Un des principaux enjeux de l'industrie aéronautique est la recherche du moteur au meilleur rendement possible, pour satisfaire des contraintes économiques, techniques et environnementales. Les turbomachines bénéficient d'un constant perfectionnement depuis plus de 60 ans, et cette technologie semble avoir atteint un plateau. Une rupture technologique est aujourd'hui nécessaire, comme la combustion à volume constant (CVC). Le gain attendu est suffisant pour tenter de remplacer les systèmes actuels où la combustion se fait à pression constante. La combustion à isovolume fait appel à des mécanismes encore rarement maitrisés dans le contexte aéronautique. Sa compréhension passe par des expérimentations et des modèles théoriques et numériques. L’objectif de cette thèse est de développer une théorie et un outil de simulation LES (Large Eddy Simulation) appliqué au cas du concept ‘thermoréacteur’. Ainsi, la première étape a consisté à mettre en place un outil de simulation 0D traduisant l’évolution d’un cycle moteur de type CVC (Combustion à Volume Constant). Certains modèles utilisés dans cet outil 0D sont basés sur des corrélations expérimentales. D'autres présentent des paramètres à déterminer à partir de simulations numériques. La simulation 3D d’un système de type CVC est envisageable aujourd’hui grâce aux progrès récents des méthodes LES. Ainsi, des simulations du thermoréacteur ont pu être réalisées, et confrontées aux résultats expérimentaux obtenus au laboratoire Pprime sur trois points de fonctionnement. Les variabilités cycle à cycle observées expérimentalement ont été analysées dans les calculs LES. Les vitesses importantes au niveau de l'allumage et le taux de résidus du cycle précédent semblent être les principaux facteurs à l'origine de ces variations cycle à cycle. / A major challenge for the aircraft industry is to improve engine efficiency and to reduce pollutant emissions for economic, technical and environmental reasons. Aeronautical gas turbines have enjoyed a constant improvement for more than 60 years. This technology seems to have reached such efficiency levels that a technological breakthrough is necessary. Constant Volume Combustion (CVC) offers significant gain in consumption and could replace classical constant pressure combustion technologies, currently used in aeronautical engines. Mechanisms involved in isovolume combustion are not accurately controlled in the context of aeronautical chambers. Experimental, theoretical and numerical studies should provide a better understanding of CVC devices. The objective of this thesis is to develop simulation tools to study the thermoreacteur concept. First, a zero-dimensional (0D) simulation tool is developed to describe the evolution of a CVC cycle. Models based on experimental correlations are used to build the 0D tool. Parameters have to be determined from numerical simulations. Today, the 3D simulation of a CVC system is possible thanks to the recent progress of the LES (Large Eddy Simulation) methods developed at CERFACS. Simulations of the thermoreacteur concept have been carried out, and compared to experimental results obtained at the Pprime laboratory. Three operating points have been calculated. The main conclusion is the existence of significant cyclic variations which are observed in the experiment and analyzed in the LES: the local flow velocity at spark timing and the level of residuals gases are the major factors leading to cyclic variations.

Les limitations récentes imposées par ICAO-CAEP, qui règlent les émissions de NOx, mènent à l’implémentation du concept de combustion lean dans les moteurs aéronautiques. Du point de vue du design, il faudrait étudier de façon plus approfondie la combustion lean et donc la Computational Fluid Dynamics(CFD) peut être un outil essentiel à ce but. Plusieurs phénomènes sont impliqués et différentes stratégies de modélisation, avec des différences en termes de coûts de calcul, sont disponibles. Néanmoins, jusqu'à présent, peu d'outils numériques peuvent prendre en compte les effets de la préparation du combustible liquide dans les calculs réactifs. Les conditions limites d’atomisation sont normalement déterminées par des approches corrélatives qui ne couvrent pas toutes les conditions de fonctionnement et les caractéristiques géométriques des brûleurs aéronautiques. Cependant, comme on peut lire dans la première partie de ma thèse, où plusieurs études de cas de littérature sont analysées, l'impact de la préparation du combustible liquide peut être extrêmement important. Les considérations basées sur des approches corrélatives ne sont pas fiables. Des méthodes prédictives focalisées sur l'atomisation du combustible sont nécessaires. Cette activité de recherche vise donc à développer un outil numérique général, utilisable dans le domaine industriel et capable de modéliser la phase liquide de son injection jusqu'à la génération d'un spray dispersé. Le modèle ELSA (Eulerian Lagrangian Spray Atomization), basé sur une approche eulérienne dans la région dense et une lagrangienne dans la zone diluée, a été choisi à ce but. Le solveur traite le combustible liquide jusqu'à la génération d'une phase dispersée et prend en compte le processus d’atomisation par l'introduction de la densité d'interface liquide-gaz. Néanmoins, si on applique cette méthode dans un environnement réactif fortement turbulent comme un brûleur aéronautique on peut rencontrer plusieurs limites. Par conséquent, on a dédié une attention particulière tout d’abord à l'étude du terme de flux turbulent à l'intérieur de l'équation de la fraction volumique liquide. Cette quantité est d'une importance primaire pour un écoulement turbulent, avec des vitesses de glissement élevées entre ses phases. Une nouvelle fermeture de second ordre pour cette variable est proposée et validée sur un cas de jet en crossflow. Ensuite, pour gérer un environnement réactif, un nouveau modèle d'évaporation est intégré dans le code et évalué par rapport aux résultats expérimentaux. Enfin, une autre méthode de dériver la distribution de la taille des gouttes dans le contexte ELSA pour l'injection lagrangienne est présentée et validée avec des simulations DNS. Pour conclure, ce travail introduit une nouvelle méthode pour une description unifiée de la combustion et de l’atomisation dans les calculs CFD. L'approche proposée devrait conduire à une description complète de l'évolution du combustible et à une caractérisation plus pertinente de l'écoulement réactif. Plusieurs aspects qui sont également mis en évidence sont améliorables et peuvent offrir des suggestions pour d’ultérieurs travaux. / The recent limitations imposed by ICAO-CAEP, regulating NOx emissions, are leading to the implementation of lean burn concept in the aero-engine framework. From a design perspective, a depth insight on lean burn combustion is required and Computational Fluid Dynamics (CFD) can be a useful tool for this purpose. Several interacting phenomena are involved and various modelling strategies, with huge differences in terms of computational costs, are available. Nevertheless, up to now few numerical tools are able to account for the effects of liquid fuel preparation inside reactive computations. Spray boundary conditions are normally determined thanks to correlative approaches that are not able to cover the wide range of operating conditions and geometrical characteristics of aero-engine burners. However, as highlighted in the first part of the dissertation, where several literature test cases are analysed through numerical calculations, the impact of liquid preparation can be extremely important. Considerations based on correlative approaches may be therefore unreliable. More trustworthy predictive methods focused on fuel atomization are required. This research activity is therefore aimed at developing a general numerical tool, to be used in an industrial design process, capable of modelling the liquid phase from its injection till the generation of a dispersed spray subject to evaporation. The ELSA (Eulerian Lagrangian Spray Atomization) model, which is based on an Eulerian approach in the dense region and a Lagrangian one in the dilute zone, has been chosen to this end. The solver is able to deal with pure liquid up to the generation of a dispersed phase and to account for the breakup process through the introduction of the liquid-gas interface density. However, several limitations of such method arise considering its application in a highly swirled reactive environment like an aero-engine burner. Therefore, particular attention has been here devoted first to the study of the turbulent liquid flux term, inside the liquid volume fraction equation. This quantity is of paramount importance for a swirled flow-field, with high slip velocities between phases. A completely innovative modelling framework together with a new second order closure for this variable is proposed and validated on a literature jet in crossflow test case. Then, to handle a reactive environment, a novel evaporation model is integrated in the code and assessed against experimental results. Finally, an alternative way to derive the Drop Size Distribution (DSD) in ELSA context for the lagrangian injection is presented and assessed by means of Direct Numerical Simulations. Ultimately, this work introduces an innovative framework towards a uni- fied description of spray combustion in CFD investigations. The proposed approach should lead to a comprehensive description of fuel evolution in the injector region and to a proper characterization of the subsequent reacting flow-field. Several improvable aspects are also highlighted, pointing the way for further enhancements.

Addressing global climate change will require increasing sustainable energy usage. Cofiring biomass fuels with coal for electrical power generation is an efficient, cost effective method of CO2 mitigation. This work is an experimental investigation of the flame structure and nitrogen chemistry differences occurring between coal, biomass and cofiring flames. A pilot-scale facility was fired with a dual-feed low-NOx burner capable of independently conveying 2 separate fuels unblended to the burner. Spatially detailed gas species measurements were made for 8 flames, including a coal, straw, finely ground straw, wood, and 4 straw/coal cofiring flames. Particle samples were also obtained from 5 of the flames. Intermittent flamelets were frequently observed in the flames. Viewing the substructure of the flame as individual flamelets provides critical insight for the interpretation of the data. The biomass and cofiring flames show larger flame volumes due to increased primary momentum, increased volatile yields, and differences in fuel particle characteristics (size and shape). The straw and cofiring flames also include secondary flame structures. The secondary flames result from delayed reaction of the straw “knees" due to differences in fuel characteristics. Biomass fuel-N was shown to evolve primarily through NH3, while the coal showed roughly equal amounts of NH3 and HCN. Due to increases in the flame volume and greater NH3 release within these larger fuel-rich regions, as well as lower fuel-N content, effluent concentrations of NO for the biomass and cofiring flames are lower than the coal flame. In-flame reduction of NO corresponds spatially to the presence of NH3, suggesting advanced reburning. Lower fuel-N contents are thought to increase the overall NO production efficiency, but this effect is uncertain for this work due to differences in flame structure and fuel-N chemistry. A mixing model based on intermittent flamelet behavior is included. The model uses dual-delta functions (DDF) to represent lean and rich eddies passing through a sampling volume. Both the beta-pdf and the DDF model were fit to data obtained in this study and compared. The beta-pdf model was unable to capture intermittent behavior. The DDF model was able to represent intermittent behavior, but produced physically unrealistic results.

No description available.

This present research effort was directed towards developing reduced order models for the dynamics of laminar flat flames, swirl stabilized turbulent flames, and in evaluating the effects of the variation in fuel composition on flame dynamics. The laminar flat flame study was conducted on instrument grade methane, propane, and ethane flames for four total flow rates from 145 cc/sec to 200 cc/sec, and five equivalence ratios from 0.5 to 0.75. The analysis was done by measuring the frequency resolved velocity perturbations, u', and the OH* chemiluminescence, as a measure of unsteady heat release rate, q'. The experimental data showed the corresponding flame dynamics to be fourth order in nature with a pure time delay. One of the resonance was shown to represent the pulsation of the flame location caused by fluctuation in the flame speed and fluctuating heat losses to the flame stabilizer. The other resonance was correlated to the dynamics of the chemical kinetics involved in the combustion process. The time delay was correlated to the chemical time delay. Upon comparing the results of the experiments with the three fuels, it was concluded that for all equivalence ratios studied, propane flame had a higher dynamic gain than methane flames. Ethane flames exhibited a higher dynamic gain than methane flame in the frequency range of 20-100 Hz. Thus, burning of propane instead of methane increased the likelihood of the occurrence of thermo-acoustic instabilities. The experimental techniques developed during the dynamic studies conducted on laminar flat flames were applied to swirl stabilized turbulent flames. Experiments were performed for QAir = 15 scfm and 20 scfm, F = 0.55, 0.6, 0.65, and S = 0.79 and 1.19. The results of fully premixed experiments showed that the flame behaved as a 8th order low pass filter. The results of the partially premixed experiment exhibited a rich spectra, which maintained its bandwidth over the entire range of frequency studied. Comparison of fully and partially premixed flames in the frequency range of 200-400 Hz, indicated that at overall lean conditions the dynamic gain of the totally premixed flames was almost an order of magnitude lower than that of the partially premixed conditions. Thus, it was concluded that combustors with fully premixed flames have a higher probability of being thermo-acoustically stable than those with partially premixed flames. / Ph. D.

Cette étude est consacrée à l'élaboration d'une méthodologie de détermination des vitessesfondamentales des flammes laminaires, en utilisant un diagnostic de Vélocimétrie par Imagerie deParticules (PIV). Ce dernier est appliqué aux écoulements réactifs avec point de stagnation, permettant lastabilisation de flammes planes, stationnaires et en conditions quasi adiabatiques. Les effets d’étirementssubits par la flamme sont également quantifiables et parfaitement maîtrisés. L’approche ici développée atout d’abord été appliquée aux mélanges méthane/air pour validation. Une comparaison exhaustive desrésultats obtenus avec les données de la littérature est effectuée. Les codes de combustion 1D (PREMIX,OPPDIF) et 2D (Fluent©) ont été utilisés afin de confirmer la fiabilité et la précision de l’approche proposée.Une attention particulière a été accordée à la caractérisation du mouvement des particules ensemencéesdans les écoulements réactifs divergents, avec notamment la prise en considération de la force dethermophorèse. La méthode développée a ensuite été appliquée à la détermination des vitesses deflammes laminaires de divers mélanges de syngas (H2+CO). Une étude comparative sur ces mélanges aété conduite en utilisant des approches expérimentales multiples comprenant : les flammes à contrecourant,les flammes à propagation sphérique ainsi que les flammes stabilisées coniques. Les résultatsobtenus pour chaque approche ont été confrontés et la sensibilité à l’étirement des flammes de syngas aété caractérisée pour une large gamme de richesses (E.R.=0.4 to 5.0) et de compositions de mélanges(5/95 to 50/50 % H2/CO). / In the context of CO2 emission reduction, the present study is devoted to the development of alaminar flame speed measurement methodology, using the Digital Particle Image Velocimetry (DPIV)diagnostic. The latter is applied to stagnation flow flames, seen to have considerable assets for suchstudies. Indeed, flames stabilized in these diverging flows are planar, steady and in near-adiabaticconditions, while subtraction of strain effects on flame is intrinsically allowed. The methodology developedherein has been applied to the well-characterized methane/air mixtures for validation. An extensivecomparison with the literature datasets has been provided. Both 1D (PREMIX, OPPDIF) as well as 2D(Fluent©) numerical tools have been used to confirm the reliability and accuracy of the developed approach.A particular attention has been given to the characterization of the seeding particle motion within thediverging flow, with consideration of the often-neglected thermophoretic force. Fundamental flame velocitiesof various syngas (H2+CO) mixtures have been investigated using multiple experimental approachesincluding the aforementioned counterflow methodology as well as spherical and conical flameconfigurations. Performed measurements from the different approaches have been confronted and flamesensitivities to stretch have been characterized for a wide range of equivalence ratios (E.R.=0.4 to 5.0) andmixture compositions (5/95 to 50/50 % H2/CO).

The work was concerned with experimental study of the turbulent flame development process of alcohol fuels, namely ethanol and butanol, in an optically accessed spark ignition research engine. The fuels were evaluated in a single cylinder engine equipped with full-bore overhead optical access operated at typical stoichiometric part-load conditions with images captured using high-speed natural light imaging techniques (or chemiluminescence). The differences in flame development between the fuels was analysed to understand better the impact of high and low alcohol content fuels on combustion. Advanced image analysis, in conjunction with Ricardo WAVE simulation, allowed for the conclusion that the faster burning exhibited by ethanol was the result of the marginally higher laminar burning velocity providing a faster laminar burn phase and accelerating the flame into the turbulent spectrum thus reducing bulk flame distortion and better in-cylinder pressure development. Such physical reactions are often over-looked in the face of chemical differences between fuels. A further study into the variation of maximum in-cylinder pressure values was conducted focussing on iso-octane and ethanol. This study identified two phenomena, namely “saw-toothing” and “creep” in which cluster of cycles feed into one another. From this it became clear that the presence of high pressure during the exhaust process had a large influence on the following cycles. This is another often overlooked phenomenon of direct cycle-to-cycle variation whereby incylinder pressures during blowdown can dictate the duration, load or stability output of the following cycle. Finally the work investigated the impact on flame development of alcohol fuels when the overlap duration was altered. While the engine produced counterintuitive figures of residual gas, ethanol was confirmed as having greater synergy with EGR by displaying less impacted combustion durations c.f. iso-octane. Care should be taken however when analysing these results due to the unique valve configuration of the engine.

No description available.

An experimental study was conducted using axisymmetric co-flow laminar diffusion flames of methane-air, methane-oxygen and ethylene-air to examine the effect of pressure on soot formation and the structure of the temperature field. A liquid fuel burner was designed and built to observe the sooting behavior of methanol-air and n-heptane-air laminar diffusion flames at elevated pressures up to 50 atm. A non-intrusive, line-of-sight spectral soot emission (SSE) diagnostic technique was used to determine the temperature and the soot volume fraction of methane-air flames up to 60 atm, methane-oxygen flames up to 90 atm and ethylene-air flames up to 35 atm. The physical flame structure of the methane-air and methane-oxygen diffusion flames were characterized over the pressure range of 10 to 100 atm and up to 35 atm for ethylene-air flames. The flame height, marked by the visible soot radiation emission, remained relatively constant for methane-air and ethylene-air flames over their respected pressure ranges, while the visible flame height for the methane-oxygen flames was reduced by over 50 % between 10 and 100 atm. During methane-air experiments, observations of anomalous occurrence of liquid material formation at 60 atm and above were recorded. The maximum conversion of the carbon in the fuel to soot exhibited a strong power-law dependence on pressure. At pressures 10 to 30 atm, the pressure exponent is approximately 0.73 for methane-air flames. At higher pressures, between 30 and 60 atm, the pressure exponent is approximately 0.33. The maximum fuel carbon conversion to soot is 12.6 % at 60 atm. For methane-oxygen flames, the pressure exponent is approximately 1.2 for pressures between 10 and 40 atm. At pressures between 50 and 70 atm, the pressure exponent is about -3.8 and approximately -12 for 70 to 90 atm. The maximum fuel carbon conversion to soot is 2 % at 40 atm. For ethylene-air flames, the pressure exponent is approximately 1.4 between 10 and 30 atm. The maximum carbon conversion to soot is approximately 6.5 % at 30 atm and remained constant at higher pressures.

The thesis addresses the influence of compressible effects on the stability of deflagration waves. Due to the quasi-isobaric nature of slow flames, compressible effects in laminar flames are usually neglected. Nevertheless, turbulent deflagrations may propagate at substantially higher speeds, suggesting that compressible effects may play a role in their dynamics. In the present thesis, the stability of diffusion-dominated high-speed deflagrations is addressed. The deflagration is studied in the thickened regime, hence addressing the long wavelength limit of these deflagrations. The deflagrations are modelled by the compressible reactive Navier-Stokes equations with a single-step Arrhenius reaction model. The 2D stability of the steady traveling-wave solution is studied by direct simulation. It is found that, as the flame compressibility becomes significant, not only does the growth rates of the cellular profile of the deflagration waves increase, but the traditional correlation of the burning velocity and the flame surface area become far less strong. Significant compression regions form in front of the nonlinear flames. These compression regions have been found to increase the growth rates by increasing the temperature of the unburned gas in front of the flames, as well as convecting the flame forward. The results show that the flame propagation velocity in references to the unburned gas was significantly faster than the burning velocity. The vorticity was given consideration, as the compressibility of flame increase one can expect the baroclinic source to be of greater significance. The vorticity was show to, in effect, increase as compressibility increases while unexpectedly having a stabilizing direction of rotation on the cellular structure of the flames.