Large Eddy Simulation And POD Analysis Of In-cylinder Turbulent Flow And Its Cycle-To-Cycle Variations In IC Engines

The in-cylinder turbulent flow of IC engine is one of the most complex among various turbulent motions. At an operating condition, the in-cylinder gas charge motion of the engine is always extremely variable and highly transient, which not only exhibits a highly unsteady state within a single cycle, but also has a high degree of randomness and variations from cycle to cycle. The in-cylinder flow field acts as a common environment for the various physical and chemical sub-processes throughout the engine combustion process, and it strongly affects the fuel economy and emission levels of the engine. The rise of the Large Eddy Simulation method (LES) broke through limitations on the conventional engine simulation. As a powerful tool, it can objectively capture transient details of the flow field for people to correctly understand the in-cylinder turbulent phenomena. By using the LES method, this dissertation has investigated in-cylinder turbulence characteristics, including transient turbulent fluctuations, coherent structures and cycle-to-cyclic variations, in order to gain more and deeper insights into characteristics of the in-cylinder turbulent flow The addressed issues and conducted work in this dissertation is as follows:First, study on the in-cylinder turbulent fluctuations.1. A dynamic Smagorinsky model was added in the KIVA-3V program to expand the original LES calculation module. Then a number of LES sub-grid models were authenticated through a cold flow test using a simplified cylinder model, and the results were compared with the one calculated by using Reynolds-averaged RNG model. In addition, the influences of grid density and numerical difference scheme on the LES performance were considered. The results show that the LES simulation is significantly more effective than the traditional Reynolds-averaged approach. Both the large scale bulk flow and the local small scale vortices random fluctuations can be effectively described. By comparison, the dynamic Smagorinsky model and sub-grid K equation model show a better performance. In addition, high-precision differential schemes and high-density grid are more conducive to the advantage of the LES method.In this study, the LES sub-grid model combined with the employed computational grid system can capture more than90%kinetic energy of the engine flow field, which can ensure the effectiveness of the simulation in the case of real IC engine.2. The intake and compression processes in an actual IC engine have been simulated by the LES method. The bulk flow field behavior, details of the local small vortices fluctuation, and energy evolution as well as relevant macroscopic flow parameters were investigated and directly compared with experimental results. It is found that the intake jet is the main dominant factor for the in-cylinder turbulent field, followed by the cylinder wall and the piston friction effects. Along with the intake jet intensity declining, the turbulent fluctuations are weakened, and the flow field anisotropy degree is declined as well. In addition, the engine speed can influence the cylinder turbulent field significantly with nearly proportional relations to the turbulent kinetic energy, energy dissipation and turbulent eddy viscosity.Second, study on the coherent structures.3. The LES method combined with the Q-criterion for vortex identification has been used to study the characteristics of coherent structures in the in-cylinder turbulent flow. By comparing with Reynolds-averaged model calculation, it is found that the LES method can factually reflect more details of in-cylinder flows, it can not only describe the evolution of large scale coherent structures, but also be able to capture the transient random motion of small scale vortices. The coherent structures can experience a series revolution:generation, development, stretch and breaking, which has a strong correlation with the evolution of the total in-cylinder turbulent kinetic energy. This indicates that the presence and evolution of the coherent structures play a key role for the evolution of the in-cylinder turbulent energy generation and the maintenance of the turbulent motion.Finally, study on cycle-to-cycle variations (CCVs)4. In-cylinder cold flow field of an actual IC engine was calculated, and100continuous cycles were simulated by LES and directly compared with the PIV measurement in order to study the in-cylinder turbulent flow CCV characteristics. The results show that the turbulent fluctuation and CCVs can be identified and distinguished effectively by using the LES method. Both phenomena have a strong positive correlation with the intake jet intensity, and the CCVs intensity is stronger than the turbulent fluctuations. In the early intake stroke, the flow field shows both intensive turbulent fluctuation and intensive CCV. But in the following stage, the flow field shows still intensive CCV but much weaker turbulent fluctuation, thus, the relative intensity keeps below15%. Some flow field macroscopic parameters, such as the swirl ratio and tumble ratio, also show strong cyclic variation characteristics. Besides that, the CCV intensity shows a nearly proportional relation to the engine speed, and is strongly influenced by the maximum valve lift (MVL). With decreasing MVL, the CCV intensity become stronger. In addition, by using the phase average method, it is confirmed that at least25sample cycles for the mean velocity and50cycles for the RMS velocity are necessary for obtaining converged and correct results in the CCV.5. Databases of multi-cycle in-cylinder turbulent fields from both simulation and PIV measurement were processed by the proper orthogonal decomposition method (POD), the original transient flow field is decomposed into a mean flow field, a coherent structure flow field, a transition flow field and a turbulent flow field by the POD quadruple decomposition proposed in this study. Then the four part field evolutions as well as CCVs are studied in a detail. The result shows that each part field not only has its own unique characteristics, but also has a close relationship with each other. Among the four parts, the CCVs of the coherent structure flow field play a key role in the whole in-cylinder turbulent flow field, while the effect of the turbulent flow field cycle variation can be neglected.