Phase discrimination and object fitting to measure fibers distribution and orientation in turbulent pipe flows

In this work, a methodology based on the analysis of single-camera, double-pulse PIV images is described and validated as a tool to characterize fiber-dispersed turbulent flows in large-scale facilities. The methodology consists of image pre-treatment (intensity adjustment, median filtering, threshold binarization and object identification by a recursive connection algorithm) and object-based phase discrimination used to generate two independent snapshots from one single image, one for the dispersed phase and one for the seeding. Snapshots are then processed to calculate the flow field using standard PIV techniques and to calculate fiber concentration and orientation statistics using an object-fitting procedure. The algorithm is tuned and validated by means of artificially generated images and proven to be robust against identified sources of error. The methodology is applied to experimental data collected from a fiber suspension in a turbulent pipe flow. Results show good qualitative agreement with experimental data from the literature and with in-house numerical data.