Bulk Storage and Shipping of Liquid Hydrogen is Hazardous

by Dr. R.G. Scurlock, Emeritus Professor of Cryogenic Engineering, University of Southampton, r.g.scurlock@soton.ac.uk


Hydrogen is very attractive as a green fuel because it burns in air to produce only water—and possibly some nitrogen oxides and hydrides—and therefore has a zero carbon footprint. As an energy store and fuel, liquid hydrogen has a high ratio of combustion energy per unit weight and is particularly useful as a space rocket fuel with minimum weight penalty.

However, its liquid density is very low and the ratio of combustion energy per unit volume of liquid is about half that of other fuels such as LNG, LPG and other hydrocarbons. The consequences of this fact will create a problem when doubling of the volume for bulk storage and transport is compared with the volume of other fuels.

The use of hydrogen gas as a replacement fuel in reciprocating and turbine engines with standard thermodynamic efficiencies is attractive, as is the development of hydrogen/air fuel cells as primary electric generators, with considerably higher thermodynamic efficiency. These applications do not necessarily require liquid hydrogen, and gas in high-pressure 100-200 bar containers and tanks may be used.

On a small scale, local plants may use electrolysis of water to make hydrogen gas an energy store and fuel. The problems start to arise upon expanding the scale of production, storage and transport when the economy of large scale at 10,000 to 100,000 m3 is required to enable the use of liquid hydrogen as a large-scale energy store and working fuel.

At this time, LH2 in bulk quantity presents extremely hazardous properties as a medium for energy storage in the public domain. Any effort to store and/or ship bulk liquid hydrogen is unsafe, and should be terminated immediately, before any serious explosive accidents occur. The collective hazards present such a high risk that a minor spill could easily escalate into a major catastrophe, with many casualties and loss of the ship.

Comparing LH2 with LNG

The existing IGC Code entitled “Draft Interim Recommendations for Carriage of Liquefied Hydrogen in Bulk” actually notes in its Annex 2 that the requirements for the carriage of liquefied hydrogen as a bulk cargo are not specified in the IGC Code. The annex then provides interim recommendations based on the results of a comparison study of similar cargoes (i.e. LNG) as well as a bibliographic survey.

The recommendations list nine special requirements for carriage of liquefied hydrogen in bulk. These are based on a comparison study of the physical properties of LH2 and LNG, and conclude that the properties and handling procedures are similar because they both are cryogenic liquids, non-toxic and generate flammable high-pressure gas.

There are, however, some fundamental differences between LH2 and LNG that need to be considered together with the important fact that no one has any working experience handling more than about 1,000 m3 of liquid hydrogen in a single carrier. NASA and its associated US contractors and research labs do have extensive experience in medium scale handling of LH2 over the past 70 years, but none use large scale quantities.

Conversely, the technology of shipping LNG in bulk has been successfully developed over the past 45 years, and LNG tankers today are carrying up to 266,000 m3 of liquid in a number of separate tanks up to 55,000 m3 each.

To consider shipping LH2 in similar quantities is a step change from existing technology because LH2 has some widely different properties compared to LNG, apart from the fact that they are both cold liquids. Care is therefore needed in making conclusions from a simple comparison. For example, one major hazard with handling large quantities of hydrogen is the extreme wide range of its flammability limit in air from four percent up to 75 percent. Tests need to be carried out to demonstrate the flammability of liquid pools and gas leaks of hydrogen under working and emergency conditions. Such tests have been carried out on LNG pools and NG leaks, long before large LNG tankers were commissioned.

In practice, LNG pools have proved surprisingly difficult to ignite. Care in tests with LH2 pools is strongly advised, since ignition will be very fast and may result in an exploding fire ball or BLEVE (boiling liquid expanding vapor explosion).



The Cryogenic Differences Between LH2 and LNG

The normal boiling point of hydrogen under a pressure of one bar at 20K is much colder—92K colder—than the normal boiling point of methane at 112K, the main component of LNG. The normal boiling point of liquefied atmospheric air is 79K. This means that, while LNG is too hot to condense any air component, any surface cooled by liquid hydrogen at 20K will be colder than the condensing temperature of any air component. As a consequence, an LH2 system has to be separated thermally from any air. Otherwise, the air will condense to a liquid and then freeze to a solid at a very low pressure, with hazardous consequences.

For example, a leak of air into an insulation space will not be observable from a pressure reading. However, all the frozen air will evaporate, on warming to say 150K, creating high pressure rising rapidly to 10 or 100 bar, depending on the size of the leak and its length of time. The inner liquid-containing wall may also collapse and spill its hydrogen liquid and vapor contents into the insulation space, further adding to the rising pressure and possible mechanical failure.

And if the cold surface is at 78K, a portion of the air, containing up to 50 percent oxygen, will condense, presenting another hazard as an oxidizing agent and fire risk.

With its extremely low viscosity, sloshing of LH2 cargo in heavy seas is a problem and may lead to membrane damage, as experienced with LNG, and is believed to arise from local vapor explosions of cold liquid sloshing into contact with the relatively warmer tank roof. A fine network of multiple anti-slosh baffles may be able to control this problem.

Problems of the Need for Vacuum Insulation with LH2

For economic and technical reasons, the total boiloff rate (BOR) from a cargo of LH2 should be in the range of 0.2 percent per day, compared with 0.1 percent for LNG, the round figure required for driving the ship’s propulsion engines. Only two percent of the cargo here is lost on a 10-day voyage. Adopting LNG technology for shipping LH2 poses a major problem because a BOR of 0.2 percent cannot be achieved using LNG insulation technology.

Ship tanks for LNG are not vacuum insulated and have several 36–48 m diameter spheres, or equivalent volume membrane tanks, and hold 25,000–55,000 m3 of liquid in each tank. These units use 1 m thick plastic foam (PF) insulation together with nitrogen purge gas, or natural boiloff gas (as in vapor-cooled suspended-deck storage tanks) as a purge gas at just above 1 bar. The heat influx through the insulation is on the order of 15 W/m2.

The equivalent with LH2 would use similar foam insulation but with non-condensable hydrogen or helium purge gas. However, the working thermal conductivity or k-value for a purged insulation is approximately equal to the static (no convection) k-value at 300K for the purge gas. Both H2 (MW = 2) and He (MW = 4) have static k-values much higher than nitrogen or natural gas. The lowest insulation k-value of a purged foam insulation for LH2 can therefore be expected to be five times greater with heat influxes of 75 W/m2, compared with the heat influx of 15 W/m2 through the insulation for LNG.

Large volumes of hydrogen purge gas within the insulation spaces between the inner wall and the liquid tight secondary membrane—and between the secondary membrane and outer shell—may not be safe or allowed by safety rules because of possible leakage.

Another problem is that while the quoted latent heat of evaporation in kilojoules per kilogram for LH2 is similar to that of LNG, the working latent heat in kilojoules per cubic meter of liquid is about seven times smaller for LH2 compared with that of LNG. In other words, for the same heat influx, the evaporation rate is seven times larger in volume of evaporated liquid because of the low density of LH2 compared with LNG.

The corresponding BOR, in terms of the ratio of boiloff volume of liquid divided by the tank volume, for the same 1 m of PF purged insulation would be in excess of five percent per day. This magnitude is not acceptable either safety-wise or economically for a delivery voyage of 10 days.

On the other hand, if double-walled vacuum insulation (VI) is used, as in LNG VI pressure tanks, then BORs in the range of 0.2 percent per day are probably attainable. Evacuated powder insulations are generally regarded as unsuitable for ship tanks due to movement and packing from the ship motions. However, NASA has been testing a new evacuated powder manufactured of miniature glass bubbles for its 200 m3 spherical tank of LH2 and yielding a BOR of 0.1 percent per day.

Unfortunately, VI insulation requires the outer wall of the double-walled tank to be strengthened to resist a collapsing pressure of 1 bar. The largest VI tanks constructed today for LNG have a volume of 1,000 m3 only, and each tank has an empty weight of 210 tons to contain 70 tons of LH2. For larger volumes, the double-walled VI tanks would be very heavy with internal supports and anti-sloshing baffles, and would probably be mechanically impractical.

Zero Boiloff with Pressurized LH2 Store at Constant Volume

If the storage tanks are strong enough to take the rising pressures, it is possible to envisage sealing the tanks against the normal isothermal boiloff at 1 bar and storing the liquid at constant volume. The tank pressure and temperature are allowed to rise over a period of days to some pressure limit below, say, 10 bar. The heat flow into the tank is therefore stored instead of being released by evaporation. Until the pressure limit is reached and the emergency vents are opened, the hazards of safely disposing of the boiloff gas into the environment is circumvented. However, the subsequent handling of pressurized flammable liquid under loading, shipping and unloading conditions is now the problem.

Thermal Insulation and Practical Boiloff Rates

The acceptable equilibrium BOR of around 0.2 percent is determined by the amount and type of insulation built into the tank structure prior to commissioning. In principle, this equilibrium boiloff may contribute to the fuel for the propulsion engines, as in standard practice with LNG.

However, additional energy is transmitted to the liquid during cooldown, loading, shipping in heavy seas and unloading, resulting in both significant increases in BOR and the major problem of disposal of the excess hydrogen gas not required as a fuel. At the appropriate terminal, the excess BOR can be absorbed via the return vapor lines to the shore tanks. If not, the excess boiloff must be vented safely into the atmosphere or flared off.

Ortho-para Conversion and Associated Boiloff

Hydrogen is essentially a mixture of ortho- and para-hydrogen, with an equilibrium concentration of 75 percent ortho to 25 percent para at ambient temperature. However, when liquefied at 20K, there is a slow but continuous transformation of the ortho-hydrogen to the lower energy para-hydrogen.

The heat of transformation is around 120 percent of the latent heat of evaporation. The boiloff generated to the complete transformation over a few days will evaporate a boiloff of around 70 percent of the LH2 cargo. The standard way out of this problem is to catalyze the ortho-para conversion to a final state of 95-99 percent para during cooling and liquefaction from ambient temperature. However, the testing of the liquid product to determine the percentage conversion by instrumental analysis is difficult; the excess BOR over a period of hours is the conventional and tedious way of measurement.

In addition, the co-existence of ortho- and para- rich liquid with different densities may lead to stratification and subsequent rollover with sudden and uncontrolled boiloff. If this happens, the tanks used for storage and shipping must be able to vent freely to avoid mechanical failure.

Hydrogen Embrittlement and Associated Leaks

Embrittlement arises from hydrogen diffusing into the atomic structure of the metalwork, liquid containment tanks and all the pipework, valves and fittings. Some of this diffusing hydrogen reacts chemically, continually forming hydrides within the metal structure and weakening its mechanical strength to the point of embrittlement and cracking.

The degree and rate of embrittlement depends on both the particular structural material of the ship, the tanks and the fittings, and also on the level of stressing. This takes place at ambient temperature and can lead to leaks of vapor and liquid for which frequent testing of empty and full tanks would be required to meet safety regulations.

There is also cryogenic embrittlement of structural steels below 150K. So LH2 must never make thermal contact with a ship’s steel structure and decks during liquid transfer and shipping operations. This latter problem is regulated by the inclusion of liquid-tight secondary membranes around the primary inner tanks in LNG tankers and drip trays/spray shields around the deck under pipework.

Diffusivity of Hydrogen and the Cold Plumes of Emergency Venting of Cold Gas

It may be argued that since hydrogen has a high diffusivity (a high rate of molecular diffusion) the vent gas will rapidly mix with air to a dilution below 4 percent the lower level of flammability in air. This may be correct at 300K but not at 20K. In an emergency boiloff the gas will be very cold with low diffusivity, and the vent gas will form a downwind white plume in which the cold hydrogen is cooling the air as it mixes and the density of the plume remains heavier than the surrounding air. There is very little convective mixing and the cold plume carries along at sea or ground level. The brilliant white coloration is due to water condensing and freezing to microscopic ice particles. The cold plume may extend for several kilometers until its increasing temperature and decreasing density provide buoyancy for the plume to rise and disappear. Inside the plume, the dilution may not be below the lower flammability level and a spark or ignition source within the plume will ignite it and possibly create a flashback to the vent. Trial tests are needed to check whether cold plumes are a hazard as predicted.

What to do with Excess Boiloff Vapor

The wide range of the flammability-limits presents a hazard that makes hydrogen require special care in handling excess boiloff gas. Options include atmospheric dilution, flare stacks and re-condensation The first way to deal with boiloff is to burn it in the ship’s propulsion engines as in LNG tankers. For 100,000 m3 cargo of LH2, this would consume around 0.2 percent per day.

To dump the excess into the atmosphere would require large exhaust fans with a volume ratio of 20-25 times the boiloff gas volume at 300K so that the dilution was below the lower flammability or explosion limit in the downwind exhaust stream. This condition would have to be met to prevent a flashback to the ship from any nearby ignition source and a subsequent fire onboard leading to a possible BLEVE explosion.

Alternatively, the excess boiloff could be burned off in remote flare stacks over open water, or over water-filled flare tanks as used at NASA’s Kennedy Space Center (KSC).

Another suggestion is to use some of the excess to drive the compressors of a set of re-condensers, mounted in the top of each tank, which reliquefy the cold boiloff gas before it exits the tank. Re-condensing the cold gas would help maintain the para-hydrogen composition and avoid reconversion and the additional refrigeration load of any orthohydrogen formed at higher temperatures.

Electrostatics, Electrical Equipment and Safe Handling Onboard and at Terminals

There are bound to be small leaks of LH2 and hydrogen vapor into the local shipping space during handling operations, with immediate production on mixing with atmospheric air of an inflammable mixture. Since the ignition energy is extremely small (~1/10 that of methane/air mixtures), the smallest electrostatic spark must be totally eliminated from the onboard and local ship environment. This means that all personnel must wear non-sparking clothes (i.e. no synthetic underwear or outer clothing that might accumulate static charge). Likewise, all footwear must be grounded individually.

All fixed and portable electrical equipment used in the cargo area of the ship will also have to be “certified safe” for explosive hydrogen/air atmospheres. Very little equipment is currently available to this standard, as hydrogen atmospheres are not common. It is not known if any modifications will be required to enable standard “safe” electrical equipment to operate in hydrogen atmospheres. The low ignition energy and wide flammable range will present
significant challenges.

Thunderstorms present a nightmarish hazard, with the whole ship needing to be protected by an overhead network of well-grounded lightning conductors, as NASA uses at KSC when space rockets are prepared for launch.

Small-scale 1000 m3 VI containers

Small-scale storage and handling of up to 1000 m3 of pressurized LNG is widely practiced, so this may be one way forward for conveying LH2. The containers are pressure vessels with vacuum insulation VI and liquid volumes 1 -10 m3 for truck and ferries and volumes of 100–1000 m3 for local distribution and bunkering. However, while the VI technology could be transferred to LH2 with some modifications, the vacuum technology is approaching a mechanical limit in volume per unit container ai 1,000 m3. So one less ambitious way forward is to use multiple containerized units with VI zero boiloff tanks as used currently for LHe. Such containerized tanks with volumes up to 80 m3 for standard container frames, together with the container carriers, would have to meet all the safety standards needed to handle LH2 as listed.


(a) In comparing the storage and handling behavior of LH2 with LNG, a clearer picture is presented by using energy terms related to volume (in m3) rather than weights (in kg). The liquid densities are very different, with LH2 density being six times less dense than LNG. While the quoted latent heats of evaporation are similar in units of kJ/kg, in working units of kJ/m3, the latent heat of LH2 is seven times smaller than for LNG; i.e. for the same heat influx, the amount of vapor generated, or BOR, is seven times larger for LH2 than for LNG.

Comparing heats of combustion in kJ/kg, LH2 is over three times higher than LNG. On the other hand, the heat of combustion in kJ/m3 of LH2 is only 50 percent that for LNG. In other words, for storing the same quantity of thermal energy, the volume of LH2 required is two times the volume of LNG and other hydrocarbon fuels.

(b) LNG storage technology with gas-purged insulation is probably unsuitable for the storage of LH2.

(c) Vacuum insulated VI tanks used for LNG could, in principle, be adapted for use with LH2, but only to 1,000 m3 capacity because the outer wall of the tank must withstand a collapsing pressure of 1 bar.

(d) VI tanks for LH2 could also be zero boiloff under pressures up to 8–10 bar, but again to a limit of 1,000 m3 capacity.

(e) Boiloff in excess of the 0.2 percent per day for fueling a tanker is a considerable problem with many hazards to be handled under safety regulations. Test programs are needed to evaluate safety precautions.

(f) The wide flammability range makes the disposal of cold hydrogen gas a major public hazard. Cold plumes downwind, and inadequate dilution to below four percent provides possibilities for flashback to the vent from distant ignition sources, leading to a fire onboard and a possible BLEVE explosion.

(g) While the energy of liquefaction of natural gas is around eight percent of the energy stored in the LNG product, the energy of liquefaction of hydrogen is 28 percent—or more—of the energy to be stored in the LH2 product.

(h) Other problems to consider include: hydrogen embrittlement in the metal structure of the tanks and ship, the incomplete ortho-para conversion in the LH2 cargo and the extremely small ignition energy, which may be activated by electrostatic sparks in personal clothing, sparks from electrical equipment and thunderstorms.

(i) A possible initial way forward is the use of multiple 1,000 m3 VI tanks to say 8 bar pressure, until adequate development of 25,000 m3 spherical tanks with 0.2 percent BOR can be achieved.