Safety

Hydrogen has the potential to act as the energy carrier of the future. It will be then produced in large amounts and will certainly need to be transported for long distances. The safest way to transport hydrogen is through pipelines. Failure of pipelines carrying gaseous hydrogen can have several effects some of which can pose a significant threat of damage to people and buildings in the immediate proximity of the failure location. This paper presents a probabilistic risk assessment procedure for the estimation of damage to people and buildings endangered by high-pressure hydrogen pipeline explosions. The procedure provides evaluation of annual probability of damage to people and buildings under an extreme event as a combination of the conditional probability of damage triggered by an explosion and the probability of occurrence of the explosion as a consequence of the pipeline failure. Physical features such as the gas jet release process flammable cloud size blast generation and explosion effects on people and buildings are considered and evaluated through the SLAB integral model TNO model Probit equations and Pressure-Impulse diagrams. For people both direct and indirect effects of overpressure events are considered. For buildings a comparison of the damage to different types of buildings (i.e. reinforced concrete buildings and tuff stone masonry buildings) is made. The probabilistic procedure presented may be used for designing a new hydrogen pipeline network and will be an advantageous tool for safety management of hydrogen gas pipelines.

This paper presents results of an experimental investigation on detonation wave propagation in semi-confined geometries. Large scale experiments were performed in layers up to 0.6 m filled with uniform and non-uniform hydrogen–air mixtures in a rectangular channel (width 3 m; length 9 m) which is open from below. A semi confined driver section is used to accelerate hydrogen flames from weak ignition to detonation. The detonation propagation was observed in a 7 m long unobstructed part of the channel. Pressure measurements ionization probes soot-records and high speed imaging were used to observe the detonation propagation. Critical conditions for detonation propagation in different layer thicknesses are presented for uniform H2/air-mixtures as well as experiments with uniform H2/O2 mixtures in a down scaled transparent channel. Finally detail investigations on the detonation wave propagation in H2/air-mixtures with concentration gradients are shown.

A series of experiments in a thin layer geometry performed at the HYKA test site of the KIT. The experiments on different combustion regimes for lean and stoichiometric H2/air mixtures were performed in a rectangular chamber with dimensions of 20 x 90 x h cm3 where h is the thickness of the layer (h = 1 2 4 6 8 10 mm). Three different layer geometries:

• a smooth channel without obstructions;
• the channel with a metal grid filled 25% of length and
• a metal grid filled 100% of length. 

Detail measurement of H2/air combustion behaviour including flame acceleration (FA) and DDT in closed rectangular channel have been done. Five categories of flame propagation regimes were classified. Special attention was paid to the analysis of critical condition for different regimes of flame propagation as function of the layer thickness and roughness of the channel. It was found that thinner layer suppresses the detonation onset and even with a roughness the flame is available to accelerate to speed of sound. The detonation may occur only in a channel thicker than 6 mm.

Experiments were performed on the influence of oxidants (air pure oxygen O2 and pure nitrous oxide N2O at atmospheric pressure) in the straight expansion tube after the burst disk on the hydrogen spontaneous ignition. The lowest pressure at which the spontaneous ignition is observed has been researched for a 4 mm diameter tube with a length of 10 cm for the two oxidant gases. The ignition phenomenon is observed with a high speed camera and the external overpressures are measured. Numerical simulations have also been conducted with the high resolution CFD approach detailed chemistry formerly developed by Wen and co-workers. Comparison is made between the predictions and the experimental data.

The introduction of hydrogen to the commercial market as alternative fuel brings up safety concerns. Its storage in liquid or cryo-compressed state to achieve volumetric efficiency involves additional risks and their study is crucial. This work aims to investigate the behaviour of cryogenic hydrogen release and to study factors that affect the vapor dispersion. We focus on the effect of ambient humidity and air's components (nitrogen and oxygen) freezing in order to identify the conditions under which these factors have considerable influence. The study reveals that the level of influence depends highly on the release conditions and that humidity can reduce conspicuously the longitudinal distance of the Lower Flammability Limit (LFL). Low Froude (Fr) number (<1000) at the release allows the generated by the humidity phase change buoyancy to affect the dispersion while for higher Fr number - that usually are met in cryo-compressed releases - the momentum forces are the dominant forces and the buoyancy effect is trivial. Simulations with liquid methane release have been also performed and compared to the liquid hydrogen simulations in order to detect the differences in the behaviour of the two fuels as far as the humidity effect is concerned. It is shown that in methane spills the buoyancy effect in presence of humidity is smaller than in hydrogen spills and it can be considered almost negligible.

At the JRC-IET on-board hydrogen tanks have been subjected to filling–emptying cycles to investigate their long-term mechanical and thermal behaviour and their safety performance. The local temperature history inside the tanks has been measured and compared with the temperatures outside and at the tank metallic bosses which is the measurement location identified by some standards. The outcome of these activities is a set of experimental data which will be made publicly available as reference for safety studies and validation of computational fluid dynamics.

Hydrogen safety is a relevant topic for both nuclear fission and fusion power plants. Hydrogen generated in the course of a severe accident may endanger the integrity of safety barriers and may result in radioactive releases. In the case of the ITER fusion facility accident scenarios with water ingress consider the release of hydrogen into the suppression tank (ST) of the vacuum vessel pressure suppression system (VVPSS). Under the assumption of additional air ingress the formation of flammable gas mixtures may lead to explosions and safety component failure.<br/>The installation of passive auto-catalytic recombiners (PARs) inside the ST which are presently used as safety devices inside the containments of nuclear fission reactors is one option under consideration to mitigate such a scenario. PARs convert hydrogen into water vapor by means of passive mechanisms and have been qualified for operation under the conditions of a nuclear power plant accident since the 1990s.<br/>In order to support on-going hydrogen safety considerations simulations of accident scenarios using the CFD code ANSYS-CFX are foreseen. In this context the in-house code REKO-DIREKT is coupled to CFX to simulate PAR operation. However the operational boundary conditions for hydrogen recombination (e.g. temperature pressure gas mixture) of a fusion reactor scenario differ significantly from those of a fission reactor. In order to enhance the code towards realistic PAR operation a series of experiments has been performed in the REKO-4 facility with specific focus on ITER conditions. These specifically include operation under sub-atmospheric pressure (0.2–1.0 bar) gas compositions ranging from lean to rich H2/O2 mixtures and superposed flow conditions.<br/>The paper gives an overview of the experimental program presents results achieved and gives an outlook on the modelling approach towards accident scenario simulation.

This study is aimed at the validation of the pressure peaking phenomenon against laboratory-scale experiments. The phenomenon was discovered recently as a result of analytical and numerical studies performed at Ulster University. The phenomenon is characterized by the existence of a peak on the overpressure transient in an enclosure with vent(s) at some conditions. The peak overpressure can significantly exceed the steady-state pressure and jeopardise a civil structure integrity causing serious life safety and property protection problems. However the experimental validation of the phenomenon was absent until recently. The validation experiments were performed at Karlsruhe Institute of Technology within the framework of the HyIndoor project. Tests were carried out with release of three different gases (air helium and hydrogen) within a laboratory-scale enclosure of about 1 m3 volume with a vent of comparatively small size. The model of pressure peaking phenomenon reproduced closely the experimental pressure dynamics within the enclosure for all three used gases. The prediction of pressure peaking phenomenon consists of two steps which are explained in detail. Examples of calculation for typical hydrogen applications are presented.

Integral type models to describe stationary plumes and jets in cross-flows (wind) have been developed since about 1970. These models are widely used for risk analysis to describe the consequences of many different scenarios. Alternatively CFD codes are being applied but computational requirements still limit the number of scenarios that can be dealt with using CFD only. The integral models however are not suited to handle transient releases such as releases from pressurized equipment where the initially high release rate decreases rapidly with time. Further on gas ignition a second model is needed to describe the rapid combustion of the flammable part of the plume (flash fire) and a third model has to be applied for the remaining jet fire. The objective of this paper is to describe the first steps of the development of an integral-type model describing the transient development and decay of a jet of flammable gas after a release from a pressure container. The intention is to transfer the stationary models to a fully transient model capable to predict the maximum extension of short-duration high pressure jets. The model development is supported by conducting a set of transient ignited and unignited spontaneous releases at initial pressures between 25bar and 400bar. These data forms the basis for the presented model development approach.

A precise understanding of the flame turbulence induced by cellular instabilities is indispensable to perform an appropriate risk assessment of hydrogen fuelled gas explosion. In this research Darrieus Landau instability (DL instability) whose effect on gas explosion is  remarkable was experimentally examined. The DL instability is essentially caused by a volumetric expansion of burned gas at flame front. Therefore in order to examine the effects of volumetric expansion ratio the experiments were conducted using H2-O2-N2-Ar gas mixtures of various volumetric expansion ratio conditions by changing N2-Ar ratio. When Ar content ratio is increased the flame temperature becomes higher and volumetric expansion ratio is increased owing to lower specific heat of Ar. The experiments were conducted in nearly unconfined conditions of laboratory-scale and large-scale. Gas mixtures were filled in a 10 cm diameter soap bubble for the laboratory-scale and in a plastic tent of thin vinyl sheet of 1m3  for the large-scale. The gas mixtures were ignited by an electric spark and blast wave and flame speed were measured simultaneously by using a pressure sensor and a high-speed video camera. The DL instability owing to volumetric expansion accelerates flame propagation. In addition the intensity of blast wave was greatly raised depending on flame acceleration which can be explained by an acoustic theory. The effects of expansion ratio and experimental scales on flame propagation and blast wave were analyzed in detail. These results are quite important to perform an appropriate consequence analysis of accidental explosion of hydrogen.