Enhancing Tomo-PIV reconstruction quality by reducing ghost particles

Abstract

A technique to enhance the reconstruction quality and consequently the accuracy of the velocity vector field obtained in Tomo-PIV experiments is presented here. The methodology involves detecting and eliminating spurious outliers in the reconstructed intensity field (ghost particles). A simulacrum matching-based reconstruction enhancement (SMRE) technique is proposed, which utilizes the characteristic shape and size of actual particles to remove ghost particles in the reconstructed intensity field. An assessment of SMRE is performed by a quantitative comparison of Tomo-PIV simulation results and DNS data, together with a comparison to Tomo-PIV experimental data measured in a turbulent channel flow at a matched Reynolds number (Reτ = 937) to the DNS study. For the simulation data, a comparative study is performed on the reconstruction quality based on an ideal reconstruction determined from known particle positions. The results suggest that a significant improvement in the reconstruction quality and flow statistics is achievable at typical seeding densities used in Tomo-PIV experiments. This improvement is further amplified at higher seeding densities, enabling the use of up to twice the typical seeding densities currently used in Tomo-PIV experiments. A reduction of spurious vectors present in the velocity field is also observed based on a median outlier detection criterion. The application of SMRE to Tomo-PIV experimental data shows an improvement in flow statistics, comparable to the improvement seen in simulations. Finally, due to the non-iterative nature of SMRE, the increase in processing time is marginal since only a single pass of the reconstruction algorithm is required.