A joint experimental and theoretical project has been initiated at Virginia Tech to study the effects of dual-mode combustion at high pressures for a two-dimensional Wolfhard-Parker burner. This thesis is the first stage of the theoretical part of the project, and contains a numerical study of laminar coflow diffusion flames stabilized on a confined Wolfhard-Parker burner. A global finite difference method is used where the nonlinear equations written on a stream function-vorticity formulation are solved with a flame sheet approach. The pseudotransient, approximative factorization method is utilized to solve the coupled system of equations. Adaptive gridding, numerical evaluation of Jacobians and iterations within time step are implemented for computational efficiency. Numerical results have been obtained for different fuels under different conditions. Comparison with measured data by Smyth et al. (1985) for a buoyancy dominated methane-air flame is made. The location of the flame front is accurately predicted. The temperature is over predicted in the fuel rich zone since pyrolysis and radiation effects have not been accounted for in the numerical model. Good agreement is observed for major species and velocities. As expected, large velocity increase and horizontal inflow of nitrogen and combustion products associated with buoyancy occur in the lower region of the flame. / Master of Science

The concurrent spreading of a boundary layer type diffusion flame is studied. The impossibility of obtaining a low velocity laminar flow without any perturbation induced by buoyancy has lead to the development of an experimental apparatus for use in micro-gravity facilities. Based on previous experimental observations, an original numerical approach has been developed showing, first the dominating role of the radiative heat transfer on the structure of the flame and second the major role of the soot on the extinction phenomenon at the flame trailing edge. The influence of the forced flow velocity, the fuel injection velocity and oxygen concentration on the geometry of the flame has been examined by imaging of CH* and OH* radicals spontaneous emission. Laser-Induced Incandescence (LII) is used to determine the soot field concentration in the flame. The soot formation has been studied by Laser Induced Fluorescence (LIF) of Polycyclic Aromatic Hydrocarbons (PAHs). The interaction between the reaction zone and the field of soot formation/oxidation is taken into account to analyze the flame length. These results can be used as the experimental input data for a future complete validation of numerical model simulating the soot formation and oxidation in this kind of flame.

An experimental study was performed to study soot formation and evolution in atmospheric, laminar, coflow, diffusion flames of Jet-A1, Synthetic Paraffinic Kerosene and their surrogates. Light extinction, rapid thermocouple insertion and thermophoretic sampling followed by transmission electron microscopy and atomic forced microscopy were used to obtain soot volume fraction profiles, temperature profiles and soot morphologies, respectively. Different soot evolution processes were observed on the flame centerline and on a streamline with a significantly different temperature history. Formation and agglomeration of the first soot particles are different on the two streamlines. Transparent liquid-like particles are produced in large volumes in the early regions of the flame centerline where T < 1500 K; these particles are undetectable by the extinction method with the wavelength of 632.8 nm. Most of the currently used computational soot models do not predict the liquid-like nature of nascent soot particles which has major effects on the modeling.

An experimental study was performed to study soot formation and evolution in atmospheric, laminar, coflow, diffusion flames of Jet-A1, Synthetic Paraffinic Kerosene and their surrogates. Light extinction, rapid thermocouple insertion and thermophoretic sampling followed by transmission electron microscopy and atomic forced microscopy were used to obtain soot volume fraction profiles, temperature profiles and soot morphologies, respectively. Different soot evolution processes were observed on the flame centerline and on a streamline with a significantly different temperature history. Formation and agglomeration of the first soot particles are different on the two streamlines. Transparent liquid-like particles are produced in large volumes in the early regions of the flame centerline where T < 1500 K; these particles are undetectable by the extinction method with the wavelength of 632.8 nm. Most of the currently used computational soot models do not predict the liquid-like nature of nascent soot particles which has major effects on the modeling.

Syngas, a gaseous mixture of H2, CO and diluents such as N2, CO2, is a clean fuel generated via gasification of coal or biomass. Syngas produced via gasification typically has low calorific values due to very high dilution levels (~60% by volume). It has been recognized as an attractive energy source for stationary power generation applications. The present work focuses on experimental and numerical investigation of syngas-air counterflow diffusion flames with varying composition of syngas. Laser-based diagnostic techniques such as Particle Imaging Velocimetry, Rayleigh thermometry and Laser-induced fluorescence have been used to obtain non-intrusive measurements of local extinction strain rates, temperature, quantitative OH and NO concentrations, respectively, for three different compositions of syngas. Complementing the experiments, numerical simulations of the counterflow diffusion flame have been performed to assess the performance of five H2/CO chemical kinetic mechanisms from the literature. The first part of the work involved determination of local extinction strain rates for six H2 /CO mixtures, with H2:CO ratio varying from 1:4 to 1:1. The extinction strain rates were observed to increase from 600 sec-1 to 2400 sec-1 with increasing H2:CO ratio owing to higher diffusivity and reactivity of the H2 molecule. Numerical simulations showed few mechanisms predicting extinction conditions within 5% of the measurements for low H2:CO ratios, however, deviations of 25% were observed for higher H2 :CO ratios. Sensitivity analyses revealed that the chain branching reactions, H+O2 <=>O+OH, O+H2 <=>H+OH and the third body reaction H+O2 +M<=>HO2 +M are the key reactions affecting extinction limits for higher H2:CO mixtures. The second phase of work involved quantitative measurement of OH species concentration in the syngas-air diffusion flames at strain rates varying from 35 sec-1 to 1180 sec-1. Non-intrusive temperature measurements using Rayleigh thermometry were made in order to provide the temperature profile necessary for full quantification of the species concentrations. The [OH] is observed to show a non-monotonous trend with increasing strain rates which is attributed to the competition between the effect of increased concentrations of H2 and O2 in the reaction zone and declining flame temperatures on the overall reaction rate. Although the kinetic mechanisms successfully captured this trend, significant deviations were observed in predictions and measurements in flames with H2:CO ratios of 1:1 and 4:1, at strain rates greater than 800 sec-1 . The key reactions affecting [OH] under these conditions were found to be the same reactions identified earlier during extinction studies, thus implying a need for the refinement of their reaction-rate parameters. Significant disagreements were observed in the predictions made using the chemical kinetic mechanisms from the literature in flames with high H2 content and high strain rate. The final phase of work focused on measurement of nitric oxide (NO) species concentrations followed by a comparison with predictions using various mechanisms. NO levels as high as ~ 48 ppm were observed for flames with moderate to high H2 content and low strain rate. Quantitative reaction pathway diagrams (QRPDs) showed thermal-NO, NNH and prompt-NO pathways to be the major contributors to NO formation at low strain rates, while the NNH pathway was the dominant route for NO formation at high strain rates. The absence of an elaborate CH chemistry in some of the mechanisms has been identified as the reason for underprediction of [NO] in the low strain rate flames. Overall, the quantitative measurements reported in this work serve as a valuable reference for validation of H2/CO chemical kinetic mechanisms, and the detailed numerical studies while providing an insight to the H2:CO kinetics and reaction pathways, have identified key reactions that need further refinement.

Simulações de chamas empregando mecanismos cinéticos detalhados são problemas computacionalmente demandantes. Por esse motivo, mecanismos reduzidos e técnicas de redução de cinética química vêm sendo desenvolvidos buscando uma melhor eficiência computacional. Mecanismos globais de poucos passos são particularmente populares pela simplicidade de programação nos códigos disponíveis. Assim, o objetivo da presente dissertação é avaliar modelagens simplificadas de cinética química na simulação numérica de chamas laminares 1D e 2D de metano diluído com dióxido de carbono. Mecanismos globais de 1, 2 e 4-passos são avaliados em comparação com o mecanismo detalhado GRI-Mech 3.0 na simulação unidimensional de chamas difusivas contra-corrente. O mecanismo global de melhor desempenho é então usado nas simulações bidimensionais de chamas difusivas tipo jato em comparação com a técnica de redução Flamelet-Generated Manifold. Observou-se que o mecanismo de 4-passos estudado apresenta bons resultados para o campo de temperaturas e para as principais espécies químicas, tanto nas simulações unidimensionais quanto nas bidimensionais. No entanto, espécies minoritárias como o CO e H2 não são bem reproduzidas. Fenômenos como posição de estabilização e penetração de oxidante na base de chamas tipo jato também não são capturadas quando o mecanismo global é usado. Por outro lado, a técnica FGM se mostrou capaz de prever tais fenômenos e resultou, adicionalmente, em um ganho computacional expressivo. / Numerical simulations of flames employing detailed kinetic mechanisms are computationally demanding problems. For this reason, reduced mechanisms and techniques of chemical kinetic reduction have been developed aiming better computational efficiency. Global mechanisms formed by few steps are particularly popular due to the simplicity of programing them in available codes. Thus, the objective of the present dissertation is to evaluate simplified chemical kinetics models in 1D and 2D numerical simulations of methane diluted with carbon dioxide laminar flames. Global mechanisms formed by 1, 2 and 4-steps are evaluated in comparison with the detailed mechanism GRI-Mech 3.0 in one-dimensional simulations of counterflow diffusive flames. The global mechanism with best performance is then used in two-dimensional simulations of diffusive jet flames for a comparison with the chemical reduction technique FGM. It was observed that a 4-step mechanism presented good results for temperature and major chemical species for both one and two-dimensional simulations. However, minor species like CO and H2 are not well reproduced. Phenomena such as stabilization position and oxygen penetration in the jet flame base are also not captured when the global mechanism is used. On the other hand, the technique Flamelet- Generated Manifold demonstrated to predict those phenomena and resulted, additionally, in an expressive computational gain.

Simulações de chamas empregando mecanismos cinéticos detalhados são problemas computacionalmente demandantes. Por esse motivo, mecanismos reduzidos e técnicas de redução de cinética química vêm sendo desenvolvidos buscando uma melhor eficiência computacional. Mecanismos globais de poucos passos são particularmente populares pela simplicidade de programação nos códigos disponíveis. Assim, o objetivo da presente dissertação é avaliar modelagens simplificadas de cinética química na simulação numérica de chamas laminares 1D e 2D de metano diluído com dióxido de carbono. Mecanismos globais de 1, 2 e 4-passos são avaliados em comparação com o mecanismo detalhado GRI-Mech 3.0 na simulação unidimensional de chamas difusivas contra-corrente. O mecanismo global de melhor desempenho é então usado nas simulações bidimensionais de chamas difusivas tipo jato em comparação com a técnica de redução Flamelet-Generated Manifold. Observou-se que o mecanismo de 4-passos estudado apresenta bons resultados para o campo de temperaturas e para as principais espécies químicas, tanto nas simulações unidimensionais quanto nas bidimensionais. No entanto, espécies minoritárias como o CO e H2 não são bem reproduzidas. Fenômenos como posição de estabilização e penetração de oxidante na base de chamas tipo jato também não são capturadas quando o mecanismo global é usado. Por outro lado, a técnica FGM se mostrou capaz de prever tais fenômenos e resultou, adicionalmente, em um ganho computacional expressivo. / Numerical simulations of flames employing detailed kinetic mechanisms are computationally demanding problems. For this reason, reduced mechanisms and techniques of chemical kinetic reduction have been developed aiming better computational efficiency. Global mechanisms formed by few steps are particularly popular due to the simplicity of programing them in available codes. Thus, the objective of the present dissertation is to evaluate simplified chemical kinetics models in 1D and 2D numerical simulations of methane diluted with carbon dioxide laminar flames. Global mechanisms formed by 1, 2 and 4-steps are evaluated in comparison with the detailed mechanism GRI-Mech 3.0 in one-dimensional simulations of counterflow diffusive flames. The global mechanism with best performance is then used in two-dimensional simulations of diffusive jet flames for a comparison with the chemical reduction technique FGM. It was observed that a 4-step mechanism presented good results for temperature and major chemical species for both one and two-dimensional simulations. However, minor species like CO and H2 are not well reproduced. Phenomena such as stabilization position and oxygen penetration in the jet flame base are also not captured when the global mechanism is used. On the other hand, the technique Flamelet- Generated Manifold demonstrated to predict those phenomena and resulted, additionally, in an expressive computational gain.

Simulações de chamas empregando mecanismos cinéticos detalhados são problemas computacionalmente demandantes. Por esse motivo, mecanismos reduzidos e técnicas de redução de cinética química vêm sendo desenvolvidos buscando uma melhor eficiência computacional. Mecanismos globais de poucos passos são particularmente populares pela simplicidade de programação nos códigos disponíveis. Assim, o objetivo da presente dissertação é avaliar modelagens simplificadas de cinética química na simulação numérica de chamas laminares 1D e 2D de metano diluído com dióxido de carbono. Mecanismos globais de 1, 2 e 4-passos são avaliados em comparação com o mecanismo detalhado GRI-Mech 3.0 na simulação unidimensional de chamas difusivas contra-corrente. O mecanismo global de melhor desempenho é então usado nas simulações bidimensionais de chamas difusivas tipo jato em comparação com a técnica de redução Flamelet-Generated Manifold. Observou-se que o mecanismo de 4-passos estudado apresenta bons resultados para o campo de temperaturas e para as principais espécies químicas, tanto nas simulações unidimensionais quanto nas bidimensionais. No entanto, espécies minoritárias como o CO e H2 não são bem reproduzidas. Fenômenos como posição de estabilização e penetração de oxidante na base de chamas tipo jato também não são capturadas quando o mecanismo global é usado. Por outro lado, a técnica FGM se mostrou capaz de prever tais fenômenos e resultou, adicionalmente, em um ganho computacional expressivo. / Numerical simulations of flames employing detailed kinetic mechanisms are computationally demanding problems. For this reason, reduced mechanisms and techniques of chemical kinetic reduction have been developed aiming better computational efficiency. Global mechanisms formed by few steps are particularly popular due to the simplicity of programing them in available codes. Thus, the objective of the present dissertation is to evaluate simplified chemical kinetics models in 1D and 2D numerical simulations of methane diluted with carbon dioxide laminar flames. Global mechanisms formed by 1, 2 and 4-steps are evaluated in comparison with the detailed mechanism GRI-Mech 3.0 in one-dimensional simulations of counterflow diffusive flames. The global mechanism with best performance is then used in two-dimensional simulations of diffusive jet flames for a comparison with the chemical reduction technique FGM. It was observed that a 4-step mechanism presented good results for temperature and major chemical species for both one and two-dimensional simulations. However, minor species like CO and H2 are not well reproduced. Phenomena such as stabilization position and oxygen penetration in the jet flame base are also not captured when the global mechanism is used. On the other hand, the technique Flamelet- Generated Manifold demonstrated to predict those phenomena and resulted, additionally, in an expressive computational gain.

La modélisation à échelle fine de la réaction sodium-eau est motivée par ses applications aux réacteurs nucléaires à neutrons rapides refroidis au sodium et aux réacteurs expérimentaux. La littérature indique que le contact entre le sodium liquide et l’eau donne naissance à un film gazeux où la réaction se produit sous la forme d’une flamme de diffusion gazeuse. Dans ce manuscrit, nous avons choisi de nous focaliser sur la combustion d’une goutte de sodium liquide immergée dans un volume infini d’eau. Plusieurs hypothèses simplificatrices sont introduites : en particulier, nous nous limitons au problème uni-dimensionnel. En supposant que le film gazeux a une masse volumique constante, une étude analytique révèle que l’état physique de l’hydroxyde de sodium a une forte influence sur le comportement du système : si la soude est entièrement gazeuse, la flamme s’éteint, tandis que si elle est entièrement condensée, une solution auto-similaire peut être exhibée et la combustion est entretenue. Un algorithme numérique est développé. Le modèle précédent est ensuite amélioré par la prise en compte de la compressibilité du film. Nous développons un algorithme bas Mach. Les calculs montrent que le film a un comportement oscillant, du fait de l’inertie de l’eau. Les taux de réaction calculés sont en bon accord avec les mesures d’Ashworth. Les conditions initiales dans le film sont inconnues : un mécanisme simplifié de formation du film est donc proposé, et une étude de sensibilité aux conditions initiales est effectuée. Les résultats ne dépendent que faiblement de l’état initial du système. / The fine-scale modeling of sodium-water reaction is motivated by its applications to sodium-cooled fast nuclear reactors and experimental irradiation reactors. As shown by several experiments from the literature, the contact between liquid sodium and water gives rise to a gaseous film where the reaction takes place in the form of a gaseous diffusion flame. In this manuscript, we have chosen to focus on the combustion of a liquid sodium drop immersed in an infinite volume of water. Several simplifying assumptions are introduced : in particular, we limit ourselves to the one-dimensional problem.Assuming the gaseous film has constant density, an analytical study shows that the physical state of sodium hydroxide has a strong influence on the behavior of the system : if soda is entirely vaporized, the flame gets choked, while, on the opposite, if it is entirely condensed, a self-similar solution can be exhibited and the combustion is sustained. A numerical algorithm is developed.Then, the previous model is improved by taking into account the gas compressibility. We develop a low Mach number algorithm. The computations show an oscillatory behavior of the one-dimensional film, due to the inertia of water. The calculated reaction rates are found to be in good agreement with Ashworth’s measurements. Initial conditions in the film are unknown : a simplified mechanism of film formation is therefore proposed, and a sensitivity analysis on initial conditions is carried out. The results are seen to be only slightly dependent on the initial state of the system.

Les lanceurs ont un impact sur la composition de l'atmosphere, et en particulier sur l'ozone stratospherique. Parmi tous les types de propulsion, les moteurs à propergol solide ont fait l'objet d'une attention particulière car leurs émissions sont responsables d'un appauvrissement significatif d'ozone dans le panache des lanceurs lors des premières heures suivant le lancement. Ce phénomène est principalement dû à la conversion de l'acide chlorhydrique, un composé chimique présent en grandes quantités dans les émissions de ce type de moteur, en chlore actif qui réagit par la suite avec l'ozone dans un cycle catalytique similaire à celui responsable du "trou de la couche d'ozone", cette diminution périodique de l'ozone en Antarctique. Cette conversion se produit dans le panache supersonique, où les hautes températures favorisent une seconde combustion entre certaines espèces chimiques du panache et l'air ambiant. L'objectif de cette étude est d'évaluer la concentration de chlore actif dans le panache d'un moteur à propergol solide en utilisant la technique des Simulations aux Grandes Echelles (SGE). Le gaz est injecté à travers la tuyère d'un moteur et une méthode de couplage entre deux instances du solveur de mécanique des fluides est utilisée pour étendre autant que possible le domaine de calcul derrière la tuyère (jusqu'à l'équivalent de 400 diamètres de sortie de la tuyère). Cette méthodologie est validée par une première SGE sans chimie, en analysant les caractéristiques de l'écoulement supersonique avec co-écoulement obtenu par ce calcul. Ensuite, le chimie mettant en jeu la conversion des espèces chlorées a été étudiée au moyen d'un modèle "hors-ligne" permettant de résoudre une chimie complexe le long de lignes de courant extraites d'un écoulement moyenné dans le temps résultant du calcul précédent (non réactif). Enfin, une SGE multi-espèces est réalisée, incluant un schéma chimique auparavant réduit afin de limiter le coût de calcul. Cette simulation représente une des toutes premières SGE d'un jet supersonique réactif, incluant la tuyère, effectuée sur un domaine de calcul aussi long. En capturant avec précision le mélange du panache avec l'air ambiant ainsi que les interactions entre turbulence et combustion, la technique des simulations aux grandes échelles offre une évaluation des concentrations des espèces chimiques dans le jet d'une precision inédite. Ces résultats peuvent être utilisés pour initialiser des calculs atmosphériques sur de plus larges domaines, afin de modéliser les réactions entre chlore actif et ozone et de quantifier l'appauvrissement en ozone dans le panache. / Rockets have an impact on the chemical composition of the atmosphere, and particularly on stratospheric ozone. Among all types of propulsion, Solid-Rocket Motors (SRMs) have given rise to concerns since their emissions are responsible for a severe decrease in ozone concentration in the rocket plume during the first hours after a launch. The main source of ozone depletion is due to the conversion of hydrogen chloride, a chemical compound emitted in large quantities by ammonium perchlorate based propellants, into active chlorine compounds, which then react with ozone in a destructive catalytic cycle, similar to those responsible for the Antartic "Ozone hole". This conversion occurs in the hot, supersonic exhaust plume, as part of a strong second combustion between chemical species of the plume and air. The objective of this study is to evaluate the active chlorine concentration in the far-field plume of a solid-rocket motor using large-eddy simulations (LES). The gas is injected through the entire nozzle of the SRM and a local time-stepping method based on coupling multi-instances of the fluid solver is used to extend the computational domain up to 400 nozzle exit diameters downstream of the nozzle exit. The methodology is validated for a non-reactive case by analyzing the flow characteristics of the resulting supersonic co-flowing under-expanded jet. Then the chemistry of chlorine is studied off-line using a complex chemistry solver applied on trajectories extracted from the LES time-averaged flow-field. Finally, the online chemistry is analyzed by means of the multi-species version of the LES solver using a reduced chemical scheme. To the best of our knowledge, this represents one of the first LES of a reactive supersonic jet, including nozzle geometry, performed over such a long computational domain. By capturing the effect of mixing of the exhaust plume with ambient air and the interactions between turbulence and combustion, LES offers an evaluation of chemical species distribution in the SRM plume with an unprecedented accuracy. These results can be used to initialize atmospheric simulations on larger domains, in order to model the chemical reactions between active chlorine and ozone and to quantify the ozone loss in SRM plumes.