Temporal Super Resolution X-ray Particle Velocimetry for Opaque Flows

X-ray imaging for studying single- and multiphase flows has been evaluated extensively over the past two decades. The advancement of X-ray Particle Velocimetry (XPV) algorithms, photon counting energy resolved x-ray imagers and laboratory sources of increased brightness make it feasible to apply these techniques to optically inaccessible flows. To date for XPV, large number of radiographs are acquired over 180 or 360 degrees, for parallel and cone beam geometries, respectively, CT reconstruction is performed, and particles are tracked in the reconstructed volumes. Locating the centroid of particles becomes challenging with motion artifacts in the flow beyond a few diameters of the particle which limits the temporal resolution achievable with XPV. In principle, a small number of radiographs contain sufficient information to track a particle, but larger numbers are needed to handle uncertainties and particle overlaps. We propose a temporal super resolution approach to XPV. We generate synthetic particle image data via Beer-Lambert modelling accounting for the energy dependent X-ray beam attenuation, scatter and detector response. The results of the simulation are validated experimentally using a laboratory setup with both XPV phantom and real flow data. The true position and velocity fields generated experimentally and synthetically are then compared to understand the current limitations for applying XPV to flows of interest.