Development and performance of robust particle image velocimetry algorithms and investigation of a model tornado-like vortex: Kinematics and proper orthogonal decomposition

In the study of fluid flows experimentally obtained data is of importance due to the complex nature of modelling complex fluid flows computationally. Most flows of academic or industrial interest are turbulent, requiring complex experimental methods to capture meaningful data. Currently the dominant method of experimentally obtaining fluid flow data is Particle Image Velocimetry (PIV). The first part of this study focuses on evaluating cross-correlation based PIV vector calculation algorithms. Algorithms are tested against synthetic images depicting Oseen vortex flow enabling calculated and exact displacements to be compared. Image parameters are varied to investigate the sensitivities of each algorithm to each parameter. The methods evaluated are standard cross-correlation, multigrid (MGRID), the products of adjacent correlation functions (CBC), particle image deformation (PID) and an original method combining PID methods and CBC methods. The new method is shown to return displacement vectors with a reduced level of error compared to existing methods, especially in regions of high spatial velocity gradients. The second part of this study focuses on recording and evaluating experimentally obtained fluid flow data. One of the most important flow features within turbulence is the vortex, to investigate vortex flow a laboratory tornado-like vortex generator was created from which PIV data could be obtained. Velocity vectors were extracted for varying flow configurations showing the impact of swirl ratio on vortex characteristics such as variability and precession of the vortex as well as the internal vortex structure. The methods developed for quantifying this variability give a greater ability to understand the bulk and internal vortex dynamics for swirling flows. Proper Orthogonal Decomposition, a correlation based method for extracting coherent structures is applied to the tornado-like flow to further show the underlying dynamics of the variations within the vortex flow and its spatial precession. Temporal characteristics of the tornado-like vortex are also investigated using velocity vectors obtained from high-speed PIV.