Investigation and modeling of the early flame kernel development in hydrogen spark-ignition engines

点击翻译按钮获取中文摘要

The transition from the ignited laminar flame kernel to turbulent flames is crucial for cycle-to-cycle variations in spark ignition engines. For mixtures with non-unity Lewis numbers, this transition can be significantly affected by differential diffusion effects due to the strong curvature of the small flame kernel. In the second funding period, hydrogen will be used as a fuel. In contrast to the iso-octane flames studied in the first funding period, which have an effective Lewis number larger than unity, lean hydrogen flames have an effective Lewis number lower than unity. As a result, in contrast to iso-octane, for hydrogen, the local burning rate in strongly convexly curved flame kernels increases due to differential diffusion. This can lead to thermodiffusive instabilities, and thus a significant increase in turbulent flame speed. Therefore, the transition from laminar flame kernels to turbulent flames is significantly different for hydrogen flames compared to iso-octane. This subproject aims to investigate and model the cycle-to-cycle variations of early flame propagation in hydrogen-fueled engines. For this purpose, Direct Numerical Simulations (DNS) of the early flame kernels as well as Large Eddy Simulations (LES) of the hydrogen-fueled engine in TP 1 will be performed. DNS will be performed for a simplified engine geometry, but for the first time under engine-relevant conditions considering the spark plug, interactions of the spark channel with the flow, and a mean velocity caused by the tumble motion. Simulations will first be carried out for a homogeneous fuel distribution using conditions derived from the experiments in TP 1. Using the mixture distribution conditions obtained from the experiments in TP 3 and the LES in TP 5, DNS with inhomogeneous fuel distribution will also be performed. Furthermore, 2D-DNS of auto-ignition in hydrogen/air mixtures will be carried out, which will be used by TP 7 for knock modeling. The resulting data will be compared with measurements from TP 1 and TP 3. Investigations and modelling of the interactions between the intrinsic instabilities and turbulence, as well as the influences of relative velocity and spark plug, will be carried out based on the optimal estimator and dissipation element analysis. DNS data with inhomogeneous fuel distribution will be used by TP 5 for model development. Here, the influence of the mixture inhomogeneities on the intrinsic instabilities is of particular interest. Models will be validated in LES of the engine in TP 1. Backward analysis of the cycle-to-cycle variations will be performed in TP 1 using LES and experiments.