Recent Changes - Search:

HySafe

SBEP

PmWiki

pmwiki.org

edit SideBar

LH2 pool spreading and vaporization

Liquefied gases are characterized by a boiling point well below the ambient temperature. If released from a pressure vessel, the pressure relief from system to atmospheric pressure results in spontaneous (flash) vaporization of a certain fraction of the liquid. Depending on leak location and thermodynamic state of the cryogen (pressure expelling the cryogen through the leak is equal to the saturation vapour pressure), a two-phase flow will develop, significantly reducing the mass released. It is connected with the formation of aerosols, which vaporize in the air without touching the ground. Conditions and configuration of the source determine features of the evolving vapour cloud such as cloud composition, release height, initial plume distribution, time-dependent dimensions, or energy balance. The phenomena that may occur after a cryogen release into the environment are shown in Fig. 1.


Fig. 1: Physical phenomena occuring upon the release of a cryogenic liquid

LH2 Vaporization

The release as a liquefied gas usually results in the accumulation and formation of a liquid pool on the ground, which expands, depending on the volume spilled and the release rate, radially away from the releasing point, and which also immediately starts to vaporize. The equilibrium state of the pool is determined by the heat input from the outside like from the ground, the ambient atmosphere (wind, insolation from the sun), and in case of a burning pool, radiation heat from the flame. The respective shares of heat input from outside into the pool are depending on the cryogen considered. Most dominant heat source is heat transport from the ground. This is particularly true for LH2, where a neglection of all other heat sources would result in an estimated error of 10-20%. For a burning pool, also the radiation heat from the flame provides a significant contribution. This is particularly true for a burning LNG pool due to its much larger emissivity resulting from soot formation (DienhartB:1995).

Upon contact with the ground, the cryogen will in a short initial phase slide on a vapor cushion (film boiling) due to the large temperature difference between liquid and ground. The vaporization rate is comparatively low and if the ground is initially water, no ice will be formed. With increasing coverage of the surface, the difference in temperatures is decreasing until at the Leidenfrost point the vapour film collapses resulting in enhanced heat transfer via direct contact (nucleate boiling). On water, there is the chance of ice formation which, however, depending on the amount of mass released, will be hindered due to the violent boiling of the cryogen, particularly if the momentum with which the cryogen hits the water surface is large. Unlike lab-scale testing (confined), ice formation was not often observed in field trials (unconfined).

The vaporization behaviour is principally different for liquid and solid grounds. On liquid grounds, the vaporization rate remains approximately constant due to natural convection processes initiated in the liquid resulting in an (almost) constant, large temperature difference between surface and cryogen indicating stable film boiling. On solid grounds, the vaporization rate decreases due to cooling of the ground. The heat flux into the pool can be approximated as being proportional to t-1/2. The vaporization time is significantly reduced, if moisture is present in the ground due to a change of the ice/water properties and the liberation of the solidification enthalpy during ice formation representing an additional heat source in the ground.

LH2 Pool Spreading

Above a certain amount of cryogen released, a pool on the ground is formed, whose diameter and thickness is increasing with time until reaching an equilibrium state. After termination of the release phase, the pool is decaying from its boundaries and breaking up in floe-like islands, when the thickness becomes lower than a certain minimum which is determined by the surface tension of the cryogen (in the range of 1 to 2 mm). The development of a hydraulic gradient results in a decreasing thickness towards the outside.

The spreading of a cryogenic pool is influenced by the type of ground, solid or liquid, and by the release mode, instantaneous or continuous. In an instantaneous release, the release time is theoretically zero (or release rate is infinite), but practically short compared to the vaporization time. Spreading on a water surface penetrates the water to a certain degree, thus reducing the effective height responsible for the spreading and also requiring additional displacement energy at the leading edge of the pool below the water surface. The reduction factor is given by the density ratio of both liquids telling that only 7% of the LH2 will be below the water surface level compared to, e.g., more than 40% of LNG or even 81% of LN2.

During the initial release phase, the surface area of the pool is growing, which implies an enhanced vaporization rate. Eventually a state is reached which is characterized by the incoming mass to equal the vaporized mass. This equilibrium state, however, does not necessarily mean a constant surface. For a solid ground, the cooling results in a decrease of the heat input which, for a constant spill rate, will lead to a gradually increasing pool size. In contrast, for a water surface, pool area and vaporization rate are maximal and remain principally constant as was concluded from lab-scale testing despite ice formation. A cutoff of the mass input finally results in a breakup of the pool from the central release point creating an inner pool front. The ring-shaped pool then recedes from both sides, although still in a forward movement, until it has completely died away.

A special effect was identified for a continuous release particularly on a water surface. The equilibrium state is not being reached in a gradually increasing pool size. Just prior to reaching the equilibrium state, the pool is sometimes rather forming a detaching annular-shaped region, propagates outwards ahead of the main pool (BrandeisJ:1983). This phenomenon, for which there is hardly experimental evidence because of its short lifetime, can be explained by the fact that in the first seconds more of the high-momentum liquid is released than can vaporize from the actual pool surface; it becomes thicker like a shock wave at its leading edge while displacing the ground liquid. It results in a stretching of the pool behind the leading edge and thus a very small thickness, until the leading edge wavelet eventually separates. Realistically the ring pool will most likely soon break up in smaller single pools drifting away as has been often observed in release tests. Whether the ring pool indeed separates or only shortly enlarges the main pool radius, is depending on the cryogen properties of density and vaporization enthalpy and on the source rate.

Also so-called rapid phase transitions (RPT) could be observed for a water surface RPTs are physical ("thermal") vapor explosions resulting from a spontaneous and violent phase change of the fragmented liquid gas at such a high rate that shock waves may be formed. Although the energy release is small compared with a chemical explosion, it was observed for LNG that RPT with observed overpressures of up to 5 kPa were able to cause some damage to test facilities.

Experimental Work

Most experimental work with cryogenic liquefied gaseous fuels began in the 1970's concentrating mainly on LNG and LPG with the goal to investigate accidental spill scenarios during maritime transportation. A respective experimental program for liquid hydrogen was conducted on a much smaller scale, initially by those who considered and handled LH2 as a fuel for rockets and space ships. Main focus was on the combustion behavior of the LH2 and the atmospheric dispersion of the evolving vapor cloud after an LH2 spill. Only little work was concentrating on the cryogenic pool itself, whereby vaporization and spreading never were examined simultaneously.

The NASA LH2 trials in 1980 (ChirivellaJE:1986) were initiated, when trying to analyze the scenario of a bursting of the 3000 m3 of LH2 containing storage tank at the Kennedy Space Center at Cape Canaveral and study the propagation of a large-scale hydrogen gas cloud in the open atmosphere. The spill experiments consisted of a series of seven trials, in five of which a volume of 5.7 m3 of LH2 was released near-ground within a time span of 35-85 s. Pool spreading on a "compacted sand" ground was not a major objective, therefore scanty data from test 6 only are available. From the thermocouples deployed at 1, 2, and 3 m distance from the spill point, only the inner two were found to have come into contact with the cold liquid, thus indicating a maximum pool radius not exceeding 3 m.

In 1994, the first (and only up to now) spill tests with LH2, where pool spreading was investigated in further detail, were conducted in Germany. In four of these tests, the Research Center Juelich (FZJ) studied in more detail the pool behavior by measuring the LH2 pool radius in two directions as a function of time (DienhartB:1995). The release of LH2 was made both on a water surface and on a solid ground. Thermocouples were adjusted shortly above the surface of the ground serving as indicator for presence of the spreading cryogen.

The two spill tests on water using a 3.5 m diameter swimming pool were performed over a time period of 62 s each at an estimated rate of 5 l/s of LH2, a value which is already corrected by the flash-vaporized fraction of at least 30%. After contact of the LH2 with the water surface, a closed pool was formed, clearly visible and hardly covered by the white cloud of condensed water vapor. The "equilibrium" pool radius did not remain constant, but moved forward and backward within the range of 0.4 to 0.6 m away from the center. This pulsation-like behavior, which was also observed by the NASA experimenters in their tests, is probably caused by the irregular efflux due to the violent bubbling of the liquid and release-induced turbulences. Single small floes of ice escaped the pool front and moved outwards. After cutting off the source, a massive ice layer was identified where the pool was boiling. In the two tests on a solid ground given by a 2 x 2 m2 aluminum sheet, the LH2 release rate was (corrected) 6 l/s over 62 s each. The pool front was also observed to pulsate showing a maximum radius in the range between 0.3 and 0.5 m. Pieces of the cryogenic pool were observed to move even beyond the edge of the sheet. Not always all thermocouples within the pool range had permanent contact to the cold liquid indicating non-symmetrical spreading or ice floes which passed the indicator.

Computer Modelling

Parallel to all experimental work on cryogenic pool behavior, calculation models have been developed for simulation purposes. At the very beginning, purely empirical relationships were derived to correlate the spilled volume/mass with pool size and vaporization time. Such equations, however, were according to their nature strongly case-dependent. A more physical approach is given in mechanistic models, where the pool is assumed to be of cylindrical shape with initial conditions for height and diameter, and where the conservation equations for mass and energy are applied [e.g., (FayJA:1978) and (BriscoeF:1980). Gravitation is the driving force for the spreading of the pool transforming all potential energy into kinetic energy. Drawbacks of these models are given in that the calculation is terminated when the minimum thickness is reached, that only the leading edge of the pool is considered, and that a receding pool cannot be simulated.

State-of-the-art modeling applies the so-called shallow-layer equations, a set of non-linear differential equations based on the conservation laws of mass and momentum, which allows the description of the transient behavior of the cryogenic pool and its vaporization. Several phases are being distinguished depending on the acting forces dominating the spreading:

  • gravitational flow determined by the inertia of the cryogen and characterized by a hydraulic gradient at the front edge;
  • gravitational viscous flow after pool height and spreading velocity have decreased making sheer forces at the boundary dominant;
  • equilibrium between surface tension and viscous forces with gravitation being negligible.

During spreading, the pool passes all three phases, whereby its velocity is steadily decreasing. For cryogens, these models need to be modified with respect to the consideration of a continuously decreasing volume due to vaporization. Also film boiling has the effect of reducing sheer forces at the boundary layer.

Based on these principles, the UKAEA code GASP (Gas Accumulation over Spreading Pools) has been created by Webber (WebberDM:1991) as a further development of the Brandeis model (BrandeisJ:1983) and (BrandeisJ:1983a). It was tested mainly against LNG and also slowly evaporating pools, but not for liquid hydrogen. Brewer also tried to establish a shallow-layer model to simulate LH2 pool spreading, however, was unsuccessful due to severe numerical instabilities except for two predictive calculations for LH2 aircraft accident scenarios with reasonable results (BrewerGD:1981).

At FZJ, the state-of-the-art calculation model, LAUV, has been developed, which allows the description of the transient behavior of the cryogenic pool and its vaporization (DienhartB:1995). It addresses the relevant physical phenomena in both instantaneous and continuous (at a constant or transient rate) type releases onto either solid or liquid ground. A system of non-linear differential equations that allows for description of pool height and velocity as a function of time and location is given by the so-called "shallow-layer" equations based on the conservation of mass and momentum. Heat conduction from the ground is deemed the dominant heat source for vaporizing the cryogen, determined by solving the one-dimensional or optionally two-dimensional Fourier equation. Other heat fluxes are neglected. The friction force is chosen considering distinct contributions from laminar and from turbulent flux. Furthermore, the LAUV model includes the possibility to simulate moisture in a solid ground connected with a change of material properties when water turns to ice. For a water ground, LAUV contains, as an option, a finite-differences submodel to simulate ice layer formation and growth on the surface. Assumptions are a plane ice layer neglecting a convective flow in the water, the development of waves, and a pool acceleration due to buoyancy of the ice layer.

The code was validated against cryogen (LNG, LH2) spill tests from the literature and against own experiments. LN2 release experiments were conducted on the KIWI test facility at the Research Center Juelich, which was used for a systematic study of phenomena during cryogenic pool spreading on a water surface. The leading edge of the LN2 pool is usually well reproduced. There is, however, a higher uncertainty with respect to the trailing edge whose precise identification was usually disturbed by waves developed on the water surface and the breakup of the pool into single ice islands when reaching a certain minimum thickness.

The post-calculations of LH2 pool spreading during the BAM spill test series have also shown a good agreement between the computer simulations and the experimental data (see Fig. 2) (DienhartB:1995).


Fig. 2: Comparison of LH2 pool measurements with respective LAUV calculations for a continuous release over 62 s at 5 l/s on water (left) and at 6 l/s on an Al sheet (right)

During the tests on water, the pool front appears at the beginning to have shortly propagated beyond the steady state presumably indicating the phenomenon of a (nearly) detaching pool ring typical for continuous releases. The radius was then calculated to slowly increase due to the gradual temperature decrease of the ice layer formed on the water surface. Equilibrium is reached approximately after 10 s into the test, until at time 62.9, i.e., about a second after termination of the spillage, the pool has completely vaporized. Despite the given uncertainties, the calculated curve for the maximum pool radius is still well within the measurement range. The ice layer thickness could not be measured during or after the test; according to the calculation, it has grown to 7 mm at the center with the longest contact to the cryogen. The spill tests on the aluminum ground (right-hand side) conducted with a somewhat higher release rate is also characterized by a steadily increasing pool radius. The fact that the attained pool size here is smaller than on the water surface is due to the rapid cooling of the ground leading soon to the nucleate boiling regime and enhanced vaporization, whereas in the case of water, a longer film boiling phase on the ice layer does not allow for a high heat flux into the pool. This effect was well reproduced by the LAUV calculation.

References

Invalid BibTex Entry!


Hydrogen Release

Edit - History - Print - Recent Changes - Search
Page last modified on December 04, 2008, at 06:20 PM