Phase-locking particle image velocimetry measurement of unsteady flow behaviors: Online dynamic mode decomposition using field-programmable gate array

Abstract

A novel online dynamic mode decomposition (DMD) approach using a field-programmable gate array (FPGA), which takes full advantage of the DMD to extract multiple unsteady events and the FPGA system for signal sampling and fast computation, was developed for phase-locking particle image velocimetry (PIV) measurements of unsteady flow behaviors. The turbulent separated and reattaching flow around a finite blunt plate with a length-to-height-ratio L/D = 6.0 was examined to demonstrate this novel approach. The wall-pressure field and the velocity field were measured using arrayed microphones and the conventional planar PIV setup, respectively. Offline DMD analysis of the wall-pressure fluctuation field was first used to identify the dominant modes corresponding to the energetically unsteady events. For each mode, the eigenmode and its mode coefficient reflected the spatial footprint pattern and temporal strength of the unsteady event, respectively. Next, trained machine learning of the mode coefficient was used to establish a phase prediction strategy. Finally, in the online analysis, the relevant eigenmode was cast into the FPGA device to serve as the reference mode for reconstruction with the sampled wall-pressure data, determining the phase signal to fire the PIV setup. High-resolution spatiotemporal evolutions of the dominant flow structures (i.e., the flapping separation bubble, the impinging leading-edge vortex, and the trailing-edge vortex street) were separately assembled. Further measurements demonstrated a clear panoramic view of the synchronous behavior of the enlarging separation bubble and the impinging leading-edge vortex. The proposed online FPGA-DMD approach can serve as a sophisticated strategy for phase-locking PIV measurements of unsteady flow behaviors.

Numerical simulation of vortex shedding from rectangular cylinders with different elongation ratios

“…Using the advanced modal analysis methods, such as proper orthogonal decomposition (Shi et al 2010, Zhang andLiu 2017) and dynamic mode decomposition (Liu and Zhang 2015), the main flow structures near rectangular cylinders with L/D = 3, 6, 9 had been successfully identified, and intermittent reattachment of the separated flows to the trailing edge of the rectangular cylinder was observed, which added great unsteadiness to the flow. Furthermore, Deng et al (2019Deng et al ( , 2020 investigated the interaction between flapping separation bubbles and large-scale vortices around a rectangular cylinder with L/D = 6 and the characteristic frequency of separation bubbles was observed to be one-third of that of L-vortices, indicating that the three successive impact processes (the amplification, transition and preparation process) were statistically coupled to one flutter period of the separation bubble, whereas the changes of these three interaction processes did not show strong periodicity.…”

Measurement techniques for aeroacoustics: from aerodynamic comparisons to aeroacoustic assimilations

“…The advantages of PIV in obtaining either statistical or time-resolved information in 2D and 3D are assessed also in many industrial configurations for the study of landing-gear noise (as for example by Neuhart et al [158] and Brès et al [159]) or for rotor configurations [160][161][162][163][164][165]. In figure 13, the results from Wang et al [166] show a peculiar usage of PIV measurements. In their setup, PIV was employed in a phaselocked configuration.…”