A new parallel implicit adaptive mesh refinement (AMR) algorithm is developed for the prediction of unsteady behaviour of laminar flames. The scheme is applied to the solution of the system of partial-differential equations governing time-dependent, two- and three-dimensional, compressible laminar flows for reactive thermally perfect gaseous mixtures. A high-resolution finite-volume spatial discretization procedure is used to solve the conservation form of these equations on body-fitted multi-block hexahedral meshes. A local preconditioning technique is used to remove numerical stiffness and maintain solution accuracy for low-Mach-number, nearly incompressible flows. A flexible block-based octree data structure has been developed and is used to facilitate automatic solution-directed mesh adaptation according to physics-based refinement criteria. The data structure also enables an efficient and scalable parallel implementation via domain decomposition. The parallel implicit formulation makes use of a dual-time-stepping like approach with an implicit second-order backward discretization of the physical time, in which a Jacobian-free inexact Newton method with a preconditioned generalized minimal residual (GMRES) algorithm is used to solve the system of nonlinear algebraic equations arising from the temporal and spatial discretization procedures. An additive Schwarz global preconditioner is used in conjunction with block incomplete LU type local preconditioners for each sub-domain. The Schwarz preconditioning and block-based data structure readily allow efficient and scalable parallel implementations of the implicit AMR approach on distributed-memory multi-processor architectures. The scheme was applied to solutions of steady and unsteady laminar diffusion and premixed methane-air combustion and was found to accurately predict key flame characteristics. For a premixed flame under terrestrial gravity, the scheme accurately predicted the frequency of the natural buoyancy induced oscillations. The performance of the proposed parallel implicit algorithm was assessed by comparisons to more conventional solution procedures and was found to significantly reduce the computational time required to achieve a solution in all cases investigated.

Syngas is a mixture of hydrogen (H2), carbon monoxide (CO) and other species including nitrogen (N2), water (H2O), methane (CH4) and higher hydrocarbons. Syngas is a highly desired product because it is very versatile. It can be used for combustion in turbines or engines, converted to H2 for use in fuel cells, turned into diesel or other high-molecular weight fuels by the Fischer-Tropsch process and used as a chemical feedstock. Syngas can be derived from hydrocarbons in the presence of oxidizer or water as in steam reforming. There are many demonstrated methods to produce syngas with or without water addition including catalytic methods, plasma reforming and combustion. The goal of this study is to add to the understanding of non-catalytic conversion of hydrocarbon fuels to syngas, and this was accomplished through two investigations: the first on fuel conversion potential and the second on the effect of preheat temperature. A primarily experimental investigation of the conversion of jet fuel and butanol to syngas was undertaken to understand the potential of these fuels for conversion. With these new data and previously-published experimental data, a comparison amongst a larger set of fuels for conversion was also conducted. Significant soot formation was observed in experiments with both fuels, but soot formation was so significant in the jet fuel experiments that it limited the range of experimental operating conditions. The comparison amongst fuels indicated that higher conversion rates are observed with smaller molecular weight fuels, generally. However, equilibrium calculations, which are often used to determine trends in fuel conversion, showed the opposite trend. In order to investigate preheat temperature, which is one important aspect of non-catalytic conversion, experiments were undertaken with burner-stabilized flames that are effectively 1-D and steady-state. An extensive set of model calculations were compared to the obtained experimental data and was used to investigate the effect of preheat temperatures that were beyond what was achievable experimentally. Throughout the range of operating conditions that were tested experimentally, the computational model was excellent in its predictions. Experiments where the reactants were preheated showed a significant expansion of the stable operating range of the burner (increasing the equivalence ratio at which the flame blew off). However, increasing preheat temperature beyond what is required for stabilization did not improve syngas yields. / text

A Bayesian framework for quantification of uncertainties has been used to quantify the uncertainty introduced by chemistry models. This framework adopts a probabilistic view to describe the state of knowledge of the chemistry model parameters and simulation results. Given experimental data, this method updates the model parameters' values and uncertainties and propagates that parametric uncertainty into simulations. This study focuses on syngas, a combination in various ratios of H2 and CO, which is the product of coal gasification. Coal gasification promises to reduce emissions by replacing the burning of coal with the less polluting burning of syngas. Despite the simplicity of syngas chemistry models, they nonetheless fail to accurately predict burning rates at high pressure. Three syngas models have been calibrated using laminar flame speed measurements. After calibration the resulting uncertainty in the parameters is propagated forward into the simulation of laminar flame speeds. The model evidence is then used to compare candidate models. Sensitivity studies, in addition to Bayesian methods, can be used to assess chemistry models. Sensitivity studies provide a measure of how responsive target quantities of interest (QoIs) are to changes in the parameters. The adjoint equations have been derived for laminar, incompressible, variable density reacting flow and applied to hydrogen flame simulations. From the adjoint solution, the sensitivity of the QoI to the chemistry model parameters has been calculated. The results indicate the most sensitive parameters for flame tip temperature and NOx emission. Such information can be used in the development of new experiments by pointing out which are the critical chemistry model parameters. Finally, a broader goal for chemistry model development is set through the adjoint methodology. A new quantity, termed field sensitivity, is introduced to guide chemistry model development. Field sensitivity describes how information of perturbations in flowfields propagates to specified QoIs. The field sensitivity, mathematically shown as equivalent to finding the adjoint of the primal governing equations, is obtained for laminar hydrogen flame simulations using three different chemistry models. Results show that even when the primal solution is sufficiently close for the three mechanisms, the field sensitivity can vary. / text

Fully understanding soot formation in flames is critical to the development of practical combustion devices, which typically operate at high pressures, and fire suppression systems in space. Flames display significant changes under microgravity and high-pressure conditions as compared to normal-gravity flames at atmospheric pressure, but the exact causes of these changes are not well-characterized. As such, the effects of gravity and pressure on the stability characteristics and sooting behavior of laminar coflow diffusion flames were investigated. To study these effects, a new highly-scalable combustion modelling tool was developed specifically for use on large multi-processor computer architectures. The tool is capable of capturing complex processes such as detailed chemistry, molecular transport, radiation, and soot formation/destruction in laminar diffusion flames. The proposed algorithm represents the current state of the art in combustion modelling, making use of a second-order accurate finite-volume scheme and a parallel adaptive mesh refinement algorithm on body-fitted, multi-block meshes. An acetylene-based, semi-empirical model was used to predict the nucleation, growth, and oxidation of soot particles. Reasonable agreement with experimental measurements for different fuels and pressures was obtained for predictions of flame height, temperature and soot volume fraction. Overall, the algorithm displayed excellent strong scaling performance by achieving a parallel efficiency of 70% on 384 processors. The effects of pressure and gravity were studied for flames of two different fuels: ethylene-air flames between pressures of 0.5–5 atm and methane-air flames between 1–60 atm. Based on the numerical predictions, zero-gravity flames had lower temperatures, broader soot-containing zones, and higher soot concentrations than normal-gravity flames at the same pressure. Buoyant forces caused the normal-gravity flames to narrow with increasing pressure while the increased soot concentrations and radiation at high pressures lengthened the zero-gravity flames. Low-pressure flames at both gravity levels exhibited a similar power-law dependence of the maximum carbon conversion on pressure which weakened as pressure was increased. This dependence decayed at a faster rate in zero gravity when pressure was increased beyond 1–10 atm.

Modeling the Response of Premixed Flames to Flow Disturbances Preetham 178 pages Directed by Dr. Tim Lieuwen Low emissions combustion systems for land based gas turbines rely on a premixed or partially premixed combustion process. These systems are exceptionally prone to combustion instabilities which are destructive to hardware and adversely affect performance and emissions. The success of dynamics prediction codes is critically dependent on the heat release model which couples the flame dynamics to the system acoustics. So the principal objective of the current research work is to predict the heat release response of premixed flames and to isolate the key non-dimensional parameters which characterize its linear and nonlinear dynamics. Explicit analytical solutions of the G- equation are derived in the linear and weakly nonlinear regime using the Small Perturbation Method (SPM). For the fully nonlinear case, the flame-flow interaction effects are captured by developing an unsteady, compressible, coupled Euler-G-equation solver with a Ghost Fluid Method (GFM) module for applying the jump conditions across the flame. The flame s nonlinear response is shown to exhibit two qualitatively different behaviors. Depending on the operating conditions and the disturbance field characteristics, it is shown that a combustor may exhibit supercritical bifurcations leading to a single stable limit cycle amplitude or exhibit sub-critical bifurcations wherein multiple stable solutions for the instability amplitude are possible. In addition, this study presents the first analytical model which captures the effects of unsteady flame stretch on the heat release response and thus extends the applicability of current models to high frequency instabilities, such as occurring during screech. It is shown that unsteady stretch effects, negligible at low frequencies (100 s of Hz) become significant at screeching frequencies (1000 s of Hz). Furthermore, the analysis also yields insight into the significant spatial dependence of the mean and perturbation velocity field induced by the coupling between the flame and the flow field. In order to meaningfully compare the heat release response across different flame configurations, this study has identified that the reference velocity (for defining the transfer function) should be based on the effective normal velocity perturbing the flame and the Strouhal number should be based on the effective residence time of the flame wrinkles.

Fully understanding soot formation in flames is critical to the development of practical combustion devices, which typically operate at high pressures, and fire suppression systems in space. Flames display significant changes under microgravity and high-pressure conditions as compared to normal-gravity flames at atmospheric pressure, but the exact causes of these changes are not well-characterized. As such, the effects of gravity and pressure on the stability characteristics and sooting behavior of laminar coflow diffusion flames were investigated. To study these effects, a new highly-scalable combustion modelling tool was developed specifically for use on large multi-processor computer architectures. The tool is capable of capturing complex processes such as detailed chemistry, molecular transport, radiation, and soot formation/destruction in laminar diffusion flames. The proposed algorithm represents the current state of the art in combustion modelling, making use of a second-order accurate finite-volume scheme and a parallel adaptive mesh refinement algorithm on body-fitted, multi-block meshes. An acetylene-based, semi-empirical model was used to predict the nucleation, growth, and oxidation of soot particles. Reasonable agreement with experimental measurements for different fuels and pressures was obtained for predictions of flame height, temperature and soot volume fraction. Overall, the algorithm displayed excellent strong scaling performance by achieving a parallel efficiency of 70% on 384 processors. The effects of pressure and gravity were studied for flames of two different fuels: ethylene-air flames between pressures of 0.5–5 atm and methane-air flames between 1–60 atm. Based on the numerical predictions, zero-gravity flames had lower temperatures, broader soot-containing zones, and higher soot concentrations than normal-gravity flames at the same pressure. Buoyant forces caused the normal-gravity flames to narrow with increasing pressure while the increased soot concentrations and radiation at high pressures lengthened the zero-gravity flames. Low-pressure flames at both gravity levels exhibited a similar power-law dependence of the maximum carbon conversion on pressure which weakened as pressure was increased. This dependence decayed at a faster rate in zero gravity when pressure was increased beyond 1–10 atm.

Les suies de combustion sont principalement connues pour leur caractère nocif, dans le cas des feux de forêt, de fumées de cheminées ou d'émissions polluantes d'un tuyau d'échappement. Cependant, le noir de carbone, un produit industriel de combustion d'hydrocarbures largement utilisé dans notre vie quotidienne. La surface d'une particule de suies ou de noir de carbon joue un rôle important tant au niveau de son utilisation que de son effet nocif. Il est donc important de connaître la masse, le volume ainsi que la morphologie des suies. En particulier, la surface des particules est un paramètre important pour prédire leur utilisation ainsi que leur effet nocif. Les suies sont généralement des agrégats présentant une structure fractale constituée d'éléments de forme sphérique, appelés particules primaires. Il est possible de connaître la surface des agrégats à partir de la distribution en taille de particules primaires (PPSD-Primary particules size distribution). Compte tenu de l'intérêt grandissant pour la surface des particules et leurs évolutions, il est aujourd’hui nécessaire d'étendre les modèles numériques pour la prévision de la PPSD. De plus, comme la taille des la particules primaires influence les processus chimiques et les processus de collision, la prise en compte de ce paramètre peut améliorer les prévisions des modèles. Les flammes multidimensionnelles laminaires, comme les flammes de diffusion, sont moins complexes que les flammes rencontrées dans les systèmes de combustion industriels. Cependant, les processus de formation de suies sont analogues dans les deux cas, ce qui rend l'étude de ces flammes intéressante. Afin d'obtenir une description détaillée des processus chimiques ayant lieu dans ces flammes tout en maintenant le coût de calcul à un niveau abordable, l'utilisation de modèles sectionnels discrets chimiques (CDSchemical discret sectional methods) est un choix approprié. Le développement de modèles CDS est au coeur de cette thèse. D'abord, une stratégie numérique pour déterminer la taille des particules primaires est présentée dans le contexte des modéles CDS. Elle repose sur la résolution d'une équation de transport pour la densité en nombre de particules primaires pour chaque section d'agrégats considérée. Pour valider la taille des particules primaires déterminée numériquement, les résultats doivent être comparés avec des données expérimentales obtenues via la technique d'Incandescence Induite par Laser résolue temporellement (TiRe-LII). Cette comparaison, dite inverse, est affectée par les incertitudes expérimentales et les hypothèses sous-jacentes au post-traitement du signal TiRe-LII pour obtenir la PSD. Pour améliorer la stratégie de validation, une nouvelle approche, dite directe, est proposée pour la validation de la PPSD à partir des données obtenues par TiRe-LII. Elle est basée sur la reconstruction numérique de l'évolution temporelle du signal d'incandescence à partir des résultats numériques et de sa comparaison avec le signal mesuré. L'efficacité de l'approche proposée est démontrée a priori en évaluant l'erreur potentiellement évitée par la nouvelle stratégie. Le modèle proposé pour le suivi des particules primaires est ensuite validé en utilisant à la fois les approches ’directe’ et ’inverse’ sur les flammes cibles issues de l'International Sooting FlameWorkshop (ISF): une flamme pré-mélangée éthylèneair et une flamme de diffusion coflow avec deux dilutions différentes. Le caractère général du modèle est discuté en effectuant une étude de sensibilité des résultats aux paramètres du modèle même. Enfin, le modèle est utilisé pour comprendre l'effet de la dilution du combustible sur la taille des particules primaires dans les flammes de diffusion en examinant les corrélations possibles entre phase gazeuse et phase solide ainsi que l'évolution temporelle des particules le long de leur trajectoires. / An image appearing when the phrase soot is heard is the smoke emitted by an exhaust pipe. The imperfect combustion of hydrocarbon fuels is a source of this harmful pollutant. The industrially controlled combustion of hydrocarbons can provide the carbon black, an industrial product widely used in our everyday life. For both its utilization and its harming effect, the surface of these combustion generated particles plays an important role, therefore, it is of interest to possess information on the particle morphology beside its mass or volume. Soot particles were found, at various conditions, to have a fractal-like structure built up from spherical shape building blocks, socalled primary particles. This increased interest in the particle surface and its evolution gives the motivation to extend numerical models to provide related information, i.e. particle surface or primary particle size. Furthermore, as the primary particle size influences the chemical and collisional processes, accounting for this parameter can improve the model predictions. The requirements for numerical models are various depending on the purpose of the simulation. Multidimensional laminar flames, like a laminar coflow diffusion flame, are less complex than flames of industrial combustion systems. However, the soot formation processes are analogous in the two cases, therefore, the investigation of these flames are of interest. In order to obtain a detailed description of the chemical processes, while keeping the computational cost in these flames at an affordable level, using chemical discrete sectional models is a suitable choice. As in their current version, these models do not provide information on the primary particle size their development in this direction is of interest. Guided by the above motivation, a numerical strategy to determine the primary particle size is presented in the context of the chemical sectional models. The proposed strategy is based on solving the transport equation of the primary particle number density for each considered aggregate section. In order to validate numerical primary particle size, the comparison to experimental data is required. Due to its numerous advantages, the Time-Resolved Laser-Induced Incandescence (TiRe-LII) technique is a nowadays popular experimental method. However, the comparison of the numerically and the experimentally obtained primary particle size may be charged with uncertainties introduced by the additional measurements or assumptions of the numerous parameters required to derive primary particle size from the detected signal. In order to improve the validation strategy, an additional approach for primary particle size distribution validation with TiRe-LII is proposed. This is based on the reconstruction of the temporal evolution of incandescence from the numerical results and its comparison with the measured signal. The effectiveness of this ’forward’ method is demonstrated a priori by quantifying the errors potentially avoided by the new strategy. The validity of the proposed primary particle tracking model is tested by both the traditional ’inverse’ and the ’forward’ method on target flames of the International Sooting Flame (ISF) Workshop. In particular a laminar premixed ethylene flame is considered first. Then, two laminar coflow ethylene flames with different dilutions are put under the scope. The sensitivity to the model parameters, such as accounting for the surface rounding and the choice of smallest aggregating particle size, is explored in both the premixed flame and in the coflow flame with highest ethylene content. To understand the effect of the fuel stream dilution on the primary particle size in the coflow flame, first, the flame-flow interaction and the effect of the dilution on the flame structure is investigated. [...]

The quenching of undesired flames by cold surfaces has been investigated for more than a century. The current quenching theory can predict simple configurations, this is not the case for real environments such as fuel management systems. Flames are sensitive to numerous parameters, such as fuel, mixture fraction, pressure, temperature, flow properties, acoustics, radiation, and surface interactions. The effects of some of these parameters are very well documented but there is a lack of information regarding the effects of acoustics and flow. This dissertation work will focus on improving the understanding of flow effect on the quenching of premixed gaseous flames. First, the effect of apparent velocity on flame quenching was investigated for different fuels and equivalence ratios. An experimental facility is designed such that the apparent flame velocity at which the flame enters and propagates through the channel can be varied without changing the initial mixture condition. High-speed (15,000 frames per second (FPS)) Schlieren and dynamic pressure measurement were used to measure the apparent flame velocity and to assess the flame quenching, respectively. This study showed that the high-speed laminar flames are harder to quench compared to self-propagating and turbulent flames. A similar trend was obtained for all the conditions investigated, lean and stoichiometric methane-air, lean propane-air, and lean ethylene-air mixtures. Further investigation was carried out to understand the quenching of high-speed laminar flames. The flame propagation through the channel was investigated using Hydroxyl (OH) planar laser induced fluorescence (PLIF). This study showed that the OH intensity fell below the detection threshold in the later part of the channel when quenching is observed. Then, the influence of heat transfer was investigated using spatial and temporal evolution of the temperature in the quenching channel. A high-speed (10 kHz) filtered Rayleigh scattering (FRS) technique was used to measure the one-dimensional time-resolved temperature profile. Three different channel heights (H = 1.3, 1.5, 2.0 mm) were investigated. Based on the evolution of the temperature profile in the quenching channel, a new parameter was identified and the importance of its evolution on the flame quenching was discussed.

With global emission regulations becoming stringent, development of new combustion technologies that meet future emission regulations is essential. In this vein, this dissertation presents the application of sensitive diagnostic tools to validate and improve chemical kinetic mechanisms that play a fundamental role in the design of new combustion technologies. First, a novel high sensitivity laser-based sensor with a wide frequency tuning range (900 – 1000 cm-1) was developed utilizing pulsed cavity ringdown spectroscopy (CRDS) technique. The novel laser-based sensor was illustrated by measuring trace amounts of multiple combustion intermediates, namely ethylene, propene, allene, and 1-butene in a static cell at ambient conditions. Subsequently, pulsed CRDS technique was utilized to develop an ultra-fast, high sensitivity diagnostic to monitor trace concentrations of ethylene in shock tube pyrolysis experiments. This diagnostic represented the first ever successful application of CRDS technique to transient species measurements in a shock tube. The high sensitivity and fast time response (10μs) diagnostic may be utilized for measuring other key neutrals and radicals which are crucial in the oxidation chemistry of practical fuels. Secondly, a quadrupole mass spectrometer (QMS) was employed to measure relative cation mole fractions in atmospheric and low-pressure (30 Torr) flames of methane/oxygen diluted in argon. Lean, stoichiometric and rich flames were 4 examined to evaluate the dependence of ion chemistry on flame stoichiometry. Spatial distribution of cations was compared with predictions of an existing ion chemistry model. Based on the extensive measurements carried out in this work, modifications were suggested to improve the ion chemistry model to enhance the fidelity of such mechanisms. In-depth understanding of flame ion chemistry is vital to model the interaction of flames with electric fields and thereby pave the way to enable active combustion control for increased efficiency and reduced emissions. Finally, a compact fast time-response time-of-flight mass spectrometer (TOFMS) was coupled to the shock tube through a pin-hole end-wall to enable timeresolved species concentration measurements. This diagnostic tool was demonstrated by investigating the decomposition of 1,3,5-trioxane over a wide range of shock conditions. Reaction rate coefficients were extracted by the best fit to the experimentally measured species time-histories. TOF-MS coupled to the shock tube is an ideal diagnostic tool for developing kinetic mechanisms for future fuels due to its ability to simultaneously measure several species during fuel pyrolysis/oxidation processes.

Following stringent mandates from environmental regulatory authorities worldwide, various steps are being implemented to ensure clean combustion with minimum emissions, including fuel dilution, mild combustion and additives. Due to the need to understand combustion characteristics in primary applications (engines and turbines) with minimum emissions, the laminar burning velocity of natural gas has been measured with CO2 dilution and a wide range of blends with higher hydrocarbons. And because it has improved anti-knock quality to reduce greenhouse gas emissions (GHGE), the demand for oxygenated gasoline is now worldwide, making a compelling case for determining combustion behavior of oxygenated gasoline doped with hydrogen, ozone and carbon monoxide. The first section of this dissertation discusses dilution of methane with CO2 at elevated pressures, providing insight into comparative laminar burning characteristics in a wide range of equivalence ratios, particularly significant at elevated initial pressure. Utilizing CHEMKIN, a detailed kinetic study has been performed that explains the varying dependence on dilution ratio controlled by initial pressure. The second phase of this work reports the laminar burning velocity measurement of commercial gasoline. A TPRFE surrogate was used here to investigate burning characteristics and to provide detailed kinetic analysis of gasoline doped with additives (hydrogen, carbon monoxide and ozone). A study was also made of the behavior of gasoline with these additives in practical applications like engine and turbines. For this purpose, laminar burning velocity was measured at elevated pressures and temperatures, by varying the concentrations of synthetic EGR, and followed by measuring turbulent burning velocity at two turbulent intensities.