In this work, a flame propagation behavior analysis was conducted for outwardly propagating flames of diethoxymethane (DEM) and ethanol as exemplary fuels with and without low-temperature chemistry (LTC). Flame temperatures were decreased by N2-dilution of the oxidizer to enhance the possible low-temperature chemistry effect, leading to decreased flame propagation speed. Slowly expanding spherical flames are known to be affected by radiation and particularly by buoyancy, deforming the flame's shape and violating underlying existing post-processing assumptions. Experiments were conducted under microgravity conditions during parabolic flight campaigns aboard the Airbus A310 Zero-G to isolate the effects of radiation from buoyancy-induced flame deformation. Extracted unstretched laminar flame speeds at initial pressures of 3 to 10 bar were compared with planar flame simulations. The approach by Yu et al. (2014) was used to correct the experimental laminar flame speeds for radiation. For lean mixtures, significant flame acceleration was observed with a strong nonlinear propagation speed dependence on stretch for both fuels. In contrast, rich DEM/oxidizer mixtures consistently maintain self-sustained flame propagation, while ethanol/oxidizer mixtures tend to extinguish. This is linked to the effectiveness of low-temperature branching reactions in DEM oxidation. Flame structure analyses for DEM/oxidizer mixtures revealed that the reaction zone broadening is correlated to the distribution of oxidized DEM radicals predominantly present at a lower temperature of T similar to 800 K.
Solid phase burning models are necessary to understand fire dynamics of wood. Its thermal degradation produces pyrolysate and char which burn during flaming and smouldering respectively, but drying is the first thermophysical process to occur when wood, or biomass in general, burns. This study characterises the behaviour of drying via a single-step first-order Arrhenius-type reaction. A kinetic model is optimised using inverse modelling on TGA data, taken from the literature, from five different biomass types at different heating rates and moisture contents. Two clusters of biomass are identified: one of high lignin and low cellulose and hemicellulose content for which drying occurs faster at around 60 degrees C, and another of low lignin and high cellulose and hemicellulose content for which drying occurs slower at around 80 degrees C. The model is validated against further TGA data, yielding errors of 8% for the prediction of the faster drying cluster and 13% for the slower drying cluster. The resulting kinetics are used in combination with a one-dimensional meso-scale model for wood burning using Gpyro, a generalised pyrolysis model for porous fuels, to study the influence of kinetics, heat transfer and mass transfer of moisture on the burning behaviour of wood. The model is compared against experimental temperature data from the literature. The relationship between the position of the pyrolysis and drying fronts with moisture content is shown. As the moisture content increases from 6% to 18%, the pyrolysis rate decreases. The drying process acts thus as a heat sink which significantly delays pyrolysis. We hence demonstrate that there exist two clusters of biomass drying behaviour and extract generalised drying kinetics for each of them.
This study investigates thermodiffusive (TD) instabilities in lean premixed hydrogen/air flames using simultaneous measurements of hydroxyl radical (OH) via planar laser-induced fluorescence (PLIF) and temperature via Rayleigh scattering. Correlations between flame front curvature and scalar (OH and temperature) gradients as surrogates for reaction rate were assessed in the presence of TD instabilities. Experimental 2D data were compared with corresponding 2D slices extracted from 3D direct numerical simulations (DNS). In both DNS and experiments, flame fronts defined by local maximum scalar gradients (gradient-based) leads to spurious results owing to very low gradients in highly negative curvature regions associated with near-extinction by TD instability. These discontinuous fronts exhibited weaker curvature-gradient correlations than the curvature-HRR (heat release rate) correlation, indicating that scalar gradients along gradient-based fronts are inappropriate surrogates for HRR. To address this limitation, two continuous flame fronts were evaluated: (1) laminar-based front, defined using temperature progress variable at local maximum gradients of scalar by laminar flame calculation, and (2) mode-based front, defined using the most probable temperature (mode value of PDF) at local maximum gradients of temperature. These fronts capture low gradient regions and exhibit stronger correlations between curvature and a surrogate for the reaction rate in both DNS and experiment. DNS analysis revealed that flame fronts based on temperatures (from both laminar-based and mode-based methods) and OH (from laminar-based method) exhibit strong correlation with HRR, with laminar-based flame front for OH showing the highest correlation. However, in experiments, laminar-based flame front for OH correlates poorly with HRR due to spatial misalignment between the temperature and OH fields. Flame fronts by temperature from both laminar-based and mode-based methods are determined to be the most reliable HRR surrogates in experiments. This study highlights that analyzing curvature-gradient correlations under TD instability requires a continuous flame front capturing negative curvature and low gradient regions.
This paper provides the first application of picosecond phase-selective laser-induced breakdown spectroscopy (PS-LIBS) to characterise nanoparticles during spray-flame synthesis. An established flame spray pyrolysis (FSP) burner producing 8 nm maghemite nanoparticles provides a reference flame condition to compare 30 picosecond and 9 to 37 nanosecond laser excitation sources. It is demonstrated that picosecond excitation provides greater than an order of magnitude increase in the PS-LIBS emission intensity over broad spectral regions (372 - 389.5 nm). This is due to the discovered power-law dependence of the PS-LIBS emission intensity on the employed irradiance. The enhanced emission enables single-shot planar imaging of iron oxide nanoparticles synthesised during FSP. Time-resolved emissions of picosecond PS-LIBS are also characterised to determine the effect of parameters, such as particle diameter, on the emission lifetime. For picosecond excitation, the emission lifetime is within 2 to 20 ns compared to the short 30 ps excitation pulse, indicating significantly reduced laser-plasma interactions over the emission lifetime. Since the laser pulse is < 1.5 % of the emission lifetime, there is a unique potential to quantify the emission decay with respect to particle attributes compared to nanosecond excitation.
This work provides new insights into the dynamics of the initial oxide lobe that forms from the native alumina layer covering an aluminum particle. This lobe is a distinctive hallmark of aluminum particle combustion — absent in hydrocarbon droplet combustion — and is commonly considered the origin of micrometric oxide residues. While it is generally accounted for in numerical simulations, this study challenges the prevailing assumption of its persistence. Instead, it demonstrates that the initial alumina lobe progressively regresses over time. The rate of this regression is closely linked to the availability of gaseous aluminum near the droplet surface and is notably delayed in nitrogen-containing environments. Moreover, this research offers a novel perspective on the final phase of the particle’s combustion, which is initiated by the formation of a surface layer composed of alumina and/or oxynitrides. The presence of nitrogen plays a key role in triggering this covering process, which subsequently induces spinning and jetting phenomena. The study elucidates the mechanisms behind these dynamics — particularly jetting — and provides quantitative data on their effects in more complex configurations involving particle suspensions.
The effect of ammonia addition on the formation and nanostructure of soot was experimentally investigated in laminar diffusion flames of n-butane. Soot volume fraction and flame temperature were measured in-situ by line-of-sight attenuation and spectral soot emission measurements respectively. Chemical, nanostructural, and electronic information of thermophoretically collected soot particles were obtained by ex-situ Raman and scanning tunneling spectroscopy measurements. The study highlights the influence of ammonia addition on soot temperature, concentration, and yield in n-butane flames. Soot levels dropped from 4.5 ppm in neat flames to below 2 ppm with a mole fraction XNH3= 0.5, with n-butane flames showing less sensitivity to ammonia than ethylene flames investigated in earlier studies. Ammonia reduced soot yields, decreasing carbon-to-soot conversion from 12% to 8% at XNH3= 0.4, and delayed the onset of the soot formation, shifting peak yields upward. Soot temperatures (1600-1900 K) increased with ammonia due to reduced radiative heat transfer from suppressed soot concentrations. These findings confirm the effectiveness of ammonia in suppressing and delaying soot formation in combustion. Finally, ammonia addition is shown to promote the formation of more graphitic soot particles, with larger aromatic domains, and the presence of nitrogen functionalities, as observed by both the Raman and photoluminescence spectra of the soot particles and by the local density of states of the carbon matrix, measured by scanning tunnelling spectroscopy.
Ammonia (NH3) has been identified as a potential carbon-free fuel to decarbonize power generation by gas turbines. However, challenges associated with flame stabilization and emissions must be solved. Two-stage, rich-lean combustion is a promising strategy that requires fine control of the secondary stage parameters, where air is injected to oxidize the remaining unburned fuel from the rich primary stage. This study investigates the effects of the fuel blend composition (NH3-CH4 and NH3-H2), NH3vol fraction (XNH3), primary (cbprimary) and global (cbglobal) equivalence ratios, and geometry of the secondary air injection (number and diameter of holes) on the morphology of the lean secondary and rich-premixed primary flames, and primary flame stability and dynamics. Experiments are conducted with a lab-scale piloted burner inspired by the AE-T100's micro gas turbine burner. By increasing the secondary air flow rate (Qsec), cbglobal varied from 0.91 down to a value that produces primary flame morphology changes (cbglobal,FC), and then to a minimum value that eventually led to flame instability followed by blowout (cbglobal,BO). These thresholds were found to depend on fuel composition and air injection geometry. High-speed chemiluminescence imaging of NH2* combined with Dynamic Mode Decomposition (DMD) revealed distinct instability mechanisms: CH4-blended flames exhibited longitudinal pulsations, while H2-blended flames showed a rotating inner core that reignites upstream reactants. These instabilities are linked with a combustion regime transition for the secondary combustion zone, from diffusion-like to premixed-like (or partially premixed), indicative of an optimum cbglobal for each geometry. Finally, increasing the number of secondary air holes (while keeping diameter constant) extended flame stability to leaner cbglobal. Data showed that the geometry of the secondary air injection is more important for stability than other varied parameters, a valuable finding for the design of future two-stage, rich-lean NH3 burners.
Hydrogen (H2) is an attractive sustainable energy source, given its carbon-free combustion. However, its future availability in large quantities remains uncertain. To address this issue, designing fuel-flexible (methane - hydrogen) combustion chambers adaptable for various fuel blends is essential. One of the main challenges in these systems is achieving low NOx emissions and maintaining flashback-free operation for high hydrogen content fuels. The present paper investigates the nitric oxide (NO) emission characteristics of technically premixed, swirl-stabilized flames across fuel mixtures, ranging from 100% methane (CH4) to 100% H2. To mitigate the flashback risk for high hydrogen content mixtures, a non-rotating axial air injection (AAI) jet is introduced along the centerline of the swirling flow. The NO distribution within the combustion chamber was measured using NO planar laser-induced fluorescence (PLIF) and validated against NO concentrations in the exhaust gas measured with an exhaust gas analyzer. Complementary hydroxyl (OH) PLIF measurements and particle image velocimetry (PIV) measurements provided insights into the flame location and flow field characteristics and the relationship between flame location and NO concentrations. The results showed that introducing AAI significantly impacts the flame shape and the NO concentration within the combustion chamber, ultimately leading to an increase in NO emissions. For cases without AAI, increasing the hydrogen molar fraction (up to XH2 = 0.4) increases the NO emissions. However, for cases with a high level of AAI (and hydrogen contents up to XH2 = 1), no clear trend was observed between the level of hydrogen content and the measured level of NO. This suggests that NO formation results from a complex interaction between local flame temperature variations, which depend on fuel-air mixing and fuel composition, and the specific mechanisms responsible for NO production.
The effect of inlet concentration of NH3 (0, 30000, 60000, and 100000 ppm) and temperature (1275, 1375, and 1475 K) on the formation of soot, gases and polycyclic aromatic hydrocarbons (PAH) from the pyrolysis of C2H4/ NH3 mixtures with constant concentration of C2H4 (30000 ppm) was analyzed. Light gases were quantified through gas chromatography, while soot was collected in a fiber filter, at the reactor outlet, and the particle diameter was analyzed. The PAH were analyzed through gas chromatograph coupled to mass spectrometer by considering the 16 PAH reported by Environmental Protection Agency (EPA) as priority pollutants. The main results show that the higher the presence of NH3 in the NH3/C2H4 mixture, the lower the soot formation. This decrease of soot is accompanied by an increase of HCN and a decrease in C2H2, indicating that the carbon has been removed from the typical soot reaction pathways. In all conditions tested, PAH formed follow the typical trend observed in earlier works with respect to temperature, i.e. as the temperature increases, the formation of PAH decreases. At 1275 K, the presence of high NH3 levels increases PAH concentrations, likely due to suppression of soot formation routes, which prevents PAH consumption and allows aromatic intermediates to accumulate rather than proceed to particle growth. The increase in NH3 concentrations at the inlet reduces the toxicity expressed as benzo[a]pyrene equivalent concentration (B[a]P-eq), under all conditions studied. Analysis of the particle distribution diameter reveals that the NH3 presence reduces diameter variations and predominantly results in smaller soot particles.
Direct numerical simulations (DNS) of laboratory-scale turbulent premixed Bunsen flames have been conducted and compared with an experimental dataset encompassing three methane-hydrogen flames with volumetric hydrogen fractions of 0%, 40% and 70%. The Bunsen flames are representative of the strict flamelet regime of combustion with low turbulence intensity, which allows for the analysis of the effects induced by the combined action of Darrieus-Landau instability and non-unity Lewis number effects. The chemical aspect of DNS is treated using irreversible one-step Arrhenius chemistry, where the effective Lewis number has been determined based on a suitable calculation of the fuel mixture Lewis number along with an appropriate blending with the oxidiser Lewis number. A comparison between 2D experimental flame imaging and 3D DNS results reveals good agreement regarding mean flame shape, flame morphology, turbulent flame wrinkling, and turbulent burning velocity. This shows that simple chemistry simulations with appropriately chosen effective Lewis numbers can accurately describe turbulent burning velocity and flame-turbulence interaction for methane-hydrogen fuel blends. Provided emissions and ignition are not within the focus of the study, this might be very useful for conducting large-scale turbulent combustion simulations or large parametric studies of carbon-free or carbon-reduced fuels, which are enablers for the transformation of the energy industry into a climate-neutral circular economy.
In anticipation of future multi-dimensional turbulent combustion calculations of dual-fuel systems with the Doubly Conditional Moment Closure (DCMC), a preliminary validation of the DCMC-calculated structures of such flames in mixture fraction-progress variable space has been performed by comparison with simulations of laminar non-premixed counterflow flames of gaseous heptane against methane/air mixtures as the oxidizer under various conditions. Two methane/air equivalence ratios were considered, one below and one slightly above the lean flammability limit of the methane/air premixed mixture, the former selected as a particularly challenging case where the interaction between the two fuels is expected to be strong since the methane-air flame cannot sustain on its own. The effect of strain was also investigated. The accuracy of the usual assumption of unity Lewis number employed in the DCMC model was examined using the laminar flame simulations and it was concluded that some discrepancies exist in intermediate species and temperature profiles compared to the case when differential diffusion was taken into account. These differences were enhanced at high strain. The heat release rate profile implies the existence of two separate reaction zones, which merge at high strain rates. The DCMC model predicted quite accurately the methane/air side of the flame in mixture fraction space, but not to the same degree of accuracy the heptane side of the flame. The results provide insights into the structure of dual-fuel flames and suggest that such systems can be successfully modelled by mixture-fraction based methods.
The sorption-enhanced methanation (SEM) process, involving in situ steam removal, has been recently proposed in a chemical looping configuration based on dual interconnected fluidized beds. By exploiting this configuration, the methanation and the sorbent hydration reactions occur in one fluidized bed, while the regeneration of the exhausted sorbent takes place in the other one. In this work, SEM of CO2 (with a stoichiometric feed ratio H2/ CO2 = 4) was studied in a batch lab-scale apparatus to test the performance of two commercial zeolites, 3A and 4A, as water sorbents. A traditional Ni/Al2O3 catalyst for methanation was synthesized and used during the experimental campaign. The selected sorbents were tested in the temperature range 200-250 degrees C, showing similar H2O capture capacities. The choice of such a narrow range of investigated temperatures is due to the physical and chemical constraints imposed by water adsorption and methanation kinetics: the Ni-based catalyst is only active above 200 degrees C, whereas the sorbent H2O capture capacity strongly decreases with increasing temperature, and above 250 degrees C a significant worsening occurs. Despite the temperature issues, a clear enhancement of methane productivity compared to traditional methanation was observed under SEM conditions, for both sorbents. As expected, this effect decreased with increasing temperatures, due to the worsening of the physical H2O adsorption performance. Both zeolites showed a quite stable and repeatable behavior with an average enhancement ranging from 22 % to 63 % and 27 % to 66 % for zeolites 3A and 4A, respectively.
Hydrogen is a promising fuel to achieve decarbonisation targets. Its broad flammability range compared to hydrocarbons allows fuel-leaner operations, which can help to mitigate thermal nitrogen oxides formation. The lean direct injection combustor (LDIC) configuration with multiple injection of fuel into crossflowing air, is one of the potential frontrunners for 100% hydrogen combustors. While the LDIC can reduce flashback risk, the global flame features are not well understood. Large Eddy Simulation (LES) with a subgrid-scale partially premixed combustion model with tabulated chemistry is employed to study the fundamental mechanisms in lifted or attached flames. The computed mixture fraction statistics in non-reacting flow and flame attributes compare quite well with measurements. The time-averaged statistics are investigated to shed light into the effects of injector count and momentum flux ratio on fuel-air mixing and global flame behaviour. It is demonstrated that the major parameter influencing the fuel distribution and flame behaviour is the jet-tocrossflow momentum flux ratio based on the individual fuel jet. The number of injectors (that are placed diametrically opposite to each other) has minimal influence on the mixture distribution of individual fuel jets but influences the flame length.
The modeling of hydrogen-air flames has become increasingly important due to ongoing efforts for the decarbonization of propulsion and power applications. Significant research has focused on developing accurate and efficient combustion models to simulate the propagation of lean hydrogen-air premixed flames. In this context, this study investigates key modeling aspects of thermodiffusively unstable turbulent premixed flames, leveraging an experimental dataset comprising Bunsen flames under various turbulence intensities and thermochemical conditions. The stability limits of the hydrogen-enriched-methane/air mixtures are assessed using highly resolved numerical simulations with detailed thermochemistry and transport. The thermodiffusive instability of the mixtures is quantified by comparing the numerical dispersion relation with the theoretical limit of the Darrieus-Landau linear growth rate. The ensuing results are consistent with the values of a thermodiffusive instability parameter E based on theoretical models, which is therefore used throughout the analysis on modeling aspects. For thermodiffusively unstable mixtures, higher flame surface density and enhanced wrinkling are observed under sufficiently high turbulence intensities. The experimental measurements on the flame front position are spatially filtered to evaluated, similarly to flame resolved numerical simulation, the wrinkling factor. A wrinkling factor model, based on the volume averaged filtered flame surface density is proposed to represent the interactions between turbulence and intrinsic flame instabilities. This empirical model incorporates a measure of the overall turbulence intensity experienced by the flame, a thermodiffusive instability parameter based on theoretical considerations, and the LES grid size.
This study investigates the influence of differential and preferential diffusion on turbulent burning velocity and flame surface area in premixed NH3/H2/N2-air flames using a three-dimensional Direct Numerical Simulation (DNS) database of statistically planar flames. The analysis covers three fuel blends-pure NH3-air, 60 %NH3/25 %H2/15 %N2-air, and 40 %NH3/45 %H2/15 %N2-air-all at an equivalence ratio of 0.81, falling within the thin reaction zones regime on the Borghi-Peters diagram. Both the normalised flame surface area (by the projected area in the mean propagation direction) and the turbulent-to-laminar burning velocity ratio are generally comparable (but not equal) across the different fuel blends. However, the normalised flame surface area is found to be smaller than the corresponding normalised turbulent burning velocity, especially in cases involving nonunity Lewis numbers. This discrepancy highlights the impact of differential and preferential diffusion. Among the mixtures studied, the NH3-air flame exhibits the highest effective Lewis number and correspondingly the lowest values of normalised flame surface area and turbulent burning velocity ratio. Despite these variations, the similarity between the normalised surface area and burning velocity across cases lends partial support to Damkohler's first hypothesis. Moreover, the second Damkohler hypothesis reasonably estimates the order of magnitude for the normalised turbulent burning velocity. Overall, the findings suggest that conventional turbulent combustion models relying upon Damkohler's hypotheses may remain valid for hydrogen-enriched ammonia-air flames under the studied conditions, despite the presence of differential and preferential diffusion effects.
The response of laminar lean H2-air premixed flames to acoustic oscillations has been investigated by performing simulations of 2D laminar flames at cent = 0.4 and cent = 0.7 for a range of oscillation frequencies, between 50 kHz and 500 kHz, at sound pressure levels of 120 dB and 130 dB. The flames are subjected to pressure waves by implementing a monopole-type acoustic source at the inflow, which acts to perturb the flame surface and leads to wrinkled flames for all of the conditions investigated. For all cases, a range of wave numbers appearing on the flame surface exhibit exponential growth at early times, consistent with the linear regime of flame evolution. This allowed for dispersion relations to be computed such that the flame response across acoustic conditions and equivalence ratios can be compared, and the interplay between the flame-acoustic interaction and hydrodynamic/thermodiffusive mechanisms can be investigated. Differences are noted in terms of the wave number at which the dispersion relation terminates, with the dispersion relation for the lower frequency cases (50 kHz for cent = 0.4 and 100 kHz for cent = 0.7) terminating at lower wave number values, indicating an immediate transition to a non-linear growth phase for high wave numbers. Higher growth rates for small wave number modes have been observed when compared with those for non-acoustically forced flames in the literature, suggesting enhancement of the thermodiffusive mechanism at these wave numbers. It is argued that the differential diffusion effects are influenced by the strong pressure gradients seen for the acoustically forced cases.
The growing need to transition from carbon-based fuels to cleaner alternatives, driven by climate change, necessitates advanced combustion technologies capable of fuel flexibility and low pollutant emissions. MILD combustion is a promising technology that meets these requirements. This work presents the methodology and process for scaling down a reactor with a cyclonic flow field operating under MILD combustion. In such reactors, fluid dynamics play a crucial role due to the recirculation of flue gases, which dilute the air-fuel mixture and preheat fresh reactants. Maintaining similar fluid dynamic behavior during scaling is essential. The scaling process began with designing the new burner. Various configurations were tested, modifying the inlet diameter, nozzle-to-wall distance, and chamber height to preserve residence time. Non-reactive CFD simulations were conducted to analyze velocity fields and select the optimal design. Velocity profiles were examined along the jet injection axis and at the centerline parallel and perpendicular to the jet injection. The selected design maintained the original geometric aspect ratio while varying the nozzle-to-wall distance to avoid wall-jet effects. The scaled reactor, constructed from vermiculite, was experimentally tested under the same conditions as the original. Temperature measurements at three positions showed similar trends and values between the two reactors. Species production also exhibited comparable trends, with slight differences in O2, CO2, and CO concentrations attributed to reduced residence time in the scaled version. NO emissions remained low but showed distinct behaviors due to differences in feeding configurations and formation mechanisms. The comparable temperature distributions and species concentration behaviors between the original and scaled reactors validate the proposed scaling methodology as a reliable approach for adapting cyclonic flow field reactors for MILD combustion.
Hydrogen is attracting increasing attention as a clean-burning, carbon-neutral fuel in heavy-duty engine applications, which usually often operate by compression ignition. However, hydrogen cannot ignite on its own under typical conditions in these engines. To address this, the concept of hydrogen-diesel dual direct injection (H2DDI) has been recently proposed in which hydrogen and diesel are introduced separately into the combustion chamber, with diesel providing the ignition source, leading to ignition occurring in stratified mixtures. This study employs numerical simulations to investigate the ignition of stratified mixtures in one-dimensional laminar mixing-layer configurations under engine-relevant conditions, where a n-dodecane fuel-rich stream is surrounded by an oxidiser stream with (dual-fuel/DF cases) and without (single fuel/SF cases) hydrogen. It is shown that the ignition of the DF cases is delayed compared to the SF cases because hydrogen consumes OH species during the low-temperature oxidation of n-dodecane. Unlike the SF cases, the DF cases exhibit a postponed initiation of the first-stage ignition relative to the most reactive ignition delay time observed in homogeneous reactor simulations. In both the SF and DF cases, diffusion-supported cool flames promote radical accumulation, accelerating the transition to the second-stage ignition. Furthermore, the study reveals that preferential diffusion effects in both the SF and DF cases, influenced by mixing conditions, significantly impact the ignition process, with the diffusion behaviours of different species playing distinct roles in the ignition of both stages. In particular, the preferential diffusion of n-dodecane is the dominant factor of overall preferential diffusion effects, delaying both the first and second-stage ignition, whereas the preferential diffusion of hydrogen only increases the maximum flame temperature.
One of the most challenging problems in the development of hydrogen gas turbine engines is predicting and mitigating self-excited combustion instabilities. In this experimental study, we employ a fuel staging method, which is implemented by splitting the total hydrogen mass flowrate into two separate fuel circuits for the transition of combustion modes between fully-premixed (FP) and technically-premixed (TP) conditions, to suppress the thermoacoustic oscillations, fundamentally through interference mechanisms of convective disturbances. We show that while the H2 flame structure and length remain unchanged, the variation in the fuel staging ratio has a significant impact on the self-excited instabilities. For the FP case, the system exhibits high-amplitude limit cycle oscillations. Remarkably, the introduction of even a small degree of fuel staging significantly attenuates the intensity of the thermoacoustic instability. To assess the system stability, here we experimentally investigate the acoustic scattering characteristics and their net power balance with respect to the staging ratio, in conjunction with physics-based analytical models. Our results demonstrate that the fuel staging ratio significantly affects the scattering properties of the burner and flame, due to the contribution of equivalence ratio fluctuations. This leads to a marked attenuation in the power amplification potential near the instability frequency, whereas the opposite effect coexists at the off-target frequency. The physics-based models derived from the analytical burner and flame scattering matrices are validated against the measurement data across all fuel staging ratios. This approach, which does not require knowledge of all the elements of the thermoacoustic system for stability predictions (e.g. the reflection coefficients of the combustor inlet and outlet), enables a preliminary assessment of the system’s thermoacoustic stability based solely on the scattering properties of the burner and its flame.
Two-dimensional (2D) Rayleigh scattering measurements are employed to investigate flame surface topology and temperature in lean premixed hydrogen/air turbulent jet flames as the Karlovitz number increases toward the distributed combustion regime. Quantitative temperature fields obtained from 2D Rayleigh thermometry are validated against one-dimensional (1D) temperature profiles measured via quasi-simultaneous 1D Raman/Rayleigh spectroscopy. The 2D temperature data are used to extract flame front locations and to establish direct correlations between local flame surface curvature and temperature characteristics through joint probability density functions (JPDFs). Results highlight the thermo-diffusive nature of the flame, with higher local temperatures frequently observed in regions of positive curvature. The JPDFs further reveal that non-uniform and asymmetric temperature distributions on the flame surface result from the combined effects of turbulence and differential diffusion. As the Karlovitz number increases, these asymmetries diminish, and temperature distributions shift, signaling a transition toward the distributed combustion regime. Based on these trends, potential temperature-based markers for thermo-diffusive instabilities are proposed, offering new insights into turbulence-chemistry interactions and their role in the onset of distributed burning in lean premixed hydrogen/air flames.
