Cluster-based tracking method for the identification and characterisation of vortices

Abstract

An unsupervised, flow-agnostic and automatic cluster-based tracking algorithm for the segmentation of vortex-dominated flows has been successfully developed. It combines the Rortex method and density-based clustering algorithms. The Rortex method differs shear from rotation and overcomes the sensitivity to user-defined thresholds that characterises current practice of vortex identification methods. The algorithm is demonstrated with experimental Stereoscopic Particle Image Velocimetry data from two cases; a high-Reynolds (≈ 106) vortex generated by a half-delta wing, and distorted flow in a scaled-model of a civil aero-engine intake under cross-wind conditions. The approach is a successful method for the segmentation of complex vortical flows under a wide range of conditions.