Methane/air premixed and partially premixed turbulent flames at high Reynolds number are characterized using Direct Numerical Simulations (DNS) with detailed chemistry in a spatially evolving slot Bunsen configuration. Two sets of simulations are performed. A first set of simulations with fully premixed inlet conditions is considered in order to assess the effect of turbulence on the flame. Four simulations are performed at increasing Reynolds number and up to 22400, defined based on the bulk velocity, slot width, and the reactants' properties, and 22 billion grid points, making it one of the largest simulations in turbulent combustion. The simulations feature finite rate chemistry with a 16 species mechanism. To perform these simulations, few preliminary steps were required: (i) two skeletal mechanisms were developed reducing GRI-3.0; (ii) a convergence study is performed to select the proper spatial and temporal discretization and (iii) simulations of fully developed turbulent channel flows are preformed to generate the inlet conditions of the jet. The study covers different aspects of flame-turbulence interaction. It is found that the thickness of the reaction zone is similar to that of a laminar flame, while the preheat zone has a lower mean temperature gradient, indicating flame thickening. The characteristic length scales of turbulence are investigated and the effect of the Reynolds number on these quantities is assessed. The tangential rate of strain is responsible for the production of flame surface in the mean and surface destruction is due to the curvature term. A second set of simulations with inhomogeneous inlet conditions is performed to study how partial premixing and turbulence interact with the flame and with each other. The jet Reynolds number is 5600, and a 33 species mechanism is used. The effect of the inlet fluctuations is reflected on heat release rate fluctuations, however the conditional mean is not affected. The flames show thickening of the preheat zone, and for the lowest level of mixing a slight thickening of the reaction zone is observed. The effect of partially mixed mixture on the NOx formation is analyzed and no major impact was found.

Moderate or Intense Low-oxygen Dilution (MILD) combustion is a combustion technology that can simultaneously improve the energy efficiency and reduce the pollutant emissions of combustion devices. It is characterised by highly preheated reactants and a small temperature rise during combustion due to the large dilution of the reactant mixture with products of combustion. These conditions are generally achieved using exhaust gas recirculation. However, the physical understanding of MILD combustion remains limited which prevents its more widely spread use. In this thesis, Direct Numerical Simulation (DNS) is used to study turbulence, premixed flames and MILD combustion to obtain these additional physical insights. In a first stage, the scale-locality of the energy cascade is analysed by applying a multiscale analysis methodology, called the bandpass filter method, on DNS of homogeneous isotropic turbulence. Evidence supporting this scale-locality were obtained and the results were found to be similar for Reynolds numbers ranging from 37 to 1131. Using the same method in turbulent premixed flames, the scale-locality of the energy cascade was still observed despite the presence of intense reactions. In addition, it was found that eddies of scales larger than the laminar flame thickness were imparting the most strain on the flame. In a second part, a methodology was developed to conduct the DNS of MILD combustion with mixture fraction variations. This methodology included the effect of mixing of exhaust gases with fuel and oxidiser in unburnt, burnt and reacting states. In addition, a specific chemical mechanism that includes the chemistry of ${\rm OH^*}$ was developed. From these DNS, the role of radicals on the inception of MILD combustion was studied. In particular, due to the reactions initiated by these radicals, the initial temperature rise in MILD combustion was occurring concurrently with an increase in the scalar dissipation rate of mixture fraction which is contrasting to conventional combustion. The reaction zones in MILD combustion were also analysed and extremely convoluted reaction zones were observed with frequent interactions among them. These interactions yielded the appearance of volumetrically distributed reactions. Furthermore, the adequacy of some species to identify these reaction zones was assessed and ${\rm OH}$ showed a poor correlation with regions of heat release. On the other hand, ${\rm OH^*}$, ${\rm HCO}$ or ${\rm OH} \times {\rm CH_2O}$ were found to be well correlated. Through the study of the flame index, the existence of non-premixed and premixed modes of combustion were also highlighted. The premixed mode was observed to be dominant but the contribution of the non-premixed mode to the total heat release was non negligible. Because of the presence of radicals and high reactant temperatures, auto-igniting regions and propagating reaction zones are both observed locally. The balance between these phenomena was investigated and it was found that this was strongly influenced by the typical lengthscale of the mixture fraction field, with a smaller lengthscale favouring sequential autoignition. Finally, using the bandpass filtering method, the effect of heat release rate in MILD combustion on the energy cascade was studied and this showed that the energy cascade was not unduly affected.

State-of-the-art reacting-flow computations have to compromise either on the detail of chemical reactions or on the dimensionality of the solution, while experiments in flames are limited by the flow accessibility and provide at best a limited number of observables. In the present work, the partially premixed laminar flame structure is examined using a detailed-chemistry, one-dimensional simulation. The computational results are compared to unpublished single-point multiscalar measurements obtained at Sandia National Labs in 2001. The study is focused on axisymmetric laminar partially-premixed methane/air flames with varying premixture strength values of 1.8, 2.2, and 3.17. The combination of computational and experimental results is used to analyze the spatial and scalar flame structure under the overarching concept of flamelets. The computations are based on the Cantera open-source software package developed at CalTech by D. Goodwin, and incorporating the GRI 3.0 chemical kinetic mechanism utilizing 325 chemical reactions and 53 species for methane combustion. Cross-transport effects as well as an optically-thin radiation model are included in the calculations. Radiation changes the flame profiles due to its effect on temperature, and the attendant effects on a number of species. Using the detailed analysis of different reaction rates, the adiabatic and radiative nitric oxide concentrations are compared. The cross-transport effects, i.e. Soret and Dufour, were studied in detail. The Soret term has a small but important effect on the flame structure through a reduction of the hydrogen mass fraction, which changes the conserved scalar values. Based on the flamelet approach and a unique formulation of the conserved scalar, the flame thermochemistry can be analyzed and understood. A number of interesting effects on the flame thermochemistry can be discerned in both experiments and computations when the premixture strength is varied. An increase in premixing results in a counterintuitive decrease in intermediate species such as carbon monoxide and hydrogen, as well as an expected increase in nitric oxide concentrations. Good agreement is found between experiments and calculations in scalar space, while the difference in dimensionality between axisymmetric measurements and opposed jet computations makes comparison in physical space tentative.

The propagation of premixed flames in turbulent flows is a problem of wide physical and technological interest, with a significant literature on their propagation speed and front topology. While certain scalings and parametric dependencies are well understood, a variety of problems remain. One major challenge, and focus of this thesis, is to model the influence of fuel/oxidizer composition on turbulent burning rates. Classical explanations for augmentation of turbulent burning rates by turbulent velocity fluctuations rely on global arguments - i.e., the turbulent burning velocity increase is directly proportional to the increase in flame surface area and mean local burning rate along the flame. However, the development of such global approaches is complicated by the abundance of phenomena influencing the propagation of turbulent premixed flames. Emphasizing key governing processes and cutting-off interesting but marginal phenomena appears to be necessary to make further progress in understanding the subject. An alternative approach to understand turbulent augmentation of burning rates is based upon so-called "leading points", which are intrinsically local properties of the turbulent flame. Leading points concepts suggest that the key physical mechanism controlling turbulent burning velocities of premixed flames is the velocity of the points on the flame that propagate farthest out into the reactants. It is postulated that modifications in the overall turbulent combustion speed depend solely on modifications of the burning rate at the leading points since an increase (decrease) in the average propagation speed of these points causes more (less) flame area to be produced behind them. In this framework, modeling of turbulent burning rates can be thought as consisting of two sub-problems: the modeling of (1) burning rates at the leading points and of (2) the dynamics/statistics of the leading points in the turbulent flame. The main objective of this thesis is to critically address both aspects, providing validation and development of the physical description put forward by leading point concepts. To address the first sub-problem, a comparison between numerical simulations of one-dimensional laminar flames in different geometrical configurations and statistics from a database of direct numerical simulations (DNS) is detailed. In this thesis, it is shown that the leading portions of the turbulent flame front display a structure that on average can be reproduced reasonably well by results obtained from model geometries with the same curvature. However, the comparison between model laminar flame computations and highly curved flamelets is complicated by the presence of negative (i.e., compressive) strain rates, due to gas expansion. For the highest turbulent intensity investigated, local consumption speeds, curvatures, strain rates and flame thicknesses approach the maximum values obtained by the laminar model geometries, while other cases display substantially lower values. To address the second sub-problem, the dynamics of flame propagation in simplified flow geometries is studied theoretically. Utilizing results for Hamilton-Jacobi equations from the Aubry-Mather theory, it is shown how the overall flame front progation under certain conditions is controlled only by discrete points on the flame. Based on these results, definitions of leading points are proposed and their dynamics is studied. These results validate some basic ideas from leading points arguments, but also modify them appreciably. For the simple case of a front propagating in a one-dimensional shear flow, these results clearly show that the front displacement speed is controlled by velocity field characteristics at discrete points on the flame only when the amplitude of the shear flow is sufficiently large and does not vary too rapidly in time. However, these points do not generally lie on the farthest forward point of the front. On the contrary, for sufficiently weak or unsteady flow perturbations, the front displacement speed is not controlled by discrete points, but rather by the entire spatial distribution of the velocity field. For these conditions, the leading points do not have any dynamical significance in controlling the front displacement speed. Finally, these results clearly show that the effects of flame curvature sensitivity in modifying the front displacement speed can be successfully interpreted in term of leading point concepts.

In the present thesis, the sensitivity of flame propagation to the turbulent motion of burning gases is investigated. The long-standing issue of the 'bending effect' is focused upon, which refers to the experimentally-observed inhibition of flame propagation velocity at high intensities of turbulence. Plausible mechanisms for the bending effect are investigated by isolating systematically the effects of turbulence intensity. By providing a novel perspective on this topic, the thesis addresses the fundamental limits of turbulent burning. The investigation employs Direct Numerical Simulation (DNS), which enables the basic conditions of burning to be controlled directly. A parametric DNS dataset is designed and generated by increasing turbulence intensity over five separate simulations. Effects of turbulent motion are isolated in this manner, such that the bending effect is reproduced in the variation of flame propagation velocity recorded. Subsequently, the validity of Damköhler's hypotheses is investigated to ascertain the mechanism of bending. Analysis of the DNS dataset highlights the significance of kinematic flame response in determining turbulent flame propagation. Damköhler's first hypothesis is found to be valid throughout the dataset, suggesting that the bending effect may be a consequence of self-regulation of the flame surface. This contradicts the dominant belief that bending occurs as a result of flame surface disruption by the action of turbulence. Damköhler's second hypothesis is found to be valid in a relatively limited regime within the dataset, its validity governed by flame-induced effects on the prescribed turbulent flow field. Therefore, this thesis presents turbulent flame propagation and the bending effect as emergent from the dynamics of a flame surface that retains its internal thermo-chemical structure. Finally, experimental validation is sought for the proposed mechanisms of bending. Comparisons have been initiated with measurements in the Leeds explosion vessel, based on which the widely accepted mechanism of bending was hypothesized twenty-five years ago. Modifications to the DNS framework warranted by this comparison have aided the development of novel computationally-efficient algorithms. The ongoing work may yield insights into the key mechanism of the bending effect in turbulent flame propagation.

Halon 1301 (CF3Br), or bromotrifluoromethane, had been widely used as fire-extinguishing agent for many years before its production and consumption were severely regulated by the Montreal Protocol due to its hazardous depletion effect to the stratospheric ozone layer. It is therefore imperative to find an effective replacement fire-fighting agent before the mandated deadline of the complete phase out of CF3Br. Currently there are intensive efforts in searching for an environmentally acceptable fire suppression replacement. This, however, is hampered by a lack of fundamental understanding of how CF3Br suppresses the chemical reactions in a flame environment so effectively. Recent experimental evidence has shown that the addition of CF3Br significantly reduced the burning velocity of premixed H2/Air flames by depleting the important radical species that are important to sustain chemical reactions. Extending this finding to understand the suppression of more complicated diffusion flames and unsteady three dimension turbulent flames in the presence of Halon 1301, however, still faces enormous challenge because of the prohibitive requirement of the computational power. The present chemical reaction mechanism for even the simplest hydrocarbon fuel (CH4) combustion involves more than 300 elemental reactions and the addition of CF3Br adds approximately 70 more elemental reactions. This large number of reactions and the associated large number of reaction species, many of which still involve uncertain reaction coefficients and thermodynamics properties, present significant computing challenges for applications in multidimensional non-premixed flames that are often encountered in practice. Therefore, it is of interest to systematically reduce the full chemical mechanism to a few global reactions while still maintaining the accuracy of the original mechanism. The present research systematically reduced the complex H2/Air/CF3Br chemical reaction mechanism with 94 initial elemental reactions to 5 global reaction steps. The reduced mechanism results in dramatic savings in computer time and is capable of predicting the major species and important steady state species with high accuracy. Through detailed sensitivity and production rate analysis the present research was able to find the key elemental reactions that are responsible for the fire suppression behavior of CF3Br. Predicted maximum concentrations of H and OH were found to correlate closely with the existing laminar burning velocity data measured for the premixed H2/Air/CF3Br flames. Better agreement with the experimental data was found when two activation energies for the two most important elementary reactions from QRRK calculations were adopted. The reduced mechanism developed through this research can be used to assist in the calculation and the understanding of fire suppression of CF3Br for more practical multidimensional nonpremixed laminar and turbulent flames, and the effort in searching for other effective fire suppressing agents.

This thesis describes an experimental investigation of the response of lean turbulent swirling flames with different degrees of premixedness (i.e. different mixture patterns) to acoustic forcing using the same burner configuration and varying only the fuel injection strategy. Special emphasis was placed on the amplitude dependence of their response. Also, the behaviour of self-excited fully premixed flames was examined. kHz OH* chemiluminescence was used to study qualitatively the heat release response of the flames, while kHz OH Planar Laser Induced Fluorescence (PLIF) was employed to understand the response of the flame structure and the behaviour of the various parts of the flame. The Proper Orthogonal Decomposition (POD) method was used to extract the dominant structures of the flame and their periodicity. In the first part of the thesis, self-excited oscillations were induced by extending the length of the duct downstream of the bluff body. It was found that the longer the duct length and the higher the equivalence ratio, the stronger the self-excited oscillations were, with the effect of duct length being much stronger. The dominant frequencies of the system were found to increase with equivalence ratio and bulk velocity and decrease with duct length. For some conditions, three simultaneous periodic motions were observed, where the third motion oscillated at a frequency equal to the difference of the other two frequencies. A novel application of the POD method was proposed to estimate the convection velocity from the most dominant reaction zone structures detected by OH* chemiluminescence imaging. For a range of conditions, the convection velocity was found to be in the range of 1.4-1.7 bulk flow velocities at the inlet of the combustor. In the second part, the response of fully premixed, non-premixed with radial fuel injection (NPR) and axial fuel injection (NPA) flames was investigated and compared. All systems exhibited a nonlinear response to acoustic forcing. The highest response was observed by the NPR flame, followed by the fully premixed and the non-premixed with axial fuel injection flame. The proximity of forced flames to blow-off was found to be critical in their heat release response, as close to blow-off the flame response was significantly lower than that farther from blow-off. In the NPR and NPA systems, it was shown that the acoustic forcing reduced the stability of the flame and the stability decreased with the increase in forcing amplitude. In the fully premixed system, the flame area modulations constituted an important mechanism of the system, while in the NPR system both flame area and equivalence ratio modulations were important mechanisms of the heat release modulations. The quantification of the local response of the various parts of the flame at the forcing frequency showed that the ratio RL (OH fluctuation at 160 Hz to the total variance of OH) was greater in the inner shear layer region than in the other parts in the case of NPR and NPA flames. In fully premixed flames, greater RL values were observed in large regions on the downstream side of the flame than those in the ISL region close to the bluff body. The ratio of the convection velocity to the bulk velocity was estimated to be 0.54 for the NPR flame, while it was found to be unity for the respective fully premixed flame. In the last part of the thesis, the response of ethanol spray flames to acoustic oscillations was investigated. The nonlinear response was very low, which was reduced closer to blow-off. The ratio RL was the highest in the spray outer cone region, downstream of the annular air passage, while RL values were very low in the inner cone region, downstream of the bluff body. Unlike NPR and fully premixed flames, in case of spray and NPA systems, it was found that forcing did not affect greatly the flame structure. The understanding of the nonlinear response of flames with different degrees of premixedness in a configuration relevant to industrial systems contributes to the development of reliable flame response models and lean-burn devices, because the degree of premixedness affects greatly the flame response. Also, the understanding of the behaviour of forced spray flames is of great interest for industrial applications, contributing to the development of thermoacoustic models for liquid fuelled combustors. Finally, the estimation of the convection velocity is of importance in the modelling of self-excited flames and flame response models, since the convection velocity affects the flame response significantly.

L'influence de l'ajout d'hydrogène sur les flammes de premelange pauvre methane-air est simulée dans cette étude. Le modèle de la chambre a haute pression Orleans - ICARE (France), a été développé. Les propriétés du front de flamme sont examinées par deux modèles de combustion turbulente prémélangée, à savoir Zimont et Flamme Cohérente Model (CFM) modèles.Toutes les études de modélisation sont effectués avec le logiciel Fluent et les résultats sont comparés aux expériences. En suite, l'influence de la pression sur les statistiques de la front de flamme prémélangée a été examinée. Les simulations montrent que l'augmentation du ratio d'équivalence a diminué la hauteur des flammes et l'épaisseur de la flamme du méthane/air flames. D'autre part, l’ajout d’hydrogene de mélange pauvre méthane-air a modifié les propriétés de la flamme prémélangée. Lorsque le pourcentage volumique de l'hydrogène dans le mélange est augmenté, la position en hauteur de la flamme est réduite et l'épaisseur de la flamme devient plus mince. En outre, il a été observé que les propriétés de la flamme prémélangée ont été modifiées avec l'opération à des conditions de pression plus élevée. / Hydrogenated premixed methane/air flames under lean conditions are simulated in this study. The model of the high pressure chamber setup of Orleans - ICARE (France) has been developed. The flame front properties are investigated by two turbulent premixed combustion models, Zimont and Coherent Flame Model (CFM) models. All modeling studies are performed with Fluent software and compared to experiments. The influence of the pressure on the premixed flame front statistics has been examined as well. The simulations show that increasing the equivalence ratio decreases the flame tip height and the flame brush thickness for methane/air flames. In addition, enriching the methane-air mixture with hydrogen modifies the premixed flame front properties. When the volumetric percentage of hydrogen in the mixture is increased, the flame-end position is reduced and flame brush thickness becomes thinner. It is also observed that the premixed flame properties have been modified with operation at higher pressure conditions.

<p>Numerical simulations have been done with the CHEMKINsoftware to study different aspects of wall effects in thecombustion of lean, laminar and premixed flames in anaxisymmetric boundary-layer flow.</p><p>The importance of the chemical wall effects compared to thethermal wall effects caused by the development of the thermaland velocity boundary layer has been investigated in thereaction zone by using different wall boundary conditions, walltemperatures and fuel/air ratios. Surface mechanisms include acatalytic surface (Platinum), a surface that promotesrecombination of active intermediates and a completely inertwall with no species and reactions as the simplest possibleboundary condition.</p><p>When hydrogen is the model fuel, the analysis of the resultsshow that for atmospheric pressure and a wall temperature of600 K, the surface chemistry gives significant wall effects atthe richer combustion case (f=0.5), while the thermal andvelocity boundary layer gives rather small effects. For theleaner combustion case (f=0.1) the thermal and velocityboundary layer gives more significant wall effects, whilesurface chemistry gives less significant wall effects comparedto the other case.</p><p>For methane as model fuel, the thermal and velocity boundarylayer gives significant wall effects at the lower walltemperature (600 K), while surface chemistry gives rather smalleffects. The wall can then be modelled as chemically inert forthe lean mixtures used (f=0.2 and 0.4). For the higher walltemperature (1200 K) the surface chemistry gives significantwall effects.</p><p>For both model fuels, the catalytic wall unexpectedlyretards homogeneous combustion of the fuel more than the wallthat acts like a sink for active intermediates. This is due toproduct inhibition by catalytic combustion. For hydrogen thisoccurs at atmospheric pressure, but for methane only at thehigher wall temperature (1200 K) and the higher pressure (10atm).</p><p>As expected, the overall wall effects (i.e. a lowerconversion) were more pronounced for the leaner fuel-air ratiosand at the lower wall temperatures.</p><p>To estimate a possible discrepancy in flame position as aresult of neglecting the axial diffusion in the boundary layerassumption, calculations have been performed with PREMIX, alsoa part of the CHEMKIN software. With PREMIX, where axialdiffusion is considered, steady, laminar, one-dimensionalpremixed flames can be modelled. Results obtained with the sameinitial conditions as in the boundary layer calculations showthat for the richer mixtures at atmospheric pressure the axialdiffusion generally has a strong impact on the flame position,but in the other cases the axial diffusion may beneglected.</p><p><strong>Keywords:</strong>wall effects, laminar premixed flames,platinum surfaces, boundary layer flow</p> / QC 20100504

Ce manuscrit constitue la synthèse d'efforts visant à évaluer l'impact des composés oxygénés contenus dans des mélanges représentatifs de Biodiesel, sur leur tendance à la production de suie d'une part, et sur les propriétés physico-chimiques et la réactivité des suies d'autre part. Pour ce faire, une production stationnaire de particules de suie par un brûleur académique générant des flammes non-prémélangées a été mise en point. Dans un premier temps, nous avons montré que la teneur ainsi que la structure des additifs oxygénés à base d'ester méthylique affectent la formation de suie dans la flamme. Ainsi, les résultats ont démontré que l'efficacité d'un carburant à réduire la tendance à la production de suie en terme d'indice YSI est de plus en plus importante lorsque le contenu et/ou la longueur de la chaîne aliphatique carbonée de l'ester méthylique augmentent dans le carburant de référence. Dans un second temps, les suies " modèles " récupérées dans la région post-flamme ont été caractérisées puis comparées entre elles ainsi qu'avec une suie Biodiesel " réelle ". Les analyses ont montré que les suies issues des " surrogates " Biodiesel contenant la teneur en ester la plus élevée et la chaîne aliphatique carbonée la plus longue présentaient la distribution de taille la plus étroite, le diamètre des agrégats le plus petit, les teneurs en oxygène et en fraction organique soluble les plus faibles et étaient moins réactives. Enfin, nous avons observé que les suies " modèles " issue du brûleur académique de SANTORO et la suie " réelle " possèdent des propriétés physico-chimiques très proches et une réactivité similaire. / The aim of this work was to evaluate the effect of oxygenated compounds concentration and structure on sooting tendencies of surrogate Diesel and Biodiesel, and to investigate the properties and the oxidative reactivity of soot obtained by their combustion using an atmospheric axi-symmetric co-flow non-premixed flame burner. Results evidenced that ester functions contained in Biodiesel surrogates reduce soot production. This decrease was more pronounced when the concentration of the oxygenated additive investigated was higher. However, it has been determined that YSI decreases when the aliphatic carbon chain of the ester additive is longer. On the other hand, physico-chemical characterizations of the generated model soot revealed that oxygen and soluble organic fraction (SOF) content decreases when the amount of methyl ester based additives increases in the reference fuel. Moreover, the behavior towards oxidation indicated that the Biodiesel-derived soot was less reactive than the Diesel-derived one. Finally, it was noticed from the results obtained from laser granulometry and TPOs that the particle size distribution and the reactivity of model soot collected from the burner are in the same range of size and of maximum oxidation temperature as soot derived from a Diesel engine functioning under specific conditions and with different type of fuel blending.