In this dissertation, some practical ignition techniques are presented that show how some problems of lean-burn combustion can be overcome. Then, to shed light on the effects of the ignition techniques described, the focus shifts to the more specific problem of the early phase of spark ignition. Thermal models of ignition are reviewed. These models treat the energy provided by the electrical discharge as a point source, delivered infinitely fast and creating a spherically symmetric ignition kernel. The thesis challenges the basis of these thermal models by reviewing the work of many investigators who have clearly shown that the temporal characteristics of the discharge have a profound effect upon ignition. Photographic evidence of the early phase of ignition, as well as other evidence from the literature, is also presented. The evidence clearly demonstrates that the morphology of spark kernels in the early phase of development is toroidal, not spherical as suggested by thermal models. A new perspective for ignition, a fluid dynamic point of view, is described. The common ignition devices are then classified according to fluid dynamics. A model describing the behaviour of spark kernels is presented, which extends a previously established mixing model for plasma jets, to the realm of conventional axial discharges. Comparison of the model behaviour to some limited data is made. The model is modified by including the effect of heat addition from combustion, and ignition criteria are discussed. / Graduate

No description available.

Cette Thèse de doctorat est consacrée à l’étude, par simulation aux grandes échelles ou LES (Larg eeddy simulation), d’un jet pariétal d’hydrogène sous-détendu dans un écoulement transverse supersonique d’air vicié. Cette configuration est représentative des conditions d’écoulement rencontrées dans les moteurs aérobies de type super-statoréacteurs (scramjets). En effet, les futurs systèmes de transport à grande vitesse dépendent fortement du développement de ce type de moteur. Dans de telles conditions, l’écoulement d’air chaud est maintenu supersonique dans la chambre de combustion afin de réduire les effets induits par l’échauffement et la dissociation de l’air. Nous étudions les processus de mélange et de combustion qui se développent en aval du jet de combustible. Ce travail s’appuie sur l’emploi d’un outil de simulation numérique haute fidélité : CREAMS (Compressible REActive Multi-species Solver) développé à l’Institut Pprime. Ce code de calcul met en oeuvre des schémas numériques d’ordre élevé : schéma Runge–Kutta d’ordre 3 pour l’intégration temporelle combiné à un schéma WENO d’ordre 7 et centré d’ordre 8 pour la discrétisation spatiale. Les simulations réalisées dans des conditions inertes permettent de caractériser l’importance des interactions choc/turbulence avec une attention particulière accordée à la description des fluctuations de composition à l’échelle non-résolue (i.e. sous-maille). Compte tenu de leur niveau de résolution,les simulations réactives permettent quant à elles une analyse détaillée du mode de stabilisation et des régimes de combustion turbulente rencontrés fournissant ainsi des informations très précieuses quant à l’adéquation des modélisations existantes pour ces conditions extrêmes. / This dissertation is devoted to the Large-eddy simulation (LES) study of a wall hydrogen underexpanded jet in a supersonic crossflow of vitiated air. This configuration is representative of flow conditions encountered in aerospace engines such as supersonic combustion ramjet (scramjets). Indeed, future of high-speed transport systems heavily depends on the development of this type of engine. Under such conditions, the high temperature flow of vitiated air is maintained supersonic in the combustion chamber to reduce effects of heating and dissociation. The mixing and combustion processes that develop downstream of the fuel jet are studied. This work is based on the use of a high fidelity numerical simulation: CREAMS (Compressible REActive Multi-species Solver) which is developed at the Pprime Institute. This computational solver makes use of high precision numerical schemes: a 3rd order Runge–Kutta scheme for the time integration combines with a 7th order WENO and 8th order centered scheme for the spatial discretisation. Non-reactive simulations allow to characterize the importance of shock/turbulence interactions with special attention paid to the description of the unresolved (i.e. sub-grid scale) scalar fluctuations. The reactive simulations allow to perform a detailed analysis of the stabilization mode and turbulent combustion regimes tha are encountered, thus providing valuable information about the possible adequacy of the available representation for these extreme conditions.

Two mathematical models have been developed to simulate two-dimensional, premixed, laminar, stationary, axisymmetric methane-air flames, and successfully validated with non-intrusive Coherent Anti-Stokes Raman Spectroscopy (CARS) temperature measurements. With the first model, the heat releaser ate model, volumetric heat release rate was generalised from one-dimensional computations. This approximation greatly simplified the set of governing equations that need to be solved. However, it cannot describe the effects of high stretch rates or of negative stretch rate. The second model made use of a number of reduced chemical kinetic schemes, with realistic elementary reactions. These were drawn from the literature and realistic transport properties have been included. With this model, based on the work of Peters (1985), the effects of stretch are automatically accounted for. Practical experimental validation was obtained with a multiple slot burner, supplied by the collaborating body, British Gas p1c. Temperature fields, obtained with the CARS technique, partially validated the reduced chemical kinetic scheme model. Some uncertainty arose in the prediction of heat loss to the burner tube. A numerical algorithm based upon the SUVIPLE method was employed,with a fully staggered grid. Various discretisation schemes were examined with the heat release model. Based on these tests, the hybrid scheme was selected for use in the reduced model. With this approach, a few reduced kinetic schemes have been selected and implemented. The most successfuol ne was the Peters( 1985) scheme. This consisted of 4 global reaction steps with 18 elementary reactions and 7 non-steady chemical species. The scheme has been employed in all the detailed computations in the present study. With this scheme, two-dimensional field solutions, for methane-air mixtures with equivalence ratios of 0.75,0.84 and 1.0, with slot widths of 2 mm, and 3 mm, and mean inlet velocities ranging from 0.3 m/s to 2.8 m/s have been obtained. Detailed flame structures have been obtained for all these conditions. Under these conditions, a number of parameters, essential in burner design and stability analysis, have been investigated. These includes flame height, flame thickness, and heat loss to the burner tube. The loss can range between 3% and 32% of the chemical energy in the premixture. The computations reveal the stretch rates acting on the flame and their effects on the burning velocity. At low flow rates the base of the flame has a negative stretch rate, while the flame tip is positively stretched. These effects are reversed at high flow rates. From the localised relationships between stretch rate and burning velocity, Markstein lengths have been evaluated,for different mixtures and the values compared with those obtained experimentally by other researchers. In general, there was good agreement despite the large scatter in the experimental values. The results further showed that the effects of the two components of flame stretch, namely flame curvature and aerodynamic straining, on burning velocities were very different. It seems appropriate to introduce two Markstein lengths to correlate burning velocity and the two components of stretch and these have been evaluated. Aerodynamic straining has a significantly larger effect on burning velocity than has flame curvature.

This study was concerned with improvement and validation of a thermodynamic spark ignition engine simulation code developed in Leeds. Experimental validation data were generated using a central ignition, disc-shaped combustion chamber variant of a ported single-cylinder research engine with full-bore overhead optical access. These data included simultaneous measurement of cylinder pressure and flame position at different operating conditions. The engine was skip fired (fired once every five cycles), to remove residuals and ensure well defined in-cylinder fuel-air mixture for simulation. Flames were imaged using a digital camera capturing the light emitted from the flame ("natural light"). New methods were developed to process the pressure and film data. Flame pictures were processed to determine enflamed area, mean flame radius and flame centroid. Parameters were also developed to describe flame "circularity" ("shape factor") and to describe asymmetry of flame approach to the cylinder walls ("active perimeter fraction", APF). Time-base crank angle records allowed evaluation of engine speed variation within a cycle and mean engine speed for a cycle. Although generated principally for model validation, the experimental results proved interesting in their own right. Middle, slow and fast cycles were defined for each condition. Analysis of these cycles suggested that there was no correlation between the initial flame centroid displacement, its locus over the flame propagation period or the flame "shape factor" and the speed of combustion and pressure development. As the flame approached the wall, the active perimeter fraction fell in a similar manner for all the middle cycles. Substantial modifications were made to a pre-existing thermodynamic engine cycle code. Deficiencies in the blowby, heat transfer and thermodynamic aspects were corrected. An additional ("Zimont") turbulent burning velocity sub-model and a new routine for the influence of engine speed variation within a cycle were incorporated into the code. The active perimeter fraction parameter function determined in the experiments was encoded to allow for the effects of flame-wall contact on entrainment rate during the late flame propagation. A radial stratified charge model was also developed. Burned gas expansion over the flame propagation period was shown to significantly change the unburned gas charge stratification from the initial variation. Two types of initial stratification (linear and parabolic distributions, rich of the centre and lean close to the wall) were imposed. Faster combustion development was observed in both cases, c. f that for equivalent homogeneous charge. Good agreement was observed between experimental results and "Zimont model" predictions at different equivalence ratios and engine speeds. Other computations using the pre-existing Leeds K and KLe correlations gave reasonable predictions at the various engine speeds and at rich conditions; however, they yielded slower results than experimentally observed for lean conditions.

No description available.

No description available.

No description available.

No description available.

No description available.