Effect of Refractive Distortions on Volumetric Reconstructions of PIV, PTV and LIF Light Fields

In this work, we analyze the effect of refractive distortions produced by a turbulent flow containing random index of refraction fluctuations on the accuracy of tomographic reconstruction utilizing standard algorithms for light field reconstruction (Tomo-PIV, Tomo-LIF) and particle tracking (Shake-the-box PTV). In order to produce images from a known emission field, an in-house implementation of a raytracer using CUDA/C++ was developed. The raytracer generates synthetic images of particle fields or continuous emission fields with refractive distortions from a known turbulent flow field for different refractive index strengths. The synthetic images are then fed to the volumetric reconstruction algorithm.

It was found that the reconstruction quality quickly starts degrading for ray displacements greater than 3 pixels as seen by the cameras, for both Tomo-PIV/LIF and Shake-the-box PTV. In particle applications, the number of ghost particles greatly increases when the ray displacement is greater than 3 pixels. The trends observed in the synthetic data will be evaluated experimentally using a water jet with temperature gradients. A theoretical foundation based on stochastic ray displacement (“random walk”) is also presented to establish the limits of the cited tomographic reconstruction techniques for arbitrary flows with refractive distortions by relating the ray distortion to the turbulent length scales of the flow.