Development and assessment of a particle tracking velocimetry (PTV) measurement technique for the experimental investigation of oil drops behaviour in dispersed oil–water two-phase flow within a centrifugal pump impeller

Abstract

The objective of the current work is to present the development of a Particle Tracking Velocimetry (PTV) algorithm for the analysis of oil drops behaviour in two-phase oil–water dispersions within a centrifugal pump impeller. The drop tracking was realized through high-speed camera images in a transparent pump prototype, which enabled the visualization of oil drops dispersed in water in all the impeller channels. The PTV algorithm is based on deep-learning techniques for image processing. The drops are detected by a combined U-Net and Convolutional Neural Network (CNN) method, with the former generating a binary image and the latter detecting valid oil drop contours. After detection, the Labelled Object Velocimetry (LOV) is adopted to calculate the instantaneous oil drop velocity. A synthetic image generator based on a Generative Adversarial Network (GAN) is then developed to assess the results from the U-Net, CNN, and LOV models. Additional validation studies are performed using the results from Perissinotto et al. (2019a). The results reveal that the presented deep-learning PTV algorithm is robust and provides consistent and reliable data for the dispersed oil phase in two-phase oil–water flows.