Multi-plane time-resolved Particle Image Velocimetry (PIV) flow field measurements in an optical Spark-Ignition Direct-Injection (SIDI) engine for Large-Eddy Simulation (LES) model validations

Abstract

In-cylinder flow characteristics play a significant role in the fuel–air mixing process of Spark-Ignition Direct-Injection (SIDI) engines. Typically, planar Particle Image Velocimetry (PIV) is used to measure a representative velocity field sectioning through the center plane of the engine cylinder. However, a single flow field offers very limited perspective regarding the Three-Dimensional nature of the flow fields. Since the in-cylinder flow is stochastically complex, large datasets of flow field measurements along multiple planes are needed to provide a complete panoramic understanding of the flow dynamics. In this study, a high-speed PIV is applied to measure the crank-angle resolved flow fields inside a single-cylinder four-valve optical SIDI engine. Five flow fields along different tumble planes are captured. These five planes include two orthogonal planes cutting through the spark plug tip, two parallel planes sectioning through middle point of the intake and exhaust valves, and one plane through the centers of two intake valves. In addition, numerical computations are carried out with Large-Eddy Simulation (LES) model in CONVERGE. With the guidance from multi-plane PIV measurements, a novel validation approach is proposed in this study. The quantitative analysis and comparison between LES simulations and PIV experiments are divided in terms of global and local comparison indices. The global comparison indices provide a quantitative single value to quickly check the overall similarity of velocity directions and magnitudes between PIV and LES results of a specific individual plane. The local comparison indices further evaluate the similarity between the flow fields of LES and PIV point by point to identify any dissimilar regions and vortex features, which are likely to indicate the complex flow structures at low-speed regions. In summary, not only can the combined data analysis approach provide a reliable way for LES model validations, it can also reveal the physical quantifications of the complex in-cylinder flow characteristics.