Vented explosion overpressures from combustion of hydrogen and hydrocarbon mixtures

Experimental data obtained for hydrogen mixtures in a room-size enclosure are presented and compared with data for propane and methane mixtures. This set of data was also used to develop a three-dimensional gasdynamic model for the simulation of gaseous combustion in vented enclosures. The experiments were performed in a 64 m³ chamber with dimensions of 4.6 × 4.6 × 3.0 m and a vent opening on one side and vent areas of either 2.7 or 5.4 m² were used. Tests were performed for three ignition locations, at the wall opposite the vent, at the center of the chamber or at the center of the wall containing the vent. Hydrogen-air mixtures with concentrations close 18% vol. were compared with stoichiometric propane-air and methane-air mixtures. Pressure data, as function of time, and flame time-of-arrival data were obtained both inside and outside the chamber near the vent. Modeling was based on a Large Eddy Simulation (LES) solver created using the OpenFOAM CFD toolbox using sub-grid turbulence and flame wrinkling models. A comparison of these simulations with experimental data is discussed.     

Experimental study of vented hydrogen deflagration with ignition inside and outside the vented volume

Experiments were carried out inside a 25 m³ vented combustion test facility (CVE) with a fixed vent area sealed by a plastic sheet vent. Inside the CVE, a 0.64 m³ open vent box, called RED-CVE was placed. The vent of the RED-CVE was left open and three different vent area were tested. Two different mixing fans, one for each compartment, were used to establish homogeneous H₂ concentrations. This study examined H₂ concentrations in the range between 8.5% vol. to 12.5% vol. and three different ignition locations, (1) far vent ignition, (2) inside the RED-CVE box ignition and (3) near vent ignition (the vent refers to the CVE vent). Peak overpressures generated inside the test facility and the smaller compartment were measured. The results indicate that the near vent ignition generates negligible peak overpressures inside the test facility as compared to those originated by far vent ignition and ignition inside the RED-CVE box. The experiments with far vent ignition showed a pressure increase with increasing hydrogen concentration which reached a peak value at 11% vol. concentration and then decreased showing a non-monotonic behaviour. The overpressure measured inside the RED-CVE was higher when the ignition was outside the box whereas the flame entered the box through the small vent.     

Vented hydrogen–air deflagration in a small enclosed volume

Since the rapid development of hydrogen stationary and vehicle fuel cells the last decade, it is of importance to improve the prediction of overpressure generated during an accidental explosion which could occur in a confined part of the system. To this end, small-scale vented hydrogen–air explosions were performed in a transparent cubic enclosure with a volume of 3375 cm³. The flame propagation was followed with a high speed camera and the overpressure inside the enclosure was recorded using high frequency piezoelectric transmitters. The effects of vent area and ignition location on the amplitude of pressure peaks in the enclosed volume were investigated. Indeed, vented deflagration generates several pressures peaks according to the configuration and each peak can be the dominating pressure. The parametric study concerned three ignition locations and five square vent sizes.     

Effect of hydrogen concentration on vented explosion overpressures from lean hydrogen–air deflagrations

Experimental data from vented explosion tests using lean hydrogen–air mixtures with concentrations from 12 to 19% vol. are presented. A 63.7-m³ chamber was used for the tests with a vent size of either 2.7 or 5.4 m². The tests were focused on the effect of hydrogen concentration, ignition location, vent size, and obstacles on the pressure development of a propagating flame in a vented enclosure. The dependence of the maximum pressure generated on the experimental parameters was analyzed. It was confirmed that the pressure maxima are caused by pressure transients controlled by the interplay of the maximum flame area, the burning velocity, and the overpressure generated outside of the chamber by an external explosion. A model proposed earlier to estimate the maximum pressure for each of the main pressure transients was evaluated for the various hydrogen concentrations. The effect of the Lewis number on the vented explosion overpressure is discussed.     

Experimental study of the effects of vent geometry on the dispersion of a buoyant gas in a small enclosure

We present an experimental study on the dispersion of helium in an enclosure of 1 m³ with natural ventilation through one vent. Three vent geometries have been studied. Injection parameters have been varied so that the injection Richardson number ranges from 2·10⁻⁶ to 9 and the volume Richardson number, which gives the ability of the release to mix the enclosure content ranges from 8·10⁻⁴ to 900. It has been found that the vertical distribution of helium volume fraction can exhibit significant gradient. Nevertheless, the results are compared to the simple analytical model based on the homogenous mixture hypothesis which gives fairly good estimates of the maximum helium volume fraction.     

Hydrogen non-premixed combustion in enclosure with one vent and sustained release: Numerical experiments

Numerical experiments are performed to understand different regimes of hydrogen non-premixed combustion in an enclosure with passive ventilation through one horizontal or vertical vent located at the top of a wall. The Reynolds averaged Navier–Stokes (RANS) computational fluid dynamics (CFD) model with a reduced chemical reaction mechanism is described in detail. The model is based on the renormalization group (RNG) k-ε turbulence model, the eddy dissipation concept (EDC) model for simulation of combustion coupled with the 18-step reduced chemical mechanism (8 species), and the in-situ adaptive tabulation (ISAT) algorithm that accelerates the reacting flow calculations by two to three orders of magnitude. The analysis of temperature and species (hydroxyl, hydrogen, oxygen, water) concentrations in time, as well as the velocity through the vent, shed a light on regimes and dynamics of indoor hydrogen fires. A well-ventilated fire is simulated in the enclosure at a lower release flow rate and complete combustion of hydrogen within the enclosure. Fire becomes under-ventilated at higher release flow rates with two different modes observed. The first mode is the external flame stabilised at the enclosure vent at moderate release rates, and the second mode is the self-extinction of combustion inside and outside the enclosure at higher hydrogen release rates. The simulations demonstrated a complex reacting flow dynamics in the enclosure that leads to formation of the external flame or the self-extinction. The air intake into the enclosure at later stages of the process through the whole vent area is a characteristic feature of the self-extinction regime. This air intake is due to faster cooling of hot combustion products by sustained colder hydrogen leak compared to the generation of hot products by the ceasing chemical reactions inside the enclosure and hydrogen supply. In general, an increase of hydrogen sustained release flow rate will change fire regime from the well-ventilated combustion within the enclosure, through the external flame stabilised at the vent, and finally to the self-extinction of combustion throughout the domain.     

Experimental investigation on unconfined hydrogen explosion with different ignition height

Experimental research is performed to investigate the effects of ignition height on explosion characteristics in a 27 m³ hydrogen/air cloud. With the ignition height decreasing, the flame propagation velocity increases gradually. The flame travels in oscillating mode and the average oscillating frequency lies between 145Hz and 155Hz. An original parameter τ, which involves flame scale and flame propagation velocity, is proposed to measure the effect of buoyancy. The higher the value of τ, the more obvious the buoyancy effect. As the ignition height increases, the critical flame scale for flame deceleration increases. The middle ignition height in the gas cloud causes the highest overpressure peak, overpressure impulse, overpressure rising and decreasing rate. As the ignition point approaches the initial gas boundary, the explosion intensity would decrease gradually. For the open space outside the flame, overpressure peak for the lower space is higher, while, the middle space experiences higher overpressure impulse.     

CFD modeling and consequence analysis of an accidental hydrogen release in a large scale facility

In this study, the consequences of an accidental release of hydrogen within large scale, (>15,000 m³), facilities were modeled. To model the hydrogen release, an LES Navier–Stokes CFD solver, called fireFoam, was used to calculate the dispersion and mixing of hydrogen within a large scale facility. The performance of the CFD modeling technique was evaluated through a validation study using experimental results from a 1/6 scale hydrogen release from the literature and a grid sensitivity study. Using the model, a parametric study was performed varying release rates and enclosure sizes and examining the concentrations that develop. The hydrogen dispersion results were then used to calculate the corresponding pressure loads from hydrogen-air deflagrations in the facility.     

A comparison study into low leak rate buoyant gas dispersion in a small fuel cell enclosure using plain and louvre vent passive ventilation schemes

Hydrogen, producing electricity in fuel cells, is a versatile energy source, but with risks associated with flammability. Fuel cells use enclosures for protection which need ventilating to remove hydrogen emitted during normal operation or from supply system leaks. Passive ventilation, using buoyancy driven flow is preferred to mechanical systems. Performance depends upon vent design, size, shape, position and number. Vents are usually plain rectangular openings, but environmentally situated enclosures use louvres for protection. The effect of louvres on passive ventilation is not clear and has therefore been examined in this paper. Comparison ‘same opening area’ louvre and plain vent tests were undertaken using a 0.144 m³ enclosure with opposing upper and lower vents and helium leaking from a 4 mm nozzle on the base at rates from 1 to 10 lpm, simulating a hydrogen leak. Louvres increased stratified level helium concentrations by typically in excess of 15%. The empirical data obtained was also used in a validation exercise with a SolidWorks: Flow Simulation CFD model, which provided a good qualitative representation of flow behaviour and close empirical data correlations.     

Hydrogen permeation from CGH2 vehicles in garages: CFD dispersion calculations and experimental validation

The time and space evolution of the distribution of hydrogen in confined settings was investigated computationally and experimentally for permeation from typical compressed gaseous hydrogen (CGH2) storage systems for buses or cars. The main goal was to examine whether hydrogen is distributed homogeneously within a garage-like facility or whether stratified conditions are developed, under certain conditions. The nominal hydrogen flow rate considered was 1.087 L/min in a bus facility with a volume of 681 m³. The release was assumed to be directed upwards from a 0.15 m diameter hole located at the middle part of the bus cylinders casing. Ventilation rates up to 0.03 air changes per hour (ACH) were considered. Simulated time periods extended up to 20 days. The CFD simulations performed with the ADREA-HF code showed that fully homogeneous conditions exist for low ventilation rates, while stratified conditions prevail for higher ventilation rates. Regarding flow structure it was found that the vertical concentration profiles can be considered as the superposition of the concentration at the floor (driven by diffusion) plus a concentration difference between floor and ceiling (driven by buoyancy forces). In all cases considered this concentration difference was found to be less than 0.5%. The dispersion experiments were performed in a large scale garage-like enclosure of 40 m³ using helium (GARAGE facility). Comparison between CFD simulations and experiments showed that the predicted concentrations were in good agreement with the experimental data. Finally, simulations were performed using two integral models: the fully homogeneous model and a two-layer model and the results were compared both against CFD and the experimental data.     

Numerical and physical requirements to simulation of gas release and dispersion in an enclosure with one vent

Numerical and physical requirements to simulations of sub-sonic release and dispersion of light gas in an enclosure with one vent are described and discussed. Six validation experiments performed at CEA in a fuel cell-like enclosure of sizes H × W × L = 126 × 93 × 93 cm with one vent, either W × H = 90 × 18 cm (vent A) or 18 × 18 cm (B) or 1 cm in diameter (C), with a vertical upward helium release from a pipe of internal diameter either 5 mm or 20 mm located 21 cm above the floor centre, were used in a parametric study comprising 17 numerical simulations. Three CFD models were applied, i.e. laminar, standard k-?, and dynamic LES Smagorinsky–Lilly, to clarify a range of their applicability and performance. The LES model consistently demonstrated the best performance in reproduction of measured concentrations throughout the whole range of experimental conditions, including laminar, transitional and turbulent releases even with large CFL numbers. The laminar and the standard k-? models were under performing in the reproduction of turbulent and laminar releases respectively, as expected, as well as in simulation of transitional flows. The laminar model demonstrated high sensitivity to the CFL (Courant–Friedrichs–Lewy) number even below the best practices limit of 40. Three different computational domains and grids were used in order to clarify the influence of mesh quality on the capability of simulations to reproduce the experimental data. It is concluded that physically substantiated choice of CFD model, the control of the CFL number (and released gas mass balance where appropriate), and the mesh quality can have a strong effect on the capability of simulations to reproduce experiments and, in general, on the reliability of CFD tools for application in hydrogen safety engineering.     

Experimental research on unconfined hydrogen explosion of different gas scale and the overpressure prediction method

In this research, unconfined hydrogen experiments are performed in 1 m³ and 27 m³ gas scale with gas concentration varying from lean-burn to rich burn. The results show that the flame travels fastest upwards and slowest downwards, which makes the flame shape irregularly spherical. The critical flame scales for the extra acceleration in the upward direction and for the deceleration in the downward direction are both smaller in 1 m³ gas scale. The acceleration exponent α is higher in the upward direction. With the gas scale increasing, the value of α increases gradually. For φ = 0.8, 1.0 and 1.5, the equivalent flame radius and the explosion overpressure in different gas scales overlap before the film rupture. According to the wrinkled laminar flame assumption and the self-similar theory, an overpressure prediction model is proposed based on the wrinkling factor Ξ_Δ. The predicted results agree well with the experimental data before the film rupture.     

Natural and forced ventilation study in an enclosure hosting a fuel cell

The purpose of the experimental work described in this paper is the determination of the ventilation requirements in enclosures containing fuel cells such that, in the case of a small non catastrophic release, the H₂ concentration in air for zone 2 ATEX (2% v/v) is not exceeded. A full scale fuel cell was placed inside the experimental facility having 25 m³ volume. Three different leaks were investigated (40, 90 and 180 Nlt/min) and H₂ concentrations were measured at five locations inside the facility. Several vent areas were examined for the cases of natural ventilation. When natural ventilation failed to ensure H₂ concentrations less than 2% v/v in the facility, mechanical ventilation using two fans was investigated.Based on the experimental set up, it was found that natural ventilation is sufficient when the air-flow calculated from ATEX guidelines is higher than 0.009 m³/s and the release flow rate corresponds to a non-catastrophic release, i.e. 40 Nlt/min. For higher release flow rates most of the ventilation configurations were not sufficient to maintain a H₂ concentration less than 2% v/v.All forced ventilation configurations examined (together with the free ventilation areas used) were sufficient to maintain a H₂ concentration below 2% v/v for 40 Nlt/min and 90 Nlt/min release flow rates. For the higher release flow rate of 180 Nlt/min, most of the forced ventilation configurations were insufficient.     

Modeling of vented deflagrations

A mathematical model of vented gas-phase deflagrations is presented. By introducing several empirical parameters, account is taken of initial turbulence in the gases, flame acceleration due to hydrodynamic instabilities prior to vent opening, and increased burning velocity due to turbulence generated by the venting process. Additionally, a mixture of burned and unburned gases is vented. Essential information needed to compute the pressure development during vented deflagrations (or in large closed vessels) is the rate of increase of flame area due to cell formation in the flame front prior to the vent opening.The model has been tested against methane/air mixtures at initial pressures of 45 psia in vessels up to 3.8 m³ in volume. Good agreement has been obtained.Further work is underway to gather data on vented deflagrations for gases such as propane, ethylene, and hydrogen (which represent a series of increasing burning velocities) and to investigate more fully the effect of initial turbulence and elevated pressures.     

Analysis of transient supersonic hydrogen release,dispersion and combustion

A hydrogen leak from a facility, which uses highly compressed hydrogen gas (714 bar, 800 K) during operation was studied. The investigated scenario involves supersonic hydrogen release from a 10 cm² leak of the pressurized reservoir, turbulent hydrogen dispersion in the facility room, followed by an accidental ignition and burn-out of the resulting H₂-air cloud. The objective is to investigate the maximum possible flame velocity and overpressure in the facility room in case of a worst-case ignition. The pressure loads are needed for the structural analysis of the building wall response. The first two phases, namely unsteady supersonic release and subsequent turbulent hydrogen dispersion are simulated with GASFLOW-MPI. This is a well validated parallel, all-speed CFD code which solves the compressible Navier-Stokes equations and can model a broad range of flow Mach numbers. Details of the shock structures are resolved for the under-expanded supersonic jet and the sonic-subsonic transition in the release. The turbulent dispersion phase is simulated by LES. The evolution of the highly transient burnable H₂-air mixture in the room in terms of burnable mass, volume, and average H₂-concentration is evaluated with special sub-routines. For five different points in time the maximum turbulent flame speed and resulting overpressures are computed, using four published turbulent burning velocity correlations. The largest turbulent flame speed and overpressure is predicted for an early ignition event resulting in 35–71 m/s, and 0.13–0.27 bar, respectively.     

Experimental study of end-vented hydrogen deflagrations in a 40-foot container

Experiments were conducted in an enclosure with the same overall dimensions as a 40-foot ISO container to study the vented hydrogen-air deflagrations. This work focuses on the effects of hydrogen concentration, ignition location and obstacles on the overpressure and the structural response of the container wall. For center ignition, three overpressure peaks, which resulted from the vent opening, Helmholtz oscillation and acoustic oscillation, respectively, were recorded inside the container without obstacles. However, with the increase of hydrogen concentration, the third overpressure peak disappears when the obstacles are added in the container. Unlike center ignition, only two overpressure peaks were observed for back ignition. Due to the difference in reactivity of hydrogen-air mixture, the first overpressure peak is generated by the vent burst for low hydrogen concentration, or the venting of flame for high hydrogen concentration. The overpressure induced by the flame-acoustic interaction was not monitored with the increase of the hydrogen concentration and the installation of obstacles for back ignition. The overpressure for back ignition is more influenced by the obstacles than that for center ignition, when hydrogen concentration is larger than 12%. The displacement-time curves share similar trends with the pressure-time curves. The first peak displacement changes linearly with the corresponding first peak overpressure. However, the displacement caused by the second overpressure peak is significantly increased, especially for high hydrogen concentration and back ignition in the case with two obstacles.     

Hydrogen dispersion phenomenon in nominally closed spaces

The hydrogen dispersion phenomenon in an enclosure depends on the ratio of the gas buoyancy-induced momentum and diffusive motions. Random diffusive motions of individual gas particles become dominative when the release momentum is low, and a uniform hydrogen concentration appears in the enclosure instead of the gas cumulation below the ceiling. The expected hydrogen behavior could be projected by the Froude number, which value ~1 predicts a decline of buoyancy. This paper justifies this hypothesis by demonstrating full-scale experimental results of hydrogen dispersion within a confined space under six different release variations. During the experiments, hydrogen was released into the test room of 60 m³ volume in two methods: through a nozzle and through 21 points evenly distributed on the emission box cover (multi-point release). Each release method was tested with three volume flow rates (3.2 × 10⁻³ m³/s, 1.6 × 10⁻³ m³/s, 3.3 × 10⁻⁴ m³/s). The tests confirm the decrease of hydrogen buoyancy and its stratification tendencies when the Mach, Reynolds, and Froud number values decrease. Because the hydrogen dispersion phenomenon would impact fire and explosive hazards, the presented experimental results could help fire protection systems be in an enclosure designed, allowing their effectiveness optimization.     

CFD modeling of hydrogen deflagration in a tunnel

In this paper CFD modeling techniques are used to simulate deflagration in homogenous, near stoichiometric hydrogen–air mixture in a model of a tunnel. The tunnel is 78.5 m long. Hydrogen–air mixture is located in a 10 m long region in the middle of the tunnel. Two cases are studied: one with a complete empty tunnel and one with the presence of four vehicles near the center of the tunnel. The combustion model is based on the turbulent flame speed concept. The turbulent flame speed is a modification of Yakhot's equation, in order to account for additional physical mechanisms. A sensitivity analysis for the ψ parameter of the combustion model and for the mesh resolution was made. The agreement between experimental and computational results concerning the value of the maximum pressure, and the time it appears, was satisfactory in both empty and non-empty tunnel case. The sensitivity analysis for the parameter of the combustion model showed that even small changes in it can have impact on the simulating results, whereas the sensitivity analysis of the mesh resolution did not reveal any significant differences. Finally, the effect of the turbulence model is examined (LES and RANS type of model). The only significant difference in the results between LES and RANS model was the arrival time of the pressure peak. A delay in the arrival time in the case of the RANS model was observed.