Safety

The Hydrogen Incident Reporting and Lessons Learned website (www.h2incidents.org) was launched in 2006 as a database-driven resource for sharing lessons learned from hydrogen-related safety events to raise safety awareness and encourage knowledge-sharing. The development of this database its first uses and subsequent enhancements have been described at the Second and Third International Conferences on Hydrogen Safety [1] [2]. Since 2009 continuing work has not only highlighted the value of safety lessons learned but enhanced how the database provides access to another safety knowledge tool Hydrogen Safety Best Practices (http://h2bestpractices.org). Collaborations with the International Energy Agency (IEA) Hydrogen Implementing Agreement (HIA) Task 19 – Hydrogen Safety and others have enabled the database to capture safety event learning’s from around the world. This paper updates recent progress highlights the new “Lessons Learned Corner” as one means for knowledge-sharing and examines the broader potential for collecting analyzing and using safety event information.

Hydrogen is a valuable option of clean fuel to keep the global temperature rise below 2°C. However one of the main barriers in its transport and use is to ensure safety levels that are comparable with traditional fuels. In particular liquid hydrogen accidents may not be fully understood (yet) and excluded by relevant risk assessment. For instance as hydrogen is cryogenically liquefied to increase its energy density during transport Boiling Liquid Expanding Vapor Explosions (BLEVE) is a potential and critical event that is important addressing in the hazard identification phase. Two past BLEVE accidents involving liquid hydrogen support such thesis. For this reason results from consequence analysis of hydrogen BLEVE will not only improve the understanding of the related physical phenomenon but also influence future risk assessment studies. This study aims to show the extent of consequence analysis influence on overall quantitative risk assessment of hydrogen technologies and propose a systematic approach for integration of overall results. The Dynamic Procedure for Atypical Scenario Identification (DyPASI) is used for this purpose. The work specifically focuses on consequence models that are originally developed for other substances and adapted for liquid hydrogen. Particular attention is given to the parameters affecting the magnitude of the accident as currently investigated by a number of research projects on hydrogen safety worldwide. A representative example of consequence analysis for liquid hydrogen release is employed in this study. Critical conditions detected by the numerical simulation models are accurately identified and considered for subsequent update of the overall system risk assessment.

The design of ventilation system has implications for the safety of life and property and the development of regulations and standards in the space with the hydrogen storage equipment. The impact of both the position and the area of a single vent on the dispersion of hydrogen in a cuboid space (with dimensions L x W x H ¼ 2.90 0.74 1.22 m) is investigated with Computational Fluid Dynamics (CFD) in this study. Nine positions of the vent were compared for the leakage taking place at the floor to understand the gas dispersion. It was shown a cloud of 1% mole fraction has been formed near the ceiling of the space in less than 40 s for different positions of the vent which can activate hydrogen sensors. The models show that the hydrogen is removed more effectively when the vent is closer to the leakage position in the horizontal direction. The study demonstrates that the vent height of 1.00 m is safer for the particular scenario considered. The area of the vent has little effect on the hydrogen concentration for all vent positions when the area of the vent is less than 0.045 m2 and the height of the vent is less than 0.61 m.

Car park fires are known to be dangerous due to the risk of fast fire spread from one car to another. In general no fatalities are recorded in such fires but they may have a great cost in relation to damaged cars and structural repair. A very recent example is the Liverpool multi-storey car park fire from December 31 2017. It destroyed 1400 cars and parts of the building structure collapsed. This questions the validity of current design praxis of car parks. Literature studies assumes a 12 minutes period for the fire spread from one gasoline fuelled car to another. Statistical research and test from the European commission of steel structures states that in an open car park at most 3-4 vehicles are expected to be on fire at the same time.<br/>A number of investigations have been made concerning vehicles performance in car park fires but only a few are concerned with hydrogen-fuelled vehicles (HFV). It is therefore important to investigate how these new vehicles may contribute to potential fire spread scenario. The aim of the paper is to report the outcome of car park fire spread simulations involving common fuelled and hydrogen fuelled cars. The case study is based on a typical car park found in Denmark. The simulation applied numerical models implemented in the Fire Dynamic Simulator (FDS). In particular the focus of the study is on the influence of the parking distance to fire spread to adjacent vehicles in case a TPRD is activated during a car fire. The results help understanding whether different design rules should be envisaged for such structures or how a sufficient safety level can be obtained by ensuring specific parking condition for the hydrogen-fuelled cars.

The reproducibility of fire test protocol in the UN Global Technical Regulation on Hydrogen and Fuel Cell Vehicles (GTR#13) is not satisfactory. Results differ from laboratory to laboratory and even at the same laboratory when fires of different heat release (HRR) rate are applied. This is of special importance for fire test of tank without thermally activated pressure relief devise (TPRD) the test requested by firemen. Previously the authors demonstrated a strong dependence of tank fire resistance rating (FRR) i.e. time from fire test initiation to moment of tank rupture on the HRR in a fire. The HRR for complete combustion at the open is a product of heat of combustion and flow rate of a fuel i.e. easy to control in test parameter. It correlates with heat flux to the tank from a fire – the higher HRR the higher heat flux. The control of only temperature underneath a tank in fire test as per the current fire test protocol of UN GTR#13 without controlling HRR of fire source is a reason of poor fire test reproducibility. Indeed a candle flame can easily provide a required by the protocol temperature in points of control but such test arrangements could never lead to tank rupture due to fast heat dissipation from such tiny fire source i.e. insufficient and very localised heat flux to the tank. Fire science requires knowledge of heat flux along with the temperature to characterise fire dynamics. In our study published in 2018 the HRR is suggested as an easy to control parameter to ensure the fire test reproducibility. This study demonstrates that the use of specific heat release rate HRR/A i.e. HRR in a fire source divided by the area of the burner projection A enables testing laboratories to change freely a burner size depending on a tank size without affecting fire test reproducibility. The invariance of FRR at its minimum level with increase of HRR/A above 1 MW/m2 has been discovered first numerically and then confirmed by experiments with different burners and fuels. The validation of computational fluid dynamics (CFD) model against the fire test data is presented. The numerical experiments with localised fires under a vehicle with different HRR/A are performed to understand the necessity of the localised fire test protocol. The understanding of fire test underlying physics will underpin the development of protocol providing test reproducibility.

Radiative heat fluxes from small to medium-scale hydrogen jet flames (<10 m) compare favorably to theoretical predictions provided the product species thermal emittance and optical flame thickness are corrected for. However recent heat flux measurements from two large-scale horizontally orientated hydrogen flames (17.4 and 45.9 m respectively) revealed that current methods underpredicted the flame radiant fraction by 40% or more. Newly developed weighted source flame radiation models have demonstrated substantial improvement in the heat flux predictions particularly in the near-field and allow for a sensible way to correct potential ground surface reflective irradiance. These updated methods are still constrained by the fact that the flame is assumed to have a linear trajectory despite buoyancy effects that can result in significant flame deformation. The current paper discusses a method to predict flame centerline trajectories via a one-dimensional flame integral model which enables optimized placement of source emitters for weighted multi-source heat flux prediction methods. Flame shape prediction from choked releases was evaluated against flame envelope imaging and found to depend heavily on the notional nozzle model formulation used to compute the density weighted effective nozzle diameter. Nonetheless substantial improvement in the prediction of downstream radiative heat flux values occurred when emitter placement was corrected by the flame integral model regardless of the notional nozzle model formulation used.

The scenario of detonative ignition in shocked mixture is significant because it is a contributor to deflagration to detonation transition for example following shock reflections. However even in one dimension simulation of ignition between a contact surface or a flame and a shock moving into a combustible mixture is difficult because of the singular nature of the initial conditions. Initially as the shock starts moving into reactive mixture the region filled with reactive mixture has zero thickness. On a fixed grid the number of grid points between the shock and the contact surface increases as the shock moves away from the latter. Due to initial lack of resolution in the region of interest staircasing may occur whereby the resulting plots consist of jumps between few values a few grid points and these numerical artifacts are amplified by the chemistry which is very sensitive to temperature leading to unreliable results. The formulation is transformed replacing time and space by time and space over time as the independent variables. This frame of reference corresponds to the self-similar formulation in which the non-reactive problem remains stationary and the initial conditions are well-resolved. Additionally a solution obtained from short time perturbation is used as initial condition at a time still short enough for the perturbation to be very accurate but long enough so that there is sufficient resolution. The numerical solution to the transformed problem is obtained using an essentially non-oscillatory algorithm which is adequate not only for the early part of the process but also for the latter part when chemistry leads to appearance of a shock and eventually a detonation wave is formed. A validation study was performed and the results were compared with the literature for single step Arrhenius chemistry. The method and its implementation were found to be effective. Results are presented for values of activation energy ranging from mild to stiff.

In the problem of the protection by the consequences of an explosion is actual for many industrial application involving storage of gas like methane or hydrogen refuelling stations and so on. A simple and economic way to reduce the peak pressure associated to a deflagration is to supply to the confined environment an opportune surface substantially less resistant then the protected structure typically in stoichiometric conditions the peak pressure reduction is around the 8 bars for a generic hydrocarbon combustion in an adiabatic system lacking of whichever mitigation system. In general the problem is the forecast of the peak pressure value (PMAX) of the explosion. This problem is faced using CFD codes modelling the structure in which the explosion is located and setting the main parameters like concentration of the gas in the mixture the volume available the size of vent area and obstacles (if included) and so on. In this work the idea is to start from empirical data to train a Neural Network (NN) in order to find the correlation among the parameters regulating the phenomenon. Associated to this prediction a fuzzy model will provide to quantify the uncertainty of the predicted value.

The formation of a flammable hydrogen-air mixture is a major safety concern especially for closed space. This hazardous situation can arise when considering permeation from a car equipped with a composite compressed hydrogen tank with a non-metallic liner in a closed garage. In the following paper a scenario is developed and analysed with a simplified approach and a numerical simulation in order to estimate the evolution of hydrogen concentration. The system is composed of typical size garage and hydrogen car’s tank. Some parameters increasing permeation rate (i.e. tank’s material thickness and pressure) have been chosen to have a conservative approach. A close look on the top of tank surface showed that the concentration grows as square root of time and does not exceed 8.2×10-3 % by volume. Also a simplified comparative analysis estimated that the buoyancy of hydrogen-air mixture prevails on the diffusion 35 seconds after permeation starts in good agreement with simulation where time is at about 80 seconds. Finally the numerical simulations demonstrated that across the garage height the hydrogen is nearly distributed linearly and the difference in hydrogen concentration at the ceiling and floor is negligible (i.e. 3×10-3 %).

The paper presents an overview of the main results of the EC NOE HySafe activity to estimate an allowable hydrogen permeation rate for automotive legal requirements and standards. The work was undertaken as part of the HySafe internal project InsHyde.<br/>A slow long term hydrogen release such as that due to permeation from a vehicle into an inadequately ventilated enclosed structure is a potential risk associated with the use of hydrogen in automotive applications. Due to its small molecular size hydrogen permeates through the containment materials found in compressed gaseous hydrogen storage systems and is an issue that requires consideration for containers with non-metallic (polymer) liners. Permeation from compressed gaseous hydrogen storage systems is a current hydrogen safety topic relevant to regulatory and standardisation activities at both global and regional levels.<br/>Various rates have been proposed in different draft legal requirements and standards based on different scenarios and the assumption that hydrogen dispenses homogeneously. This paper focuses on the development of a methodology by HySafe Partners (CEA NCSRD. University of Ulster and Volvo Technology) to estimate an allowable upper limit for hydrogen permeation in automotive applications by investigating the behaviour of hydrogen when released at small rates with a focus on European scenario. The background to the activity is explained. reasonable scenarios are identified a methodology proposed and a maximum hydrogen permeation rate from road vehicles into enclosed structures is estimated The work is based on conclusions from the experimental and numerical investigations described by CEA NCSRD and the University of Ulster in related papers.