Fast Flame Research Articles

Chapman–Jouguet deflagrations and their transition to detonation

We study experimentally fast flames and their transition to detonation for five different hydrocarbons, namely methane, ethane, ethylene, acetylene, and propane with oxygen as the oxidizer. Following the interaction of a detonation wave with a column of cylinders of varying blockage ratio, the experiments demonstrate that the fast flames established are Chapman–Jouguet deflagrations, in excellent agreement with the self-similar model of Radulescu et al. (2005). The experiments indicate that these Chapman–Jouguet deflagrations dynamically restructure and amplify into fewer stronger modes until the eventual transition to detonation. The transition length to a self-sustained detonation was found to correlate very well with the mixtures’ sensitivity to temperature fluctuations, reflected by the χ parameter introduced by Radulescu, which is the product of the non-dimensional activation energy Ea/RT and the ratio of chemical induction to reaction time ti/tr. Correlation of the measured deflagration to detonation transition (DDT) lengths determined that the relevant characteristic time scale from chemical kinetics controlling DDT is the energy release or excitation time tr. Correlations with the cell size also capture the dependence of the DDT length on χ for fixed blockage ratios.

Critical role of blockage ratio for flame acceleration in channels with tightly spaced obstacles

A conceptually laminar mechanism of extremely fast flame acceleration in obstructed channels, identified by Bychkov et al. [“Physical mechanism of ultrafast flame acceleration,” Phys. Rev. Lett. 101, 164501 (2008)], is further studied by means of analytical endeavors and computational simulations of compressible hydrodynamic and combustion equations. Specifically, it is shown how the obstacles length, distance between the obstacles, channel width, and thermal boundary conditions at the walls modify flame propagation through a comb-shaped array of parallel thin obstacles. Adiabatic and isothermal (cold and preheated) side walls are considered, obtaining minor difference between these cases, which opposes the unobstructed channel case, where adiabatic and isothermal walls provide qualitatively different regimes of flame propagation. Variations of the obstructed channel width also provide a minor influence on flame propagation, justifying a scale-invariant nature of this acceleration mechanism. In contrast, the spacing between obstacles has a significant role, although it is weaker than that of the blockage ratio (defined as the fraction of the channel blocked by obstacles), which is the key parameter of the problem. Evolution of the burning velocity and the dependence of the flame acceleration rate on the blockage ratio are quantified. The critical blockage ratio, providing the limitations for the acceleration mechanism in channels with comb-shaped obstacles array, is found analytically and numerically, with good agreement between both approaches. Additionally, this comb-shaped obstacles-driven acceleration is compared to finger flame acceleration and to that produced by wall friction.

Simulations of gasoline engine combustion and emissions using a chemical-kinetics-based turbulent premixed combustion modeling approach

This work presents a turbulent premixed combustion modeling approach which is based on chemical kinetics. In this approach, the smallest length scales are of the order of 0.1–1.0 mm for typical engine simulations with a Reynolds-averaged Navier–Stokes turbulence model and, after adaptive mesh refinement technology is used to consider the magnitude of the subgrid field, the Reynolds-averaged Navier–Stokes turbulent flow field can be well resolved. For solution of the flame front, an artificially thickened laminar flame concept is introduced to balance the computational accuracy and the computational cost. Around the artificially thickened laminar flame front, a special grid resolution strategy is designed, i.e. using much finer resolution in the normal direction of the flame front and typical adaptive mesh refinement resolution in the other two perpendicular directions. Then, chemical kinetics can be applied to the chemistry process which occurs in the flame front. To use this chemical-kinetics-based turbulent premixed combustion modeling approach better, a good chemical kinetics mechanism is very important. For this reason a practical primary-reference-fuel chemical kinetics mechanism is improved and validated in present work. The newly improved mechanism resolves several issues in the existing mechanisms, including unrealistically fast autoignition reactions and limited laminar flame speed validation. After reoptimization of those laminar-flame-speed-related reactions, the new mechanism can correctly compute the laminar flame speeds for a wide range of Ford spark ignition engines and for various operating conditions. Using this combustion modeling approach together with the new mechanism, simulations of the combustion and the emissions of several spark ignition engines for typical operating conditions were carried out. The simulated in-cylinder pressures, the simulated burn rates, and the simulated emissions including the brake specific carbon monoxide emissions, the nitrogen oxide emissions, and the unburned hydrocarbon emissions are compared with the experimental data, and very good agreement is found without tuning any model constants.

Combustion instability characteristics of H2/CO/CH4 syngases and synthetic natural gases in a partially-premixed gas turbine combustor: Part I—Frequency and mode analysis

This paper presents the results of combustion instability characteristics of H2/CO/CH4 synthetic gases and synthetic natural gases by testing a scale-downed industrial GE7EA gas turbine that features a partially-premixed swirl-stabilized combustion. H2/CO/CH4 syngases ranging from 0% to 100% and synthetic natural gases containing H2 from 0% to 30% were tested and analyzed in view of H2, which has distinctive combustion characteristics of fast burning and high flame temperature. Interestingly, as features of the partially-premixed flames that were adapted to inhibit flashback from fast-burning H2-containing syngas, high multi-mode combustion instabilities such as 2nd, 3rd, 4th, and over were observed without fundamental longitudinal mode. To investigate these high multi-mode combustion instabilities, combustion properties such as the laminar burning velocity, adiabatic flame temperature and ignition delay time were calculated for the all tested fuels and their dependence within them and dependence on instability frequencies were closely analyzed. Test results demonstrated that H2 contents are the root cause of the high-order instability mode and shifting within modes, since fast burning H2 affects the chemical time and flame temperature of syngas more than that of CO and CH4. The relationships between the combustion properties have also illustrated that the sensitive change of combustion properties are responsible for generating flame oscillations of high multi-mode and sensible frequency and mode shifting. Therefore, the laminar burning velocity, adiabatic flame temperature and ignition delay time should be considered as important parameters when designing a gas turbine combustor; moreover, deciding operating conditions for new fuels and careful management of fuel quality, particularly in view of H2, is required when operating integrated gasification combined cycle plants and synthetic natural gas plants.

Premixed flame propagation in vertical tubes

Analytical treatment of the premixed flame propagation in vertical tubes with smooth walls is given. Using the on-shell flame description, equations for a quasi-steady flame with a small but finite front thickness are obtained and solved numerically. It is found that near the limits of inflammability, solutions describing upward flame propagation come in pairs having close propagation speeds and that the effect of gravity is to reverse the burnt gas velocity profile generated by the flame. On the basis of these results, a theory of partial flame propagation driven by a strong gravitational field is developed. A complete explanation is given of the intricate observed behavior of limit flames, including dependence of the inflammability range on the size of the combustion domain, the large distances of partial flame propagation, and the progression of flame extinction. The role of the finite front-thickness effects is discussed in detail. Also, various mechanisms governing flame acceleration in smooth tubes are identified. Acceleration of methane-air flames in open tubes is shown to be a combined effect of the hydrostatic pressure difference produced by the ambient cold air and the difference of dynamic gas pressure at the tube ends. On the other hand, a strong spontaneous acceleration of the fast methane-oxygen flames at the initial stage of their evolution in open-closed tubes is conditioned by metastability of the quasi-steady propagation regimes. An extensive comparison of the obtained results with the experimental data is made.

Flame Visualization and Mechanism of Fast Flame Propagation through a Meso-scale Packed Porous Bed in a High-pressure Environment

Flame Visualization and Mechanism of Fast Flame Propagation through a Meso-scale Packed Porous Bed in a High-pressure Environment

Fast fire flame detection in surveillance video using logistic regression and temporal smoothing

Real-time detection of fire flame in video scenes from a surveillance camera offers early warning to ensure prompt reaction to devastating fire hazards. Many existing fire detection methods based on computer vision technology have achieved high detection rates, but often with unacceptably high false-alarm rates. This paper presents a reliable visual analysis technique for fast fire flame detection in surveillance video using logistic regression and temporal smoothing. A candidate fire region is determined according to the color component ratio and motion cue of fire flame obtained by background subtraction. The candidate fire region is examined for genuine fire flame in terms of the proposed fire probability computed using logistic regression of prominent features of size, motion, and color information. Temporal smoothing is employed to reduce false alarm rates at a slight decrease in sensitivity. Experiments conducted on various benchmarking databases demonstrate that the proposed scheme successfully distinguishes fire flame from the background as well as moving fire-like objects in real-world indoor and outdoor video surveillance settings. The average time to fire detection was fastest among the state-of-the-art video-based fire flame detection techniques for comparison.

Opposed-flow Flame Spread Over Solid Fuels in Microgravity: the Effect of Confined Spaces

Effects of confined spaces on flame spread over thin solid fuels in a low-speed opposing flow is investigated by combined use of microgravity experiments and computations. The flame behaviors are observed to depend strongly on the height of the flow tunnel. In particular, a non-monotonic trend of flame spread rate versus tunnel height is found, with the fastest flame occurring in the 3 cm high tunnel. The flame length and the total heat release rate from the flame also change with tunnel height, and a faster flame has a larger length and a higher heat release rate. The computation analyses indicate that a confined space modifies the flow around the spreading flame. The confinement restricts the thermal expansion and accelerates the flow in the streamwise direction. Above the flame, the flow deflects back from the tunnel wall. This inward flow pushes the flame towards the fuel surface, and increases oxygen transport into the flame. Such a flow modification explains the variations of flame spread rate and flame length with tunnel height. The present results suggest that the confinement effects on flame behavior in microgravity should be accounted to assess accurately the spacecraft fire hazard.

Hydrogen combustion in a flat semi-confined layer with respect to the Fukushima Daiichi accident

Hydrogen accumulations at the top of a containment or reactor building may occur due to the interaction of molten corium and water followed by a severe accident of a nuclear reactor (TMI, Chernobyl, Fukushima Daiichi). The hydrogen that is released from the reactor accumulates usually as a stratified semi-confined layer of hydrogen–air mixture. A series of large scale experiments on hydrogen combustion and explosion in a semi-confined layer of uniform and non-uniform hydrogen–air mixtures in the presence of obstructions or without them was performed at the Karlsruhe Institute of Technology (KIT). Different flame propagation regimes from slow subsonic to relatively fast sonic flames and then to detonations were experimentally investigated in different geometries and then simulated with COM3D code with respect to evaluate the amount of hydrogen that was involved in the Fukushima Daiichi Accident (FDA).The experiments were performed in a horizontal semi-confined layer with the dimensions 9×3×0.6m with/without obstacles opened from below. The hydrogen concentration in the mixtures with air was varied in the range of 10–34vol.% without or with a gradient of 20–60vol.%H2/m. Effects of hydrogen concentration gradient, layer thickness, obstruction geometry, average and maximum hydrogen concentration on the flame propagation regimes were investigated with respect to evaluate the maximum pressure loads on internal structures. Blast wave strength and dynamics of propagation after the explosion of the hydrogen–air mixture layer were numerically simulated to reproduce the hydrogen explosion process during the Fukushima Daiichi Accident (Unit 1).

Structure and flame speed of dilute and dense layered coal-dust explosions

Multidimensional time-dependent simulations were performed to study the interaction of a stoichiometric methane–air detonation with layers of coal dust. The simulations solved equations representing a Eulerian kinetic-theory-based granular multiphase model applicable to dense and dilute particle volume fractions. These equations were solved using a high-order Godunov-based method for compressible fluid dynamics. Two dust layer concentrations were considered: loose with an initial volume fraction of 1%, and dense with an initial volume fraction of 47%. Each layer was simulated with two types of dust: reactive coal and inert ash. Burning of the coal particles results in a coupled complex consisting of an accelerating shock leading a coal-dust flame. The overall structure of the shock–flame complex resembles that of a premixed fast flame with length scales on the order of several meters. The large length scales are direct results of time needed to lift, mix, heat, and autoignite the particle. The flame speeds are large and much larger than the gas-phase velocity. Large spikes of flame speed are characteristic of the 47% case. These spikes and high flame speed are caused by pockets of coal dust autoigniting ahead of the flame. The flame is choked in the 1% case due to the gas-phase products exceeding the sonic velocity with respect to the flame. The 47% case is choked due to attenuation of pressure waves as they propagate through particles. Inert layers of dust substantially reduce the overpressure, impulse, and speed produced by propagating blast wave. The results also show that loose layers of dust are far more dangerous than dense layers. The shock and flame are more strongly coupled for loose layers, propagate at higher velocity, and produce large overpressures and impulses.

Pulsating instability and self-acceleration of fast turbulent flames

A series of three-dimensional numerical simulations is used to study the intrinsic stability of high-speed turbulent flames. Calculations model the interaction of a fully resolved premixed flame with a highly subsonic, statistically steady, homogeneous, isotropic turbulence. The computational domain is unconfined to prevent the onset of thermoacoustic instabilities. We consider a wide range of turbulent intensities and system sizes, corresponding to the Damköhler numbers Da = 0.1 − 6.0. These calculations show that turbulent flames in the regimes considered are intrinsically unstable. In particular, we find three effects. (1) Turbulent flame speed, ST, develops pulsations with the observed peak-to-peak amplitude STmax/STmin>10 and a characteristic time scale close to a large-scale eddy turnover time. Such variability is caused by the interplay between turbulence, which continuously creates the flame surface, and highly intermittent flame collisions, which consume the flame surface. (2) Unstable burning results in the periodic pressure build-up and the formation of pressure waves or shocks, when ST approaches or exceeds the speed of a Chapman-Jouguet deflagration. (3) Coupling of pressure gradients formed during pulsations with density gradients across the flame leads to the anisotropic amplification of turbulence inside the flame volume and flame acceleration. Such process, which is driven by the baroclinic term in the vorticity transport equation, is a reacting-flow analog of the mechanism underlying the Richtmyer-Meshkov instability. With the increase in turbulent intensity, the limit-cycle instability discussed here transitions to the regime described in our previous work, in which the growth of ST becomes unbounded and produces a detonation.

Influence of hydrodynamic instabilities on the propagation mechanism of fast flames

The present work investigates the structure of fast supersonic turbulent flames typically observed as precursors to the onset of detonation. These high speed deflagrations are obtained after the interaction of a detonation wave with cylindrical obstacles. Two mixtures having the same propensity for local hot spot formation were considered, namely hydrogen–oxygen and methane-oxygen. It was shown that the methane mixture sustained turbulent fast flames, while the hydrogen mixture did not. Detailed high speed visualizations of nearly two-dimensional flow fields permitted to identify the key mechanism involved. It was found that the shock reflections in methane permitted to drive strong jets behind periodically formed Mach shocks on the front of the deflagration. These provided local enhancement of the reactivity through mixing, conducive to new hot spot formation. The hot spot formation forward sustained the shock reflection processes conducive to further jetting action. The subsequent organization of the front in stronger fewer modes eventually lead to the transition to detonation. In the hydrogen system, for similar thermo-chemical parameters, the absence of these jets did not permit to establish such fast flames. This jetting slip line instability in shock reflections (and lack thereof in hydrogen) was correlated with the value of the isentropic exponent and its control of Mach shock jetting (Mach & Radulescu, PROCI, 2011).

Relativistic outflow from two thermonuclear shell flashes on neutron stars

We study the exceptionally short (32-41 ms) precursors of two intermediate-duration thermonuclear X-ray bursts observed with RXTE from the neutron stars in 4U 0614+09 and 2S 0918-549. They exhibit photon fluxes that surpass those at the Eddington limit later in the burst by factors of 2.6 to 3.1. We are able to explain both the short duration and the super-Eddington flux by mildly relativistic outflow velocities of 0.1$c$ to 0.3$c$ subsequent to the thermonuclear shell flashes on the neutron stars. These are the highest velocities ever measured from any thermonuclear flash. The precursor rise times are also exceptionally short: about 1 ms. This is inconsistent with predictions for nuclear flames spreading laterally as deflagrations and suggests detonations instead. This is the first time that a detonation is suggested for such a shallow ignition column depth ($y_{\rm ign}$ = 10$^{10}$ g cm$^{-2}$). The detonation would possibly require a faster nuclear reaction chain, such as bypassing the alpha-capture on $^{12}$C with the much faster $^{12}$C(p,$\gamma$)$^{13}$N($\alpha$,p)$^{16}$O process previously proposed. We confirm the possibility of a detonation, albeit only in the radial direction, through the simulation of the nuclear burning with a large nuclear network and at the appropriate ignition depth, although it remains to be seen whether the Zel'dovich criterion is met. A detonation would also provide the fast flame spreading over the surface of the neutron star to allow for the short rise times. (...) As an alternative to the detonation scenario, we speculate on the possibility that the whole neutron star surface burns almost instantly in the auto-ignition regime. This is motivated by the presence of 150 ms precursors with 30 ms rise times in some superexpansion bursts from 4U 1820-30 at low ignition column depths of ~10$^8$ g cm$^{-2}$.

Basic oxygen furnace steelmaking end-point prediction based on computer vision and general regression neural network

End-point prediction is one of the most difficult problems in basic oxygen furnace (BOF) steelmaking process. To address this problem, some researchers have proposed some methods based on flame image processing and pattern classification. Because of the dynamically changing flame and real-time needs during the blowing process, there are still some issues that need to be solved. We propose a novel method based on accurate and fast multi flame features extraction and general regression neural network (GRNN). Firstly, flame images were acquired, and then the background of each image was removed via color similarity determination algorithm; secondly, color, texture, and boundary features were extracted; the fast and robust boundary and texture features were extracted by using the proposed methods, and these features were tested for their validity to the end-point prediction via comparing them with some other similar methods; finally, the prediction model was built using multi-features and GRNN. The experimental results demonstrated that it is accurate and fast to use the proposed method to the BOF end-point predict.

Extraction induced by emulsion breaking as a tool for Ca and Mg determination in biodiesel by fast sequential flame atomic absorption spectrometry (FS-FAAS) using Co as internal standard

This paper reports, for the first time, the determination of Ca and Mg in biodiesel after their extraction from samples using the novel extraction induced by emulsion breaking. The quantification of the analytes in the extracts was performed by flame atomic absorption spectrometry operating at fast sequential mode (FS-FAAS) and using cobalt as internal standard. Several parameters that could affect the extraction efficiency of the method were studied such as the influence of surfactant (Triton X-114) and acid (HNO3) in the extractant solution. Also, operating parameters of the spectrometer (flame stoichiometry and observation height) were evaluated in order to allow sensitive measurements of Ca, Mg and Co in the same conditions. Quantitative extraction of the analytes can be obtained by using 20mL of biodiesel and 4mL of a solution containing 2.1molL−1HNO3 and 7.5% m/v Triton X-114. In these conditions, the emulsions can be broken in 10min of heating at 90°C. The limits of quantification for Ca and Mg were 0.047 and 0.013μgmL−1, respectively. Five samples of soybean-based biodiesel were analyzed and recovery tests were performed, obtaining recovery percentages in the range of 88%–106%.

Effects of direct injection timing of ethanol fuel on engine knock and lean burn in a port injection gasoline engine

Ethanol is a promising alternative fuel for internal combustion engines due to its renewable feature. To make the use of ethanol fuel more effective and efficient, ethanol direct injection plus gasoline port injection (EDI+GPI) has been investigated in recent years. By directly injecting ethanol into the engine, the advantages of ethanol fuel such as high latent heat of vaporization, fast laminar flame speed, wide flammability and better low temperature combustion stability can be well utilized to enhance engine anti-knock ability and improve lean burn performance. For an engine equipped with direct injection (DI) system, start of injection (SOI) timing is an important control parameter which directly affects the engine performance. This paper reports the investigation to the effect of ethanol fuel SOI timing on knock mitigation and lean burn. Experiments were conducted on a 250cc single cylinder spark ignition (SI) engine equipped with EDI+GPI system. Ethanol fuel SOI timing before and after the inlet valve closing, defined as early and late injection timings (EEDI and LEDI) were investigated in engine conditions at knock limited spark advance (KLSA) and lean burn limit. The experimental results showed that LEDI was effective on suppressing engine knock and permitting more advanced spark timing. EEDI was less effective than LEDI on mitigating knock due to the increased heat transfer from cylinder wall to gases. The mixture quality may be deteriorated in LEDI conditions which resulted in low engine efficiency and high emissions. Volumetric efficiency was increased and combustion duration was reduced in EEDI conditions. The combined effects of improved volumetric efficiency, reduced combustion duration and moderately advanced spark timing resulted in increased engine thermal efficiency in EEDI conditions. In lean burn, EEDI was more effective on extending lean burn limit. The maximum lambda achieved in EEDI condition was 1.29 when ethanol energy ratio (EER) was 24% and SOI timing was 290 CAD BTDC. LEDI only slightly increased lean burn limit which was just over stoichiometric air–fuel ratio (AFR). In EEDI conditions, IMEP was greater and combustion stability (COV) was better than that in LEDI conditions. The emissions in EEDI conditions were also lower than that in LEDI conditions.

Nitrogen-Doped Carbon Nanoparticles by Flame Synthesis as Anode Material for Rechargeable Lithium-Ion Batteries

Nitrogen-doped turbostratic carbon nanoparticles (NPs) are prepared using fast single-step flame synthesis by directly burning acetonitrile in air atmosphere and investigated as an anode material for lithium-ion batteries. The as-prepared N-doped carbon NPs show excellent Li-ion stoarage properties with initial discharge capacity of 596 mA h g(-1), which is 17% more than that shown by the corresponding undoped carbon NPs synthesized by identical process with acetone as carbon precursor and also much higher than that of commercial graphite anode. Further analysis shows that the charge-discharge process of N-doped carbon is highly stable and reversible not only at high current density but also over 100 cycles, retaining 71% of initial discharge capacity. Electrochemical impedance spectroscopy also shows that N-doped carbon has better conductivity for charge and ions than that of undoped carbon. The high specific capacity and very stable cyclic performance are attributed to large number of turbostratic defects and N and associated increased O content in the flame-synthesized N-doped carbon. To the best of our knowledge, this is the first report which demonstrates single-step, direct flame synthesis of N-doped turbostratic carbon NPs and their application as a potential anode material with high capacity and superior battery performance. The method is extremely simple, low cost, energy efficient, very effective, and can be easily scaled up for large scale production.

Fast determination of trace elements in organic fertilizers using a cup-horn reactor for ultrasound-assisted extraction and fast sequential flame atomic absorption spectrometry

A fast and accurate method based on ultrasound-assisted extraction in a cup-horn sonoreactor was developed to determine the total content of Cd, Co, Cr, Cu, Mn, Ni, Pb and Zn in organic fertilizers by fast sequential flame atomic absorption spectrometry (FS FAAS). Multivariate optimization was used to establish the optimal conditions for the extraction procedure. An aliquot containing approximately 120 mg of the sample was added to a 500 µL volume of an acid mixture (HNO3/HCl/HF, 5:3:3, v/v/v). After a few minutes, 500 µL of deionized water was added and eight samples were simultaneously sonicated for 10 min at 50% amplitude, allowing a sample throughput of 32 extractions per hour. The performance of the method was evaluated with a certified reference material of sewage sludge (CRM 029). The precision, expressed as the relative standard deviation, ranged from 0.58% to 5.6%. The recoveries of analytes were found to 100%, 109%, 96%, 92%, 101%, 104% and 102% for Cd, Cr, Cu, Mn, Ni, Pb and Zn, respectively. The linearity, limit of detection and limit of quantification were calculated and the values obtained were adequate for the quality control of organic fertilizers. The method was applied to the analysis of several commercial organic fertilizers and organic wastes used as fertilizers, and the results were compared with those obtained using the microwave digestion procedure. A good agreement was found between the results obtained by microwave and ultrasound procedures with recoveries ranging from 80.4% to 117%. Two organic waste samples were not in accordance with the Brazilian legislation regarding the acceptable levels of contaminants.

HCNG 엔진의 터보차저 변경에 따른 전부하 출력 및 배출가스 특성 연구

Hydrogen-natural gas blends(HCNG) engine is optimizing technology of performance and emission characteristics with use of hydrogen's fast flame speed and wide flammability limit. As lean-burn limit is extended, the improvement in thermal efficiency and harmful emissions can be achieved. However, the extension of lean-burn limit under a wide open throttle operation point could be realized with the increase in boosting capacity in a lean-burn engine with turbo-charging system. In the present study, the power output characteristics of HCNG engine with turbo-charger change is assessed and feasibility of the increase in boost- ing capacity is evaluated. The turbo-charger design with high efficiency at higher flow rate rather than higher boosting pressure makes efficient operation possible at relatively rich mixture condition.

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.