Tabulated chemical kinetics for efficient and detailed simulations of diesel engine combustion

Efficient and detailed computational tools to simulate engine combustion are of great importance. The internal combustion engine will remain the primary mean for transportation in the decades to come. Especially diesel engines are, and become increasingly more, popular because of their high fuel-efficiency. Unfortunately they produce relatively high amounts of soot and NOx. Although the emissions from combustion engines have decreased dramatically (orders of magnitude) in the past decades, legislation continues to impose ever more demanding limits. Computational tools could help to rapidly design high-efficiency and low-emission engines. However, the computational effort for the needed detailed simulations is prohibitive. One of the major challenges is to accurately model the chemical reactions in an efficient manner. Chemistry reduction methods have been introduced to reduce computational costs while preserving the essentials of chemical kinetics. The Flamelet Generated Manifold (FGM) technique developed at Eindhoven University of Technology is a tabulation approach which has been applied for the simulation of different laminar and statistically-steady turbulent flames before. FGM is especially appealing for the simulation of combustion devices such as diesel engines because it is known for lowering computational costs tremendously and at the same time giving an accurate description of chemistry. Therefore, the central question in this thesis is how to include vital engine combustion phenomena, such as ignition, in a generic way to make FGM a useful tool for diesel engines. In this thesis, the FGM method is raised to a level at which it can be applied to simulate the complex processes in diesel engines. The motivation and scope are given in the general introduction in Chapter 1, and the proposed approach is described in Chapter 2. The novel extensions are developed and validated systematically in conjunction with a variety of Computational Fluid Dynamics (CFD) approaches as presented in the subsequent chapters. First, the small scale processes are examined in simplified systems. Gradually, more of the large scale physics are included towards device scale simulations.Developments in modern engine technology are moving towards a regime with fuel injection uncoupled from combustion. Auto-ignition is an essential characteristic in these systems. The accurate prediction of this chemical process is of paramount importance. In a tabulation method, a canonical combustion configuration that best represents the local chemical process in a multidimensional system is chosen. Each canonical configuration assumes a typical mixing condition which may be different from the mixing in the multidimensional case. Since only chemistry information is stored in the table, the mixing model in the canonical configuration may introduce errors. Therefore, in Chapter 3 the validity of the FGM method for ignition tabulation at mixing conditions that are relevant for engines is investigated with Direct Numerical Simulations (DNS) of 0D, 1D and 2D igniting systems. An FGM table generated with homogeneous reactor simulations is able to correctly predict reaction progress in perfectly stirred reactor cases which include mixing. On the other hand, an FGM table generated with a single igniting counterflow diffusion flame at a constant strain rate predicts the trend in auto-ignition delay for varying strain rates qualitatively correct. Ignition in a 2D mixing layer, where a straining field due to vortical structures exists, is also well predicted with this FGM. To improve the quantitative auto-ignition prediction near the ignition limit, an extra controlling variable is needed. In the next step, in Chapter 4 the method is applied to Large-Eddy Simulations (LES) of a diesel spray in a constant volume. This three-dimensional turbulent simulation setup is the typical target application. The question is whether the developed approach is able to predict macroscopic characteristics that are affected by chemistry. The focus is set on the ignition delay and the quasi-steady lift-off distance of the flame. Furthermore, the effects of models for turbulence-chemistry interaction are discussed. To this purpose, Flamelet Generated Manifolds are constructed using (igniting) counter-flow diffusion flames computed with two different reaction mechanisms. The chemistry is parameterized as a function of the mixture fraction and a reaction progress variable. LES of an extensive set of igniting n-heptane liquid sprays are performed of which well-documented experiments exist in literature (Engine Combustion Network). The main spray characteristics, spray penetration depth, ignition delay time, and flame lift-off length, are compared to measured values. They correspond very well for varying ambient oxygen content and ambient temperature. The results show that the ignition delay trend is captured with an igniting laminar flame in the chemistry table and that subgrid closure assumptions can have a large influence on the results. The final intended application is the simulation of a full engine with moving pistons. The temporal variation of in-cylinder pressure and temperature due to piston motion greatly impact all processes, thus also combustion. The conditions at the start of injection are considerably different from the conditions at the start of ignition. The pressure/temperature evolution in between influences the ignition delay itself. To take this phenomenon into account, in Chapter 5 pressure has been added to the FGM as an additional controlling variable next to mixture fraction and progress variable. An FGM table has been generated from igniting diffusion flames computed at certain expected pressure levels (and the corresponding approximated in-cylinder bulk temperature). The obtained pressure dependent tables have been applied to Reynolds averaged Navier-Stokes (RANS) simulations of a moving piston section. The study shows that a moderate amount of pressure levels in the FGM database are sufficient to represent the effect of the large pressure variation during compression and combustion on FGM tabulation. Finally, in Chapter 6 the achievements are summarized and suggestions for major improvement points are formulated. Overall, FGM’s good performance at each stage of model development and examination demonstrates the potential of the method to become a powerful tool in designing clean and fuel-efficient compression ignition engines