Bubble Trajectory in a Bubble Column Reactor using Combined Image Processing and Artificial Neural Network

Abstract

Abstract The applications of Bubble column reactors in gas-liquid multiphase reactions are widely observed in process industries. Biochemical reactions such as wet oxidation and algae bio-reactions are carried out in bubble column reactors. In this article, an image processing based comprehensive algorithm is developed to identify the trajectory of bubbles in a bubble column reactor. Photographs of bubbles moving up in a bubble column reactor are recorded for different velocities using a high speed camera. An algorithm is developed to plot the trajectory of the bubble. The developed algorithm can be used with experimental and numerical results to trace the trajectory of bubbles. The algorithm is applied to the results of volume of fluids (VOF) simulation to identify the bubble path in Newtonian and non-Newtonian fluids. Based on the algorithm, numerical results obtained on Newtonian fluids are used to train an Artificial Neural Network (ANN) to find the temporal position of the bubble. Superficial fluid velocities, nozzle diameter and time are the input parameters. The trained Levenberg-Marquardt based neural network can find the position of the bubble at any instant of time. The designed algorithm can study the dynamics and position of a bubble in process applications carried out in a bubble column reactor.