Modern spark ignition engines are downsized and boosted to meet stringent emission standards and growing customer demands on performance and fuel economy. They operate under high intake pressures and close to their limits to engine knock. As the intake pressure is increased knock becomes the major barrier that prevents further improvement on downsized boosted spark ignition engines. It is generally accepted that knock is caused by end gas autoignition ahead of the propagating flame. The propagating flame front has been identified as one of the most influential factors that promote the occurrence of autoignition. Systematic understanding and numerical relation between the propagating flame front and the occurrence of knock are still lacking. Additionally, knock mitigation strategy that minimizes compromise on engine performance needs further researching. Therefore the objectives of the current research consist of two steps: 1). study of turbulent flame propagation in both naturally aspirated SI engine. 2) study of the relationship between flame propagation and the occurrence of engine knock for downsized and boosted SI engine. The aim of the current research is, firstly, to find out how turbulent flames propagate in naturally aspirated and boosted S.I. engines, and their interaction with the occurrence of knock; secondly, to develop a mitigation method that depresses knock intensity at higher intake pressure. Autoignition of hydrocarbon fuels as used in spark ignition engines is a complex chemical process involving large numbers of intermediate species and elementary reactions. Chemical kinetics models have been widely used to study combustion and autoignition of hydrocarbon fuels. Zero-dimensional multi-zone models provide an optimal compromise between computational accuracy and costs for engine simulation. Integration of reduced chemical kinetics model and zero-dimensional three-zone engine model is potentially a effective and efficient method to investigate the physical, chemical, thermodynamic and fluid dynamic processes involved in in-cylinder turbulence flame propagation and knock. The major contributions of the current work are made to new knowledge of quantitative relations between intake pressure, turbulent flame speed, and knock onset timing and intensity. Additionally, contributions have also been made to the development of a knock mitigation strategy that effectively depresses knock intensity under higher intake pressure while minimizes the compromise on cylinder pressure, which can be directive to future engine design.

Rare events have very low probability of occurrence but can have significant impact. Earthquakes, volcanoes, and stock market crashes can have devastating impact on those affected. In industry, engineers evaluate rare events to design better high-reliability systems. The objective of this work is to increase efficiency in design optimization for rare events using metamodeling and variance reduction techniques. Opportunity exists to increase deterministic optimization efficiency by leveraging Design of Experiments to build an accurate metamodel of the system which is less resource intensive to evaluate than the real system. For computationally expensive models, running many trials will impede fast design iteration. Accurate metamodels can be used in place of these expensive models to probabilistically optimize the system for efficient quantification of rare event risk. Monte Carlo is traditionally used for this risk quantification but variance reduction techniques such as importance sampling allow accurate quantification with fewer model evaluations. Metamodel techniques are the thread that tie together deterministic optimization using Design of Experiments and probabilistic optimization using Monte Carlo and variance reduction. This work will explore metamodeling theory and implementation, and outline a framework for efficient deterministic and probabilistic system optimization. The overall conclusion is that deterministic and probabilistic simulation can be combined through metamodeling and used to drive efficiency in design optimization. Applications are demonstrated on a gas turbine combustion autoignition application where user controllable independent variables are optimized in mean and variance to maximize system performance while observing a constraint on allowable probability of a rare autoignition event.

Chemical kinetics has been widely acknowledged as a fundamental theory in analysis of chemical processes and the corresponding reaction outputs and rates. The study and application of chemical kinetics thus provide a simulation tool to predict many characteristics a chemical process. Oxidation of hydrocarbon fuels applied in internal combustion engines is a complex chemical process involving a great number of a series of chained reaction steps and intermediate and simultaneous species. Symbolic and Numerical description of such a chemical process leads to the development and application of chemical kinetics models. The up-to-date application of chemical kinetics models is to the simulation of autoignition process in internal combustion engines. Multi-zone thermodynamic combustion modelling has been regarded as a functional simulation approach to studying combustion process in IC engines as a decent compromise between computation accuracy and efficiency. Integration of chemical kinetics models into multi-zone models is therefore a potential modelling method to investigate the chemical and physical processes of autoignition in engine combustion. This research work has been therefore concerned with the development, validation and application of multi-zone chemical kinetic engine models in the simulation of autoignition driven combustion in SI and HCCI engines. The contribution of this work is primarily made to establish a mathematical model based on the underlying physical and chemical principles of autoignition of the fuel-air mixture in SI and HCCI engines. Then, a computer code package has been developed to numerically solve the model. The derived model aims at improving the understanding of autoignition behaviour under engine-like conditions and providing an investigative tool to autoignition characteristics. Furthermore, as part of the ongoing program in the research of free piston engines, the results of this work will significantly aid in the investigation and simulation of the constant volume autoignition applied in free piston engines.

Practical gas turbine engine combustors create extremely non-uniform flowfields, which are highly stratified making it imperative that similar environments are well understood. Laser diagnostics were utilized in a variety of stratified environments, which led to temperature or chemical composition gradients, to better understand autoignition, extinction, and flame stability behavior. This work ranged from laminar and steady flames to turbulent flame studies in which time resolved measurements were used. Edge flames, formed in the presence of species stratification, were studied by first developing a simple measurement technique which is capable of estimating an important quantity for edge flames, the advective heat flux, using only velocity measurements. Both hydroxyl planar laser induced fluorescence (OH PLIF) and particle image velocimetry (PIV) were used along with numerical simulations in the development of this technique. Interacting triple flames were also created in a laboratory scale burner producing a laminar and steady flowfield with symmetric equivalence ratio gradients. Studies were conducted in order to characterize and model the propagation speed as a function of the flame base curvature and separation distance between the neighboring flames. OH PLIF, PIV and Rayleigh scattering measurements were used in order to characterize the propagation speed. A model was developed which is capable of accurately representing the propagation speed for three different fuels. Negative edge flames were first studied by developing a one-dimensional model capable of reproducing the energy equation along the stoichiometric line, which was dependent on different boundary conditions. Unsteady and laminar negative edge flames were also simulated with periodic boundary conditions in order to assess the difference between the steady and unsteady cases. The diffusive heat loss was unbalanced with the chemical heat release and advective heat flux energy gain terms which led to the flame proceeding and receding. The temporal derivative balanced the energy equation, but also aided in the understanding of negative edge flame speeds. Turbulent negative edge flame velocities were measured for extinguishing flames in a separate experiment as a function of the bulk advective heat flux through the edge and turbulence level. A burner was designed and built for this study which created statistically stationary negative edge flames. The edge velocity was dependent on both the bulk advective heat flux and turbulence levels. The negative edge flame velocities were obtained with high speed stereo-view chemiluminescence and two dimensional PIV measurements. Autoignition stabilization was studied in the presence of both temperature and species stratification, using a simple laminar flowfield. OH and CH2O PLIF measurements showed autoignition characteristics ahead of the flame base. Numerical chemical and flow simulations also revealed lower temperature chemistry characteristics ahead of the flame base leading to the conclusion of lower temperature chemistry dominating the stabilization behavior. An energy budget analysis was conducted which described the stabilization behavior.

Autoignition of non-premixed methane-air mixtures is investigated using first-order Conditional Moment closure (CMC). In CMC, scalar quantities are conditionally averaged with respect to a conserved scalar, usually the mixture fraction. The conditional fluctuations are often of small order, allowing the chemical source term to be modeled as a function of the conditional species concentrations and the conditional enthalpy (temperature). The first-order CMC derivation leaves many terms unclosed such as the conditional scalar dissipation rate, velocity and turbulent fluxes, and the probability density function. Submodels for these quantities are discussed and validated against Direct Numerical Simulations (DNS). The CMC and the turbulent velocity and mixing fields calculations are decoupled based on the frozen mixing assumption, and the CMC equations are cross-stream averaged across the flow following the shear flow approximation. Finite differences are used to discretize the equations, and a two-step fractional method is implemented to treat separately the stiff chemical source term. The stiff ODE solver LSODE is used to solve the resulting system of equations. The recently developed detailed chemical kinetics mechanism UBC-Mech 1.0 is employed throughout this study, and preexisting mechanisms are visited. Several ignition criteria are also investigated. Homogeneous and inhomogeneous CMC calculations are performed in order to investigate the role of physical transport in autoignition. Furthermore, the results of the perfectly homogeneous reactor calculations are presented and the critical value of the scalar dissipation rate for ignition is determined. The results are compared to the shock tube experimental data of Sullivan et al. The current results show good agreement with the experiments in terms of both ignition delay and ignition kernel location, and the trends obtained in the experiments are successfully reproduced. The results were shown to be sensitive to the scalar dissipation model, the chemical kinetics, and the ignition criterion.

Autoignition of non-premixed methane-air mixtures is investigated using first-order Conditional Moment closure (CMC). In CMC, scalar quantities are conditionally averaged with respect to a conserved scalar, usually the mixture fraction. The conditional fluctuations are often of small order, allowing the chemical source term to be modeled as a function of the conditional species concentrations and the conditional enthalpy (temperature). The first-order CMC derivation leaves many terms unclosed such as the conditional scalar dissipation rate, velocity and turbulent fluxes, and the probability density function. Submodels for these quantities are discussed and validated against Direct Numerical Simulations (DNS). The CMC and the turbulent velocity and mixing fields calculations are decoupled based on the frozen mixing assumption, and the CMC equations are cross-stream averaged across the flow following the shear flow approximation. Finite differences are used to discretize the equations, and a two-step fractional method is implemented to treat separately the stiff chemical source term. The stiff ODE solver LSODE is used to solve the resulting system of equations. The recently developed detailed chemical kinetics mechanism UBC-Mech 1.0 is employed throughout this study, and preexisting mechanisms are visited. Several ignition criteria are also investigated. Homogeneous and inhomogeneous CMC calculations are performed in order to investigate the role of physical transport in autoignition. Furthermore, the results of the perfectly homogeneous reactor calculations are presented and the critical value of the scalar dissipation rate for ignition is determined. The results are compared to the shock tube experimental data of Sullivan et al. The current results show good agreement with the experiments in terms of both ignition delay and ignition kernel location, and the trends obtained in the experiments are successfully reproduced. The results were shown to be sensitive to the scalar dissipation model, the chemical kinetics, and the ignition criterion.

In the present work, the processes of steady combustion and autoignition of hydrogen are investigated using the Conditional Moment Closure (CMC) model with a Reynolds Averaged Navier-Stokes (RANS) Computational Fluid Dynamics (CFD) code. A study of the effects on the flowfield of changing turbulence model constants, specifically the turbulent Schmidt number, Sct, and C epsilon 1 of the k − epsilon model, are investigated. The effects of two different mixing models are explored: the AMC model, which is commonly used in CMC implementations, and a model based on the assumption of inhomogeneous turbulence. The background equations required for implementation of the CMC model are presented, and all relevant closures are discussed. The numerical implementation of the CMC model, in addition to other techniques aimed at reducing computational expense of the CMC calculations, are provided. The CMC equation is discretised using finite volume (FV) method. The CFD and CMC calculations are fully coupled, allowing for simulations of steady flames or flame development after the occurrence of autoignition. Through testing of a steady jet flame, it is observed that the flowfield calculations follow typical k − epsilon model trends, with an overprediction of spreading and an underprediction of penetration. The CMC calculations are observed to perform well, providing good agreement with experimental measurements. Autoignition simulations are conducted for 3 different cases of turbulence constants and 7 different coflow temperatures to determine the final effect on the steady flowfield. In comparison to the standard constants, reduction of Sct results in a reduction of the centreline mixing intensity within the flowfield and a corresponding reduction of ignition length, while reducing C 1 results in an increase of centreline mixing intensity and an increase in the ignition length. All scenarios tested result in an underprediction of ignition length in comparison to experimental results; however, good agreement with the experimental trends is achieved. At low coflow temperatures, the effects of mixing intensity within the flowfield are seen to have the largest influence on ignition length, while at high coflow temperatures, the chemical source term in the CMC equation increases in magnitude, resulting in very little difference between predictions for different sets of turbulence constants. The inhomogeneous mixing model is compared using the standard turbulence constants. A reduction of ignition lengths in comparison to the AMC model is observed. In steady state simulation of the autoigniting flow, the inhomogeneous model is observed to predict both lifted flames and fully anchored flames, depending on coflow temperature.

In the present work, the processes of steady combustion and autoignition of hydrogen are investigated using the Conditional Moment Closure (CMC) model with a Reynolds Averaged Navier-Stokes (RANS) Computational Fluid Dynamics (CFD) code. A study of the effects on the flowfield of changing turbulence model constants, specifically the turbulent Schmidt number, Sct, and C epsilon 1 of the k − epsilon model, are investigated. The effects of two different mixing models are explored: the AMC model, which is commonly used in CMC implementations, and a model based on the assumption of inhomogeneous turbulence. The background equations required for implementation of the CMC model are presented, and all relevant closures are discussed. The numerical implementation of the CMC model, in addition to other techniques aimed at reducing computational expense of the CMC calculations, are provided. The CMC equation is discretised using finite volume (FV) method. The CFD and CMC calculations are fully coupled, allowing for simulations of steady flames or flame development after the occurrence of autoignition. Through testing of a steady jet flame, it is observed that the flowfield calculations follow typical k − epsilon model trends, with an overprediction of spreading and an underprediction of penetration. The CMC calculations are observed to perform well, providing good agreement with experimental measurements. Autoignition simulations are conducted for 3 different cases of turbulence constants and 7 different coflow temperatures to determine the final effect on the steady flowfield. In comparison to the standard constants, reduction of Sct results in a reduction of the centreline mixing intensity within the flowfield and a corresponding reduction of ignition length, while reducing C 1 results in an increase of centreline mixing intensity and an increase in the ignition length. All scenarios tested result in an underprediction of ignition length in comparison to experimental results; however, good agreement with the experimental trends is achieved. At low coflow temperatures, the effects of mixing intensity within the flowfield are seen to have the largest influence on ignition length, while at high coflow temperatures, the chemical source term in the CMC equation increases in magnitude, resulting in very little difference between predictions for different sets of turbulence constants. The inhomogeneous mixing model is compared using the standard turbulence constants. A reduction of ignition lengths in comparison to the AMC model is observed. In steady state simulation of the autoigniting flow, the inhomogeneous model is observed to predict both lifted flames and fully anchored flames, depending on coflow temperature.

To contribute to the design of next-generation high performance and low emission combustion devices, this study provides a series of high fidelity numerical simulations of turbulent premixed combustion and autoignition with different clean fuels. The first part of the thesis consists of the direct numerical simulations (DNS) of the lean hydrogen-air turbulent premixed flames at a wide range of Karlovitz number (Ka) conditions up to Ka = 1,126. Turbulence-chemistry interaction is discussed in terms of statistical analysis of the turbulent flame speed and flame structure. Global and local flame speed are separately studied through the fuel consumption speed and displacement speed of the flame front, respectively, and the results are compared with the reference laminar flames as well as similar studies in the literature. The global flame structure is assessed via cross-sectional and conditional averages, and modeling implication is further discussed. Detailed analysis of the local flame structure along the positive and negative curvature is also conducted, providing an understanding of the different behavior of local heat release response. Finally, as the modeling perspectives for Reynolds-averaged Navier-Stokes (RANS) and large eddy simulations (LES), the mean quantities of major species, intermediate species, density, the reaction rate of the progress variable, and heat release rate are assessed in the context of the probability density function (PDF). The second part of the thesis consists of applications of the advanced mathematical tool called the computational singular perturbation (CSP). A skeletal chemical mechanism is developed using the CSP algorithm for the autoignition of methanol and dimethyl ether blends, and the ignition delay time and laminar flame speed are validated for a wide range of mixture conditions. A series of autoignition simulations are carried out in the canonical counter flow mixing layer using the developed skeletal mechanism, and detailed analyses of the autoignition for the methanol and dimethyl ether blends at a wide range of strain rate conditions are provided using the CSP diagnostics tools for a wide range of chemical and fluid combinations.

No description available.