Medium-scale experiments on vented hydrogen deflagration

A series of medium-scale experiments on vented hydrogen deflagration was carried out at the KIT test side in a chamber of 1 × 1 × 1 m³ size with different vent areas. The experimental program was divided in three series: (1) uniform hydrogen–air mixtures; (2) stratified hydrogen–air mixtures within the enclosure; (3) a layer deflagration of uniform mixture. Different uniform hydrogen–air mixtures from 7 to 18% hydrogen were tested with variable vent areas 0.01–1.0 m². One test was done for rich mixture with 50% H₂. To vary a gradient of concentration, all the experiments with a stratified hydrogen–air mixtures had about 4%H₂ at the bottom and 10 to 25% H₂ at the top of the enclosure. Measurement system consisted of a set of pressure sensors and thermocouples inside and outside the enclosure. Four cameras combined with a schlieren system (BOS) for visual observation of combustion process through transparent sidewalls were used. Four experiments were selected as benchmark experiments to compare them with four times larger scale FM Global tests (Bauwens et al., 2011) and to provide experimental data for further CFD modelling. The nature of external explosion leading to the multiple pressure peak structure was investigated in details. Current work addresses knowledge gaps regarding indoor hydrogen accumulations and vented deflagrations. The experiments carried out within this work attend to contribute the data for improved criteria for hydrogen–air mixture and enclosure parameters to avoid unacceptable explosion overpressure. Based on theoretical analysis and current experimental data a further vent sizing technology for hydrogen deflagrations in confined spaces should be developed, taking into account the peculiarities of hydrogen–air mixture deflagrations in presence of obstacles, concentration gradients of hydrogen–air mixtures, dimensions of a layer of flammable cloud, vent inertia, etc.     

Numerical study of the effect of obstacles on the spontaneous ignition of high-pressure hydrogen

A numerical simulation of the spontaneous ignition of high-pressure hydrogen in a duct with two obstacles on the walls is conducted to explore the spontaneous ignition mechanisms. Two-dimensional rectangular ducts are adopted, and the Navier–Stokes equations with a detailed chemical kinetic mechanism are solved by using direct numerical simulations. In this study, we focus on the effects of the initial pressure of hydrogen and the position of the obstacles on the ignition mechanisms. Our results demonstrate that the presence of obstacles significantly changes the spontaneous ignition mechanisms producing three distinct ignition mechanisms. In addition, the position of the obstacles drastically changes the interaction of shock waves with the contact surface, and spontaneous ignition may take place at a relatively low pressure in some obstacle positions, which is attributed to the propagation direction and interaction timing of two reflected shock waves.     

LES – DFSD modelling of vented hydrogen explosions in a small-scale combustion chamber

Accidental explosions are a plausible danger to the chemical process industries. In the event of a gas explosion, any obstacles placed within the path of the flame generate turbulence, which accelerates the transient flame and raises explosion overpressure, posing a safety hazard. This paper presents numerical studies using an in-house computational fluid dynamics (CFD) model for lean premixed hydrogen/air flame propagations with an equivalence ratio of 0.7. A laboratory-scale combustion chamber is used with repeated solid obstacles. The transient compressible large eddy simulation (LES) modelling technique combined with a dynamic flame surface density (DFSD) combustion model is used to carry out the numerical simulations in three-dimensional space. The study presented uses eight different baffle configurations with two solid obstructions, which have area blockage ratios of 0.24 and 0.5. The flame speed, maximum rate of pressure-rise as well as peak overpressure magnitude and timing are presented and discussed. Numerical results are validated against available published experimental data. It is concluded that, increasing the solid obstacle area blockage ratio and the number of consecutive baffles results in a raised maximum rate of pressure rise, higher peak explosion overpressure and faster flame propagation. Future model development would require more experimental data, probably in a more congested configuration.     

Experimental investigation of fast flame propagation in stratified hydrogen–air mixtures in semi-confined flat layers

This paper presents results of an experimental investigation on fast flame propagation and the deflagration-to-detonation transition (DDT) and following detonation propagation in a semi-confined flat layer filled with stratified hydrogen–air mixtures. The experiments were performed in a transparent, rectangular channel open from below. The combustion channel has a width of 0.3 m and a length of 2.5 m. The effective layer thickness in the channel was varied by using different linear hydrogen concentration gradients. The method to create quasi-linear hydrogen concentration gradients that differ in the range and slope is also presented. The ignited mixtures were accelerated quickly to sonic flame speed in the first obstructed part of the channel. The interaction of the fast flame propagation with different obstacle set-ups was studied in the second part of the channel. The experimental results show an initiation of DDT by one additional metal grid in the obstructed semi-confined flat layer. Detonation propagation and failed detonation propagation were observed in obstructed and unobstructed parts of the channel.     

Validation of geometry modelling approaches for offshore gas dispersion simulations

Computational Fluid Dynamics (CFD) codes are widely used for gas dispersion studies on offshore installations. The majority of these codes use single-block Cartesian grids with the porosity/distributed-resistance (PDR) approach to model small geometric details. Computational cost of this approach is low since small-scale obstacles are not resolved on the computational mesh. However, there are some uncertainties regarding this approach, especially in terms of grid dependency and turbulence generated from complex objects. An alternative approach, which can be implemented in general-purpose CFD codes, is to use body-fitted grids for medium to large-scale objects whilst combining multiple small-scale obstacles in close proximity and using porous media models to represent blockage effects. This approach is validated in this study, by comparing numerical predictions with large-scale gas dispersion experiments carried out in DNV GL's Spadeadam test site. Gas concentrations and gas cloud volumes obtained from simulations are compared with measurements. These simulations are performed using the commercially available ANSYS CFX, which is a general-purpose CFD code. For comparison, further simulations are performed using CFX where small-scale objects are explicitly resolved. The aim of this work is to evaluate the accuracy and efficiency of these different geometry modelling approaches.     

A study of the blast wave shape from elongated VCEs

Elongated congestion patterns are common at chemical processing and petroleum refining facilities due to the arrangement of processing units. The accidental vapor cloud explosion (VCE) which occurred at the Buncefield, UK facility involved an elongated congested volume formed by the trees and undergrowth along the site boundary. Although elongated congested volumes are common, there have been few evaluations reported for the blast loads produced by elongated VCEs. Standard VCE blast load prediction techniques do not directly consider the impact of this congested volume geometry versus a more compact geometry.This paper discusses an evaluation performed to characterize the blast loads from elongated VCEs and to identify some significant differences in the resulting blast wave shape versus those predicted by well-known VCE blast load methodologies (e.g., BST and TNO MEM). The standard blast curves are based on an assumption that the portion of the flammable gas cloud participating in the VCE is hemispherical and located at grade level. The results of this evaluation showed that the blast wave shape for an elongated VCE in the near-field along the long-axis direction is similar to that for an acoustic wave generated in hemispherical VCEs with a low flame speed. Like an acoustic wave, an elongated VCE blast wave has a very quick transition from the positive phase peak pressure to the negative phase peak pressure, relative to the positive phase duration. The magnitude of the applied negative pressure on a building face depends strongly on the transition time between the positive and negative phase peak pressures, and this applied negative phase can be important to structural response under certain conditions. The main purpose of this evaluation was to extend previous work in order to investigate how an elongated VCE geometry impacts the resultant blast wave shape in the near-field. The influence of the normalized flame travel distance and the flame speed on the blast wave shape was examined. Deflagration and deflagration-to-detonation transition regimes were also identified for unconfined elongated VCEs as a function of the normalized flame travel distance and flame speed attained at a specified flame travel distance.     

Explosion pressure prediction via polynomial mathematical correlation based on advanced CFD Modelling

Computational Fluid Dynamics CFD can be used as a powerful tool supporting engineers throughout the steps of the design. The combination of CFD with response surface methodology can play an important role in such cases. During the conceptual engineering design phase, a quick response is always a matter of urgency. During this phase even a sketch of the geometrical model is rare. Therefore, the utilisation of typical response surface developed for congested and confined environment rather than CFD can be an important tool to help the decision making process, when the geometrical model is not available, provided that similarities can be considered when taking into account the characteristic of the geometry in which the response surface was developed. The present work investigates how three different types of response surfaces behave when predicting overpressure in accidental scenarios based on CFD input. First order, partial second order and complete second order polynomial expressions are investigated. The predicted results are compared with CFD findings for a classical offshore experiment conducted by British Gas on behalf of Mobil and good agreement is observed for higher order response surfaces. The higher order response surface calculations are also compared with CFD calculations for a typical offshore module and good agreement is also observed.     

Explosibility of hydrogen–graphite dust hybrid mixtures

To evaluate the hazard of combined hydrogen/dust explosions under severe accident conditions in International Thermonuclear Experimental Reactor (ITER), standard method of 20-L-sphere was used to measure the explosion indices of 4-μm fine graphite dust in lean hydrogen/air mixtures. The mixtures were ignited by a weak electric spark. The tested fuel concentrations were 8–18 vol% H₂ and 25–250 g/m³ dust. If the hydrogen content is higher than 10 vol%, the dust constituent can be induced to explode by the hydrogen explosion initiated by a weak electric spark. Depending on the fuel component concentrations, the explosions proceed in either one or two stages. In two-stage explosions occurring at low hydrogen and dust concentrations, the mixture ignition initiates first a fast hydrogen explosion followed by a slower phase of the dust explosion. With increasing dust concentration, the dust explodes faster and can overlap the hydrogen-explosion stage. At higher hydrogen concentrations, the hybrid mixtures explode in one stage, with hydrogen and dust reacting at the same time scale. Maximum overpressures of hybrid explosions are higher than those observed with hydrogen alone; maximum rates of pressure rise are lower in two-phase explosions and, generally, higher in one-stage explosions, than those characteristic of the corresponding H₂/air mixtures.