A low-cost low-viscosity refractive-index-matching pair -- PIV error measurement and sensitivity analysis

Abstract Refractive-index matching (RIM) techniques are often employed in order to perform optical flow measurements around solid bodies: the RI of a transparent solid object is matched with that of the working fluid. The solid becomes "invisible" and full optical access inside or behind the object is granted free of distortions. In this paper, a low-cost and low-viscosity RIM pair is proposed. It combines the polymer THV (tetrafluoroethylene, hexafluoropropylene and vinylidene fluoride), which is commercially available and manufactured in various shapes, with NaCl (table salt) added to water for the working fluid. The low cost of table salt combined with the viscosity of the solution close to water allows this RIM pair to be extended for large scale, high Reynolds number experiments, such as open-channel flows. Here, a set of validation experiments is presented which was designed to analyse the errors and matching sensitivity of PIV measurements in an actual flow rather than using a still target pattern -- the common testing method. The flow -- a dipole impacting a patch of vertical tubes -- is measured simultaneously by two PIV systems, where one camera takes images through a patch of THV tubes while the other camera has a free optical path to the same measurement laser plane. Comparing the results between the two cameras gives an estimate of the PIV measurement error. To further analyse the PIV errors and their dependence on the tubes' geometry and their sensitivity to slight differences in the refractive indices between the solid and the fluid, a mathematical model of light refraction is employed. The higher thickness of the tube is found to increase the PIV error less than expected and sometimes even decrease it for a given resolution. It is found that the PIV error is inversely proportional to the PIV correlation box size relative to the tube size. This conclusion is also valid for other geometries, where edges or curved surfaces are present. That is, close-up measurements (higher resolution) around objects require higher precision of the RI matching.