Investigation of In-Cylinder Engine Flow Quadruple Decomposition Dynamical Behavior Using Proper Orthogonal Decomposition and Dynamic Mode Decomposition Methods

Abstract

In order to study the in-cylinder flow characteristics, one hundred consecutive cycles of velocity flow fields were investigated numerically by large eddy simulation, and the proper orthogonal decomposition (POD) algorithm was used to decompose the results. The computed flow fields were divided into four reconstructed parts, namely mean part, coherent part, transition part, and turbulent part. Then, the dynamic mode decomposition (DMD) algorithm was used to analyze the characteristics of the reconstructed fields. The results show that DMD method is capable of finding the dominant frequencies in every reconstructed flow part and identifying the flow structures at equilibrium state. In addition, the DMD results also reveal that the reconstructed parts are related to each other through the break-up and attenuation process of unstable flow structures, while the flow energy cascade occurs among these parts through different scale vortex generation and dissipation process.