Cyclic Analysis of In-Cylinder Vortex Interactions Based on Data-Driven Detection and Characterization Framework

The complex vortex flow interactions are critical to affect the fuel-air mixing and combustion stability in direct-injection engine. However, due to the strong cyclic variations inside engine, the multiscale swirl flow characteristics with cyclic details are difficult to be sufficiently revealed. Therefore, a vortex detection and characterization framework, including physical and data-driven methods, is implemented to elucidate the cyclic vortex interaction process. In this study, a high-speed time-resolved particle image velocimetry is applied to record the spatiotemporal flow behavior under three different swirl ratio conditions. First, the presence of vortex motion is detected at each crank angle for each engine cycle. Results show that the vortex interaction processes under different swirl ratio conditions exhibit distinctive characteristics. The presence of multiple vortices and their interactions are found to trigger dramatic changes and variations in swirl flow behavior. Then, the individual-cycle analysis of the vortex interaction effects on flow characteristics is conducted. The vortex characteristics including vortex location, strength, and size are examined with cyclic detail using data-driven unsupervised clustering. Results indicate that the vortex merging is the main source inducing the vortex characteristics variations. Furthermore, the occurrence and duration of the vortex merging process are found to be closely related to the intake swirl ratio and valve lift profile. Increased swirl ratio and valve lift cause vortex to merge earlier and reduce the merging duration. This finding provides a potential idea to alleviate the cyclic variation issue by controlling the vortex merging process.