Optimization and testing of the tomographic method of velocity measurement in the flow volume

Abstract

The optic noncontact method of velocity field measurement in the flow volume is considered in this paper for the purposes of hydroaerodynamic experiment. The essence of this method is measurement of particles motion in the flow during short periods between laser pulses. This study offers and implements several algorithmic optimizations, allowing data processing time reduction. It is shown that application of threshold background filtering on the recorded projections (particle images) and fast estimation of initial intensity distribution in the volume allows increasing the speed of tomographic reconstruction algorithm two or three times. Reconstruction accuracy and errors in determination of particle shift were studied in this work using artificial images. The described tomographic method for the velocity field estimation in the flow volume was used for diagnostics of a turbulent submerged jet flowing into a narrow channel. The application of developed approaches in experiment allowed us to obtain spatial distribution of the average velocity field and instantaneous velocity fields in the measurement area.

A maximum entropy reconstruction technique for tomographic particle image velocimetry

“…In these regions the amplitude of the mean normal velocity component (W ) is 0.2 voxels. Therefore, the random error of the W component is <0.2 voxels for both reconstruction schemes due to the reproducible mean structure of the flow, which was also confirmed by the planar PIV measurements [20].…”

“…Figure 11 demonstrates the comparison between two mean velocity fields referring to the secondary flows in a slot jet shear layer [20,21]. Visually, both contour plots for the mean crosswise velocity component only have minor differences.…”

Characterization of Near-Wall Turbulent Flows by Tomographic Piv

“…This MLOS-SMART achieved better performance than MART. This fact was after confirmed by [Bilsky et al, 2011] and [Thomas et al, 2014], albeit the latter authors detected a slight quality loss.…”

“…For the SMART reconstruction method, the convergence was obtained with 15 iterations, in agreement with [Bilsky et al, 2011]. Around 28 iterations were required to obtain negligible changes in the reconstruction volume yielding a < Q reproj > of 87.0% and a SN R E of 2.17.…”

“…When a synthetic data set is provided, the uncertainty in velocity measurements can be calculated from the ground truth displacement or velocity field (e.g., [Elsinga et al, 2006, Worth and Nickels, 2008, Michaelis et al, 2010, Atkinson et al, 2011, Bilsky et al, 2011, Schanz et al, 2013a, Lynch and Scarano, 2014). The mean error norm can be calculated from the difference of the synthetic velocity field, u u u 0 = (u 0 , v 0 , w 0 ), and the corresponding velocity field computed from the pair of reconstructed volumes,…”