Multi-camera single-plane PIV imaging in two-phase flow for improved dispersed-phase concentration

Abstract

This paper presents a multi-camera method to reconstruct the instantaneous position of large dispersed-phase particles in systems where the optical depth is of order O(1), with a specific emphasis on problems in sediment transport. Although much work has been performed in multi-camera three-dimensional reconstruction methods, the majority of prior work has been focused on small tracer particles appropriate for single-phase PIV. The large difference in the size between typical tracer particles and suspended sediment (usually 10 to 100 times) gives sediment particles distinct image characteristics, which violates the assumptions made by most 3D reconstruction techniques in current use and thus motivates us to develop a technique to accommodate the unique image signature of sediment particles. Inspired by the work of Maas et al. (1993), Khalitov and Longmire (2002), Spinewine et al. (2003), Knowles and Kiger (2012), this paper introduced a multi-camera thin light sheet imaging method to accurately measure the dispersed phase concentration up to optical densities of close to O(1). The work is an extension of a prior single-camera method (Knowles and Kiger 2012) that utilizes particle image characteristics to identify particles and, when appropriately calibrated, provide a measure of the effective measurement volume thickness. By introducing multiple camera perspectives, stereophotogrammetry methods (Maas et al. 1993; Spinewine et al. 2003) can be combined with the particle image characteristics to provide (1) increased accuracy in determining individual particle locations and (2) increased reliability in identifying all of the dispersed-phase objects in the face of larger increased volume fraction. The method is calibrated through the use of a fixed solid/gel suspension test cell that mimics the optical properties of a solid/water suspension. The static arrangements of the particles allow for a repeatable volume scan of the cell. This is subsequently used to produce an accurate map** of the particle locations within the test volume, which serves as the reference set for evaluating the performance of the new method. Comparisons are made over a range of volume fractions from $C = 1 \times 10^{-4}$ to $C = 1.2 \times 10^{-2}$ for a fixed spherical particle size of $D = 240 \, \mu m$. The new method is able to provide an accuracy of $2 \%$ up to a volume fraction of $C \approx 8 \times 10^{-3}$, which is an order of magnitude greater than the single-camera method used previously. Finally, the proposed technique is applied to an oscillatory sheet flow to demonstrate its advantage over the standard 3D-PTV and single-camera methods.