a piping venting case (even if manual) into the same stack where both the vent case and the safety relief device can open simultaneously, the vent stack must be designed for all these cases.

Documents such as CGA G5.5 and API 521 provide guidance, requirements, and best practices for the design of vent systems, including addressing the thrust forces caused by high flow and velocity. 

There are advantages to reducing the amount and rate of hydrogen released.  Relief devices should not be oversized with an unnecessary safety factor.  Reduction of flow will lower reaction forces and the mass of hydrogen in the released cloud during an event.  A method to reduce the amount of hydrogen released is to use relief devices that reclose rather than emptying the entire inventory (i.e., rupture discs). However, rupture discs may still be needed if the storage vessels are subject to an engulfing fire to ensure the pressure is reduced before the mechanical integrity of the vessels is lost.

Pressure rating and set pressures of the devices are also a concern. For example, a 3000-psig set pressure device with the typical 10% allowable back pressure, would allow up to 300-psig in the vent header. If a lower set pressure device were connected to the same header, then the backpressure would not allow it to operate properly, leading to possible overpressure of the process system. Best practice would be to use different headers on systems that operate at significant differences in pressure.  A best practice is using a maximum 5:1 ratio to determine when a separate higher-pressure vent system is needed. For example, discharge from a device set at 100 psig should not be piped into the same vent system as a device set above 500 psig, even if the piping is nominally sized to allow minimal backpressure.

Another consideration during the hazard review is to ensure that vent headers do not create a common mode failure such that redundant devices could be blocked from a single failure. Care must also be taken so incompatible materials (e.g. hydrogen and oxygen) aren’t vented on a common manifold and that contamination (e.g. compressor oil) doesn’t affect other portions of the system where a source of contamination is present. 

When designing a vent system, the designer must perform a process safety analysis to ensure the hydrogen cannot flow to unexpected locations. It is never a good design to tie a hydrogen vent system into a building ventilation system. Maintenance is also an issue since vent headers can be an overlooked cross tie between portions of systems that otherwise are properly isolated on the upstream side. For example, if maintenance is being performed on a relief device, and a separate device activates elsewhere on the same header, then backflow could create a hazard while the vent piping is disassembled. One simple best practice is to vent hydrogen vertically where possible, using momentum to supplement its natural buoyancy. Wind can affect hydrogen flow, so it does not always rise straight up after release. Top works should also be designed to minimize plugging due to weather and other causes.

Numerous incidents have occurred where vent systems failed or leaked during device activation. A best practice is to pressure test vent systems above the expected backpressures to ensure integrity per ANSI B31.3.  Additionally, it is critical to support the vent stack to ensure it maintains its mechanical integrity during venting and weather events.

It is difficult to control fluid velocity in a vent header since nearly all relief systems will operate at a pressure that will result in choked/sonic flow at some location in the discharge piping or outlet.  There is generally no velocity limit within the piping, but from a practical perspective, high velocities will result in high, excessive pressure drop.  The velocity is only limited by the requirement to maintain maximum pressure drop of no more than 10% of the relief device set point in the vent system.

Both types of valve actuators are used, and both offer advantages and disadvantages for a given application.